previous experiments indicate that cdk5

• Open access
• Published: 23 November 2021

Three decades of Cdk5

• Ping-Chieh Pao 1 , 2 &
• Li-Huei Tsai 1 , 2  

Journal of Biomedical Science volume  28 , Article number:  79 ( 2021 ) Cite this article

11k Accesses

59 Citations

13 Altmetric

Metrics details

Cdk5 is a proline-directed serine/threonine protein kinase that governs a variety of cellular processes in neurons, the dysregulation of which compromises normal brain function. The mechanisms underlying the modulation of Cdk5, its modes of action, and its effects on the nervous system have been a great focus in the field for nearly three decades. In this review, we provide an overview of the discovery and regulation of Cdk5, highlighting recent findings revealing its role in neuronal/synaptic functions, circadian clocks, DNA damage, cell cycle reentry, mitochondrial dysfunction, as well as its non-neuronal functions under physiological and pathological conditions. Moreover, we discuss evidence underscoring aberrant Cdk5 activity as a common theme observed in many neurodegenerative diseases.

Discovery of Cdk5

Cyclin-dependent kinase 5 (Cdk5) is a proline-directed serine/threonine protein kinase. Thirty years ago, Cdk5 was first discovered on the basis of its close sequence homology to the human cell division cycle protein 2 (Cdc2, also known as Cdk1), a regulator of cell cycle progression [ 1 , 2 , 3 ]. The kinase activity of Cdc2 is detected in proliferating cells and activates prior to or during S-phase [ 4 ]. Cdk5 was originally inferred to function in cell cycle regulation based on its 60% sequence identity and similar substrate specificities to Cdc2 [ 1 , 2 , 3 ]. However, three decades of research have demonstrated that Cdk5 plays many key roles in post-mitotic neurons.

Unlike other Cdks, Cdk5 activity is most apparent in adult mouse brain [ 5 ], where most of the cells are terminally differentiated. During brain development, the expression and kinase activity of Cdk5 gradually increase from embryonic day 11 (E11) and peak at E17, which correlates with the main phase of neuronal differentiation in the developing mouse neocortex [ 5 ]. In contrast, the expression pattern and kinase activity of Cdc2 decline as development proceeds [ 5 ]. These findings suggest functions for Cdk5 in brain development and neuronal differentiation.

Regulation of Cdk5: regulatory subunit

The Cdk family comprises 21 members, most of which depend on the association with specific cyclin partners to become constitutively active protein kinases [ 6 ]. Surprisingly, the subunit required for Cdk5 activity is a non-cyclin protein named p35. p35 was identified as the Cdk5 regulatory subunit and lacked the conserved amino acid sequences typically found in cyclins [ 7 , 8 , 9 ]. p35 physically associates with Cdk5 in brain lysates and is capable of activating Cdk5 upon direct binding [ 7 ]. The expression of Cdk5 and p35 transcripts overlaps spatially and temporally in the developing mouse neocortex, and p35 is primarily expressed in the post-mitotic neurons [ 7 ]. Single-cell transcriptomic analysis of the adult mouse brain indicates that the expression of Cdk5 is not restricted to neurons [ 10 ] (Fig.  1 ). Nonetheless, Cdk5r1 , the gene encoding p35, shows marked expression predominately in the neuronal linage [ 10 ] (Fig.  1 ).

Single-cell transcriptomic analysis of the expression of Cdk5 , Cdk5r1 (the gene encoding p35), and Cdk5r2 (the gene encoding p39) in young and aged mouse brain. The data are based on Ximerakis et al., 2019, Nat. Neurosci., and the graphs were generated by Single Cell Portal website ( https://singlecell.broadinstitute.org/single_cell/study/SCP263/aging-mouse-brain ). Astrocyte lineage: neural stem cells, astrocyte-restricted precursors and astrocytes. Ependymal cells: ependymocytes, hypendymal cells, tancytes, and choroid plexus epithelial cells. Immune cells: microglia, monocytes, macrophages, dendritic cells, and neutrophils. Neuronal lineage: neuronal-restricted precursors, immature neurons, mature neurons, and neuroendocrine cells. Vasculature cells: endothelial cells, pericytes, hemoglobin-expressing vascular cells, vascular smooth muscle cells, vascular and leptomeningeal cells, and arachnoid barrier cells

Another Cdk5 regulatory subunit, p39, was later discovered in part by its high degree of sequence identity to p35 [ 11 ]. While p39 shares many characteristics with p35 including its high expression in the brain, direct binding to Cdk5, and the ability to activate Cdk5 [ 11 , 12 , 13 ], there are distinct properties between p35 and p39. During brain development, the expression of p35 is high from embryonic stage to postnatal stage, whereas that of p39 starts increasing postnatally [ 14 ]. p35 and p39 exhibit different regional distribution patterns with p35 expressed most prominently in the cerebral cortex and cerebellum while p39 highly expressed in the cerebellum, brain stem, and spinal cord [ 14 ]. Moreover, p39 protein displays a higher protein stability and lower binding affinity for Cdk5, relative to p35 [ 15 , 16 ]. Studies have identified functions that are preferentially regulated by Cdk5/p39 but not Cdk5/p35, regardless of their redundant roles in the nervous system. Deletion of p39 or Cdk5 in cultured neurons causes defective dendritic morphogenesis whereas no abnormality was observed in cultured neurons lacking p35 expression [ 17 ]. Cdk5/p39 also plays a dominant role in Rac1-induced lamellipodia formation [ 18 ].

Regulation of Cdk5: membrane targeting

Subcellular fractionation assays reveal the enrichment of p35 and p39 proteins in membrane-bound fractions of cultured neurons [ 12 , 19 ]. The membrane-associated localization of p35 and p39 is promoted by the myristoylation signal on a conserved glycine located at position two of these two proteins [ 20 ]. Myristoylated p35 and p39 recruits Cdk5 to cell membrane [ 20 , 21 ], whereas harboring mutated myristoylation signal fails to retain their membrane-targeting distribution [ 20 , 21 ]. Thus, under normal circumstances, active Cdk5 complexes primarily reside on the cell membrane.

Regulation of Cdk5: post-translational modifications

While binding to a regulatory subunit is obligatory for Cdk5 activation, its activity can be further modulated by a variety of post-translational modifications. The phosphorylation of Cdk5 at two different residues within the ATP-binding sites leads to opposing effects. Phosphorylation at Tyr 15 is stimulatory but inhibitory at Thr 14 [ 22 , 23 ]. Moreover, phosphorylation of Ser 159 in the T-loop of Cdk5 is critical for p35 binding [ 24 ], and acetylation at Lys 33 impairs Cdk5 activity due to the loss of ATP binding [ 25 ]. Recent evidence indicates a role for S-nitrosylation in Cdk5 regulation. S-nitrosylation of Cdk5 at Cys 83, a residue locating in the ATP-binding pocket, suppresses Cdk5 activity [ 26 ]. S-nitrosylation of p35 at Cys 92 induces its degradation via the proteasome, which in turn reduces Cdk5 activity [ 27 ].

Roles of Cdk5 in the nervous system

There has been rapid progress in our understanding of Cdk5 function in the nervous system.

Prior review articles on Cdk5 have provided a comprehensive coverage of its roles in neuronal migration, neurite outgrowth, axonal guidance, and synaptic plasticity . In this review, we focus on the recent findings of Cdk5, particularly its roles in neuronal/synaptic functions and circadian clock regulation under physiological conditions, as well as its pathological links to DNA damage, cell cycle reentry, mitochondrial dysfunction, and oxidative stress. The Cdk5 substrates discussed in this review and their functional categories are listed in Table 1 , and the reader is referred to several excellent reviews with a more comprehensive list of known Cdk5 substrates [ 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 , 36 , 37 ].

Neuronal migration

Gene-targeting studies demonstrate a role for Cdk5 in brain development, particularly in neuronal migration [ 38 , 39 , 40 ]. Cdk5 modulates neuronal migration through multiple pathways. Cdk5/p35 interacts with N-cadherin adhesion complex. Pharmacological inhibition of Cdk5 enhances N-cadherin-mediated cell adhesion, which hinders neuronal migration [ 41 ]. A recent study shows that Cdk5 phosphorylates RapGEF2 at S1124, and subsequently activates Rap1, a key factor modulating neuronal migration [ 42 ]. Moreover, semaphorin-3A (Sema3A)-elicited neuronal migration requires Cdk5 phosphorylating Synapsin III at Ser 404 [ 43 ]. Additional substrates of Cdk5 functioning in neuronal migration include NUDEL, FAK, disrupted-in-Schizophrenia-1 (Disc1), doublecortin (Dcx), and Dix-domain containing 1 (Dixdc1), and p27 [ 19 , 44 , 45 , 46 , 47 , 48 , 49 ].

Neurite outgrowth

On the basis of its unique expression and activity pattern in the developing mouse brain, Cdk5 is believed to have roles in neurogenesis. Indeed, Cdk5 is essential for neurite growth and axonal formation. Cdk5 and p35 show co-localization with actin filaments in axonal growth cones, and inactivation of Cdk5 in cultured neurons inhibits neurite outgrowth [ 50 ]. Consistently, disrupting Cdk5 function causes axon patterning defects in the Drosophila model [ 51 ], and p35-null mice show altered axonal and dendritic trajectories [ 52 ]. In contrast, co-expression of Cdk5 and p35 increases neurite length in cultured neurons [ 50 ]. Cdk5 can promote neurite outgrowth through phosphorylating Pak1, a kinase that regulates the dynamics of actin and microtubule fibers [ 53 ], whereas neurite growth elicited by the serotonin 6 receptor (5-HT6R) requires receptor phosphorylation at Ser 350 by Cdk5 [ 54 ]. Cdk5 also modulates axonal outgrowth by phosphorylating GRAB, a guanine nucleotide exchange factor for Rab8 [ 55 ]. Recent findings suggest that Cdk5 regulates dendritic morphogenesis by an adaptor protein WD repeat and FYVE domain-containing 1 (WDFY1) [ 17 ]. Cdk5 can also enhance neurite outgrowth through the association with Cables and c-Abl complex [ 22 ], as well as the phosphorylation of Map1b [ 56 ], a microtubule-binding protein, and Axin [ 57 ], a scaffold protein of the Wnt pathway.

Synaptic plasticity

Impaired long-term depression and depotentiation of long-term potentiation have been shown in p35-deficient mice [ 58 ], implying a role for Cdk5/p35 complex in synaptic plasticity. Synaptic plasticity reflects modification of the efficacy or strength of synaptic transmission in response to neuronal activity and is integral to memory formation. Regulation of synaptic plasticity occurs through multiple mechanisms at both pre- and post-synaptic levels [ 59 ]. Presynaptically, Cdk5 controls exocytosis, endocytosis, and Ca 2+ influx. Phosphorylation of Munc-18 by Cdk5 results in the dissociation of Munc-18 from Syntaxin 1A, facilitating synaptic fusion and release [ 60 ]. Cdk5 also regulates endocytosis through Dynamin I and Amphiphysin I [ 61 , 62 ], two components of Clathrin-mediated endocytosis. Cdk5 also induces Ca 2+ influx into the pre-synaptic cytoplasm via voltage-dependent Ca 2+ channels (VDCCs), which increases the probability of channel opening and facilitates the release of neurotransmitters [ 63 , 64 ].

Change in the numbers or properties of post-synaptic receptors is one of the mechanisms for modulating synaptic plasticity at the post-synapses. Cdk5 phosphorylates the NMDA receptor subunit NR2A at Ser1232, thereby enhancing NMDA receptor function [ 65 , 66 ]. PSD-95 is a postsynaptic scaffold protein that tethers NMDA receptors to the cytoskeleton. Cdk5 catalyzes PSD-95 at three residues near the N-terminal domain, which causes a reduction in co-clustering of PSD-95 and neuronal ion channels [ 67 ]. Conversely, inhibition of Cdk5 function enhances PSD-95 clustering [ 67 ]. Cdk5-mediated phosphorylation of PSD-95 increases the degradation of PSD-95 by the ubiquitin-proteosome pathway [ 68 ]. Recent work reveals that Cdk5 regulates activity-dependent dendritic spine remodeling through multiple mechanisms. Cdk5-mediated phosphorylation of scaffold protein Liprinα1 at Thr 701 declines in response to neuronal activity, which is linked to enhanced excitatory synaptic function by promoting the binding of Liprinα1to PSD-95 and PSD-95 synaptic localization [ 69 ]. Moreover, Cdk5 is crucial for BDNF-TrkB signaling. Phosphorylation of TrkB on Ser 478 by Cdk5 increases activity-dependent structural plasticity and spatial memory [ 70 ]. Conversely, Cdk5 impairs activity-dependent dendric spine maintenance through a pseudokinase CaMKv, as phosphorylation of CaMKv by Cdk5 at Thr345 is associated with reduced spine density [ 71 ].

Synaptic homeostasis

Synaptic homeostasis is a compensatory process that allows neurons to adapt to altered levels of network activity. Neurons potentiate synaptic efficacy when inputs are dampened and downmodulate their firings when inputs are heightened [ 72 ]. Thus, homeostatic mechanisms ensure that neurons maintain their firings within an optimal range and protect network stability despite recurrent alterations in input activity. Cdk5 is important for synaptic scaling, a principal mechanism underlying homeostatic plasticity. At presynaptic terminals, synaptic vesicles recycle into recycling or resting pools. Recycling pools are available for release upon neuronal activation, whereas resting pools remain silent [ 73 ]. Long-term suppression of neuronal activity reduces presynaptic Cdk5 levels, and inhibition of Cdk5 is associated with unlocked resting vesicles and an increased pool of recycling vesicles [ 74 ], indicative of synaptic strengthening.

At the postsynaptic levels, Cdk5 has been implicated in depressing synaptic strength following heightened neuronal activity [ 75 ]. Upon increased network activity, Cdk5 phosphorylates spine-associated Rap guanosine triphosphatase-activating protein (SPAR), a postsynaptic scaffold protein regulating actin dynamics and promoting the growth of dendritic spines [ 76 ]. Priming phosphorylation of SPAR at S1328 by Cdk5 induces Plk2-mediated phosphorylation of SPAR, leading to ubiquitin-dependent degradation of SPAR and synaptic weakening [ 75 ]. Collectively, these observations underscore the role of Cdk5 in modifying synaptic scaling by regulating the partition of presynaptic vesicles and the degradation of postsynaptic SPAR scaffold protein.

Circadian clocks

Circadian clocks are oscillators that synchronize daily cycles of behavior and physiology. The suprachiasmatic nucleus (SCN) of the hypothalamus is the master circadian pacemaker in mammals and entrains the peripheral clocks across the body [ 77 ]. Circadian clocks are generated in a transcriptional autoregulatory feedback loop by the circadian machinery. The core circadian machinery consists of the transcriptional activators CLOCK and BMAL1 and the repressors PER1/2 and CRY1/2. CLOCK/BMAL1 heterodimer activates the transcription of a set of circadian clock genes, including Per1/2 and Cry1/2 . Newly synthesized PER and CRY proteins heterodimerize, translocate into the nucleus, and inhibit CLOCK/BMAL1 activity through direct binding, resulting in the subsequent repression of downstream target genes [ 77 ]. Thus, circadian clock genes display an oscillatory expression pattern. Perturbation of circadian clocks compromises brain function, and circadian dysfunction is a common symptom of various neurodegenerative diseases including Alzheimer’s disease (AD) [ 78 ].

Cdk5 has been implicated in the regulation of circadian clocks [ 79 , 80 ]. The running wheel test is a method to record circadian rhythm, whereby wild-type mice start wheel running precisely at the beginning of the dark phase. Mice injected with adeno-associated virus expressing shRNA against Cdk5 in the SCN show earlier onset of wheel running activity, which phenocopies mice harboring Per2 silencing [ 79 ]. Moreover, rhythmic expression of CLOCK target genes, including Per1 and Per2 , is disturbed in p35 heterozygous knockout mice [ 80 ]. Cdk5 interacts with and phosphorylates CLOCK at residues T451 and T461. Cdk5-mediated phosphorylation causes nuclear translocation and transcriptional activation of CLOCK [ 80 ]. In addition, Cdk5 phosphorylates PER2 at S394 residue, which stabilizes PER2 protein and promotes its nuclear translocation [ 79 ]. Collectively, Cdk5 regulates circadian clocks by its phosphorylation on several components of the core clock machinery.

Dysregulation of Cdk5: Calpain-dependent proteolytic cleavage of p35 to p25

Aberrant Cdk5 activity caused by p25 accumulation contributes to the pathogenesis of various neurodegenerative diseases [ 81 ]. p25 is a 208-residue carboxy-terminal fragment of p35. The mechanism underlying the cleavage of p35 to p25 has been well-characterized. Neurotoxic insults such as ischemia, the addition of hydrogen peroxide, glutamate or ionomycin, cause calcium influx and trigger the activation of a cysteine protease named calpain [ 82 ]. Calpain cleaves p35 at Phe 98 /Ala 99 sequence and generates p25 in a calcium-dependent manner. Accordingly, increasing intracellular calcium level stimulates p25 generation, whereas removing calcium prevents p25 accumulation [ 82 ].

p25 causes constitutive activation and mislocalization of Cdk5. p25 activates Cdk5 through direct binding, and p25 has an approximately 5- to 10-fold longer protein half-life compared to p35 [ 20 ], thereby prolonging Cdk5 activation. Moreover, p25 lacks the myristoylation signal that normally tethers Cdk5 to the membrane. Immunohistochemical and cell fractionation analysis demonstrate that p25 is enriched in the nuclear and perinuclear regions of the cell [ 20 ]. These findings suggest that p25 promotes Cdk5 hyperactivation and redirects Cdk5 to a wider array of substrates under pathological contexts (Fig.  2 ).

Activation of Cdk5 and its functions under physiological or pathological conditions

p25-mediated neurotoxicity and neurodegeneration

Dysregulation of Cdk5 activity leads to neurotoxicity and neurodegeneration [ 83 ]. Cultured neurons overexpressing p25 exhibit morphological deterioration and apoptotic cell death, characterized by degenerated neurites and fragmented nuclei [ 20 ]. To understand the consequence of Cdk5 hyperactivation in vivo, several groups have generated mouse models of p25 overexpression [ 84 , 85 ]. Different genetic approaches converge on similar pathological phenotypes.

CK-p25 mice are one such p25 overexpression model. CK-p25 mice overexpress an inducible human p25 under the control of a forebrain-specific CamKII promoter, which is expressed predominantly in excitatory pyramidal neurons. The expression of p25 is induced upon the removal of the tetracycline derivative doxycycline from the animal diet [ 85 ]. Following p25 induction in excitatory neurons, CK-p25 mice exhibit progressive neurodegenerative phenotypes. Immunohistochemical analysis in CK-p25 mice following acute p25 induction (2-week) reveals substantial DNA damage of double-strand breaks (DSBs) in p25-expressing neurons [ 86 ]. Neurons containing DSBs also express ectopic cell cycle markers [ 86 ], suggesting Cdk5 hyperactivation may lead to cell cycle reentry, which has been reported in post-mortem AD patient brain samples [ 87 ]. Neuronal DNA damage is associated with a dramatic morphological and transcriptional response in microglia, including structural remodeling of processes/cell bodies as well as upregulated expression of genes regulating cell division [ 85 , 88 ].

Aberrant Cdk5 activity is also found to trigger neuroinflammation. Neurons overexpressing p25 markedly produce and secrete a soluble lipid known as lysophosphatidylcholine, which activates glia and induces the expression of cytokines and chemokines [ 89 ]. Prolonged p25 induction is accompanied by increased tau phosphorylation and elevated levels of Aβ peptide from the cleavage of APP [ 85 ]. CK-p25 mice also exhibit severe neuronal loss and reduced hippocampal LTP induction, together with behavioral alterations including locomotor hyperactivity, increased anxiety, and memory impairment [ 85 , 90 ]. In Drosophila , dysregulated Cdk5 disrupts autophagy and augments the expression of anti-microbial peptides. Anti-microbial peptides cause the hyperactivation of innate immune response and are linked to dopaminergic neuronal death in Drosophila [ 91 ]. In summary, overexpression of p25 in excitatory neurons triggers several AD-like pathological hallmarks.

Cdk5 dysfunction in Alzheimer’s disease

AD is the leading cause of senile dementia, which is pathologically characterized by the accumulation of amyloid plaques and neurofibrillary tangles (NFTs) [ 92 ]. Several lines of evidence support a strong association between Cdk5 dysregulation and AD pathogenesis [ 93 ]. p25 accumulates in NFT-bearing neurons and in brain lysates from AD patients [ 20 , 94 ]. Relative to age-matched non-AD individuals, Cdk5 immunoprecipitated from AD brain tissue displays a greater activity when histone H1 was used as a substrate [ 20 ]. Similar observations were made in cellular and mouse models of AD. Introducing Aβ 42 peptide to cultured neurons leads to p25 generation and neuronal death [ 82 ], and accordingly, the Cdk5 small molecule inhibitor butyrolactone or the calpain inhibitor calpeptin alleviates Aβ-induced neuronal death [ 82 , 95 ]. A plethora of AD mouse models also reveal dysregulation of Cdk5 via p25 accumulation [ 96 , 97 ]. Here, we discuss pertinent aspects of how aberrant Cdk5 activity may contribute to AD progression.

APP processing and Aβ production

Dysregulation of Cdk5 activity promotes amyloid plaque deposition by promoting amyloidogenic APP processing. The amyloidogenic APP pathway includes sequential cleavage by β- and γ-secretases, which ultimately produce a secreted form of APP (sAPPβ), C-terminal fragments (CTF99 and CTF89) and Aβ peptides [ 98 ]. Aβ peptides form fibrillar aggregates that form amyloid plaques [ 99 ]. The first observation linking aberrant Cdk5 activity to Aβ production was made in the CK-p25 mouse model [ 100 ]. ELISA analysis demonstrates an increase in endogenous mouse Aβ levels after p25 induction. Immunolabeling of two antibodies recognizing Aβ peptide, 4G8 and 6E10, reveals intracellular accumulation of Aβ in neurons expressing p25 [ 100 ]. Studies indicate that Cdk5/p25 enhances Aβ production through STAT3-mediated transcriptional regulation of BACE1, a gene encoding β-secretase. Cdk5/p25 phosphorylates STAT3 at S727 residue and in turn activates BACE1 transcription [ 100 , 101 , 102 ]. Significant upregulation of BACE1 immunoreactivity and increased β-secretase processing of APP are found in CK-p25 mouse brain [ 100 ]. In addition, Cdk5 phosphorylates APP at T668, which facilitates the BACE1 cleavage of APP to increase Aβ generation [ 103 ]. As discussed above, Aβ peptide is able to induce p25 accumulation and Cdk5 overactivation which can result in a feed-forward reaction, whereby aberrant Cdk5 activity in turn amplifies Aβ-associated pathologies. In the context of familial AD, it is likely that elevated Aβ levels might be the causal factor initiating this cascade. On the other hand, Cdk5/p25 is linked to Aβ-induced synaptic depression. Cdk5/p25 leads to DARPP-32 inhibition and PP1/Calcineurin activation, which promotes AMPA receptor subunit GluA1 (Ser 845) dephosphorylation and impacts negatively on AMPAR endocytosis [ 104 ].

A previous report established a knock-in mouse model deficient in p25 generation named Δ p35KI mice [ 104 ]. In Δ p35KI mice, a mutant p35 resistant to calpain cleavage was designed by substituting an alaine [ 99 ] residue at the cleavage site with leucine and removing six amino acid residues (A 93 NLSTF 98 ) adjacent to the calpain-cleavage site [ 104 ]. 5XFAD mice are an AD mouse model that exhibit increased p25 levels, and the blockade of p25 generation by crossing with Δ p35KI mice rescues AD pathology in 5XFAD, Δ p35KI mice. Compared to 5XFAD mice, the 5XFAD, Δ p35KI mice display a reduction in soluble Aβ peptide levels, plaque deposition, and inflammatory cytokine expression. Furthermore, the 5XFAD, Δ p35KI mice show an improvement in synaptic plasticity and memory function relative to the 5XFAD mice [ 104 ]. Collectively, these findings highlight therapeutic potential of targeting Cdk5 hyperactivation in AD.

Tau is a microtubule-associated protein mainly distributed in axons. Tau stabilizes neuronal microtubules and regulates microtubule dynamics involved in axonal outgrowth and transport [ 105 ]. Tauopathy is a pathological feature characterized by the deposition of abnormal tau aggregation in the brain, and is present in a wide variety of neurodegenerative diseases such as AD and frontotemporal dementia (FTD). Hyperphosphorylation of tau is well known to enhance tau aggregation and form NFTs [ 105 ].

Cdk5 is a tau protein kinase: Cdk5 co-purifies with tau and phosphorylates tau in vitro [ 106 ]. Overexpression of Cdk5/p25 increases tau phosphorylation in cultured neurons compared to overexpression of the Cdk5/p35 complex [ 20 ]. Emerging evidence suggests that Cdk5 phosphorylates tau on many sites including T181, S202, T205, T212, T217, S235, S396, and S404. Notably, many of the Cdk5 target sites are hyperphosphorylated in post-mortem AD brain samples [ 107 , 108 ].

Mutations in the MAPT gene (the gene encoding tau) are associated with FTD and Parkinsonism linked to chromosome 17 (FTDP-17) [ 107 ]. Importantly, mutant tau protein shows higher propensity for phosphorylation and aggregation than the wild-type tau protein [ 105 ]. Increased calpain activity, p25 protein accumulation, and Cdk5 hyperactivity are observed in P301L and P301S tauopathy mouse models, which harbor FTD-associated tau mutations [ 109 , 110 ]. Crossing tau P301S mice with Δ p35KI mice reduces tau phosphorylation at residues T181 and S202 [ 110 ]. Compared to the tau P301S mice, brain extracts from the tau P301S, Δ p35KI compound mice display a reduction in tau seeding activity [ 110 ], which is believed to be critical for tau aggregation and propagation observed in AD brains. Importantly, neuronal loss and synaptic dysfunction are also ameliorated relative to the tau P301S mice [ 110 ]. Similarly, pharmacological inhibition of calpain by calpastatin reduces tau phosphorylation/aggregation and delays disease progression in mice expressing tau P301L [ 109 ].

A previous report utilized FTD patient-derived iPSCs that harbor the tau P301L mutation, and generated isogenic lines in which the tau leucine mutation was reverted to proline, as well as targeted knock-in Δ p35 in these lines using CRISPR/Cas9 genome-editing [ 110 ]. Compared to the brain organoids derived from an isogenic control iPSC line, tau P301L organoids display increased phosphorylated tau levels and p25 protein accumulation. Inhibiting p25 generation (tau P301L, Δ p35KI ) attenuates tau phosphorylation and increases the expression of synaptophysin [ 110 ]. Together, these findings highlight an important role for Cdk5 hyperactivity in tau-associated pathologies.

DNA damage and cell cycle reentry

Increased DNA damage and cell cycle reentry have been observed in AD human brains [ 111 ]. Most studies on Cdk5 function have focused on DNA DSBs, although we note that many kinds of DNA damage emerge in the context of neurodegeneration. Dysfunction of Cdk5 is prominently linked to elevated DNA DSBs and the expression of cell cycle genes in neurons. In the CK-p25 mouse model, acute p25 induction leads to an increase in the number of neurons bearing DNA DSBs marked by immunoreactivity to γH2AX, a histone modification in the vicinity of DSB sites [ 86 ]. Elevated DNA damage precedes the development of pathology and symptoms in CK-p25 mice, suggesting that DNA damage is a predisposing factor in the functional decline of the brain.

DNA DSBs in CK-p25 mice may be critically linked to the inhibition of histone deacetylase 1 (HDAC1) [ 86 ], a member of class I histone deacetylases. HDAC1 maintains genomic integrity in neurons by deacetylating histone H3 at lysine 56 (H3K56) and histone H4 at lysine 16 (H4K16) [ 112 ]. Upon the induction of DSBs, HDAC1 is rapidly recruited to DNA break sites to catalyze H3K56 and H4K16 deacetylation, which represses transcription of adjacent genes and promotes DSB repair through non-homologous end joining (NHEJ) [ 112 ], the predominant DSB repair pathway in neurons [ 113 ]. Furthermore, HDAC1 activity is reduced in CK-p25 mouse brains after acute p25 induction, and HDAC1 harbors a stronger binding affinity to p25 compared to p35 [ 86 ]. p25 interacts with the catalytic domain of HDAC1 [ 86 ], yet the mechanism underlying p25-mediated HDAC1 inactivation remains elusive.

Post-mitotic neurons normally do not exhibit cell cycle activity. However, re-expression of cell cycle-related proteins including Cyclin B and proliferating cell nuclear antigen (PCNA) has been documented in hippocampal pyramidal neurons of post-mortem AD brains [ 87 ]. Fluorescence in situ hybridization analysis indicates that ~ 3.7% of neurons analyzed in human AD brains exhibit S-phase activity, where their DNA has been duplicated. In contrast, age-matched healthy controls show no such abnormalities [ 87 ]. DNA damage is associated with cell cycle reentry in neurons [ 114 ]. Likewise, exposure of genotoxic reagents ( e.g., etoposide, methotrexate, and homocysteine) to cultured neurons increases DNA damage and Cdc25a levels, a marker of S-phase entry and DNA replication [ 114 ]. Importantly, the addition of Aβ 42 peptides to cultured neurons also induces DNA damage and cell cycle reentry [ 114 ], suggesting that Aβ may trigger this cascade during AD pathogenesis.

Cdk5 dysfunction has been linked to cell cycle reentry. Several cell-cycle markers are aberrantly upregulated in p25-expressing neurons in the CK-p25 mouse model [ 86 ]. Moreover, the small molecule Cdk5 inhibitor roscovitine attenuates DNA damage-induced cell cycle reentry in cultured cerebellar granule neurons [ 115 ]. Similarly, roscovitine or the calpain inhibitor MDL28170 diminishes Aβ-induced cell cycle reentry in rat cultured neurons [ 116 ]. Recent evidence reveals that Cdk5-mediated cell cycle reentry in neurons might also involve the β-catenin pathway. Hyperactive Cdk5 disrupts GSK3β-induced degradation of β-catenin, promoting β-catenin-mediated nuclear translocation and activation of cell-cycle machinery [ 117 ]. Collectively, these observations suggest Cdk5 dysfunction is a critical element in DNA damage and cell cycle reentry, two well-established pathological features of AD.

Mitochondrial dysfunction

Mitochondria utilize oxidative phosphorylation to produce ATP and adapt the metabolic needs of cells by fusion and fission. Mitochondrial fusion joins mitochondria together, while fission separates one mitochondrion into two or more [ 118 , 119 ]. Mitochondrial fusion is frequently found in metabolically active cells, and enables the formation of an extended mitochondrial network. Conversely, mitochondrial fission segregates components of the mitochondrial network, facilitating the removal of damaged components through mitophagy [ 118 , 119 ]. As such, imbalanced fusion/fission leads to mitochondrial dysfunction and degeneration. Emerging evidence links Cdk5 hyperactivity to excessive mitochondrial fission under pathological conditions, such as neurotoxic insults and neurodegenerative diseases [ 120 , 121 , 122 , 123 , 124 , 125 , 126 , 127 ].

Dynamin-related protein 1 (Drp1) is a GTPase that regulates mitochondrial fission. Drp1 is recruited from the cytosol to the mitochondrial outer membrane (MOM), where it assembles into ring-like structures that wrap around the MOM and incise the membrane following GTP hydrolysis [ 118 ]. In pathological conditions, Cdk5 phosphorylates Drp1 at S616, which increases its mitochondrial translocalization and GTPase activity, ultimately accelerating mitochondrial fission [ 120 , 121 , 122 , 123 , 124 , 125 , 126 , 127 ]. Excessive mitochondrial fission is associated with mitochondrial defects and neuronal death. Thus, pharmacological or genetic inhibition of Cdk5 restores mitochondrial ATP production and confers neuroprotection by attenuating Drp1-induced mitochondrial fission in disease models [ 120 , 121 , 122 , 123 , 124 , 125 , 126 , 127 ].

Cdk5 dysfunction in other neurodegenerative diseases

Parkinson’s disease.

Parkinson’s’ disease (PD) is a chronic movement disorder characterized by Lewy body formation, mitochondrial dysfunction, and loss of dopaminergic neurons in the substantia nigra [ 128 ]. Compared to age-matched healthy individuals, the brain extracts from PD patients exhibit increased calpain cleavage activity and p25 accumulation [ 129 ]. Among the most widely used models of PD are those that employ toxins, including 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) and paraquat [ 130 ]. Upregulated Cdk5 expression/activity and increased p25 generation have been observed in the MPTP mouse model [ 131 ]. Administration of a pan-Cdk synthetic inhibitor flavopiridol attenuates MPTP-induced degeneration of nigral dopaminergic neuron and reduces motor impairments in this model [ 131 ]. Viral-mediated expression of dominant-negative Cdk5 or a peptide inhibitor of Cdk5 prevents the death of dopaminergic neurons in the MPTP mouse model [ 131 , 132 ]. Similarly, a Cdk5 peptide inhibitor blocks Cdk5 hyperactivity and promotes dopaminergic neuronal survival in the nematode worm C. elegans after exposure to paraquat [ 133 ].

Several reports link Cdk5 hyperactivity to oxidative stress in PD. Reactive oxygen species (ROS) including hydrogen peroxide are byproducts of normal mitochondrial metabolism, and elevated levels of ROS damage lipids, proteins, and DNA, impeding a wide array of cellular processes [ 134 ]. A large group of antioxidant enzymes catalyze ROS into stable non-toxic molecules, which protects cells from damage [ 134 ]. Oxidative stress is a condition where ROS levels accumulate from an imbalance of ROS production and antioxidant capacity. Peroxidases are antioxidant enzymes that break down hydrogen peroxide into less reactive molecules [ 134 ]. Prior studies revealed that Cdk5 phosphorylates peroxidase 1 and 2 (Prx1 T90 and Prx2 T89) and represses their peroxidase activity [ 120 , 135 ]. Moreover, increased phosphorylation of Prx2 at T89 residue has been observed in brain samples from PD patients [ 135 ]. These studies suggest Cdk5 modulates oxidative stress by regulating antioxidant enzymes.

Other pathways have been suggested for Cdk5 hyperactivity in PD pathogenesis involving mitochondrial defects and Parkin dysfunction. In a non-human primate model of PD, aberrant Cdk5 is proposed to phosphorylate Drp1 at S616, which accelerates mitochondrial dysfunction and neurotoxicity [ 121 ]. Cdk5 also regulates PD pathogenesis through Parkin, an E3 ubiquitin ligase. Parkin dysfunction is thought to govern PD progression, as Parkin mutations have been identified in patients with autosomal recessive form of PD. Reduced Parkin activity and the presence of Parkin aggregates in the Lewy body are evident in PD human brains [ 136 ]. Several studies show that the phosphorylation of Parkin at S131 by Cdk5 notably decreases its ubiquitin ligase activity and increases Parkin aggregation [ 137 ], subsequently leading to neurotoxicity. Together, these findings emphasize how Cdk5 hyperactivation impacts PD pathogenesis through oxidative stress, mitochondrial defects, and Parkin dysfunction.

Amyotrophic lateral sclerosis

Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease that results in selective loss of motor neurons in the spinal cord, brainstem and cerebral cortex. Patients with ALS suffer from severe motor deficits and paralysis, which lead to death within years after disease onset [ 138 ]. Marked Cdk5 immunoreactivity was observed in degenerating neurons in spinal cord of patients with sporadic ALS, as well as in a familial ALS case harboring mutant superoxide dismutase type 1 ( SOD1 ) gene [ 139 ]. Moreover, p25 accumulation and Cdk5 hyperactivity have been shown in the spinal cord of SOD1 G93A mouse model of ALS, along with the hyperphosphorylation of tau and neurofilament (NF) [ 140 , 141 ]. Interestingly, in the two lines of SOD1 G37R mice that exhibit different disease severity, Cdk5 activity correlates with lethality [ 140 ]. Conversely, inhibiting Cdk5 hyperactivity improves motor deficits, delays pathology, and extends survival in SOD1 G93A mice [ 141 ].

Huntington’s disease

In Huntington’s disease (HD), expansion of a polyglutamine domain of the huntingtin (Htt) protein leads to Htt aggregation and selective loss of medium spiny neurons in the striatum [ 142 ]. Brain extracts from an HD rat model display a greater calpain activity (which increases p25 levels), and upregulated p25 levels have been shown in both cellular and rodent model of HD [ 143 , 144 ]. Pharmacological inhibition of Cdk5 by roscovitine decreases DARPP-32 phosphorylation at T75, which has been linked to stabilize dendritic spines and attenuate depressive-like behavior in the HD mouse model [ 145 ]. Notably, converging evidence also underscores the neuroprotective roles for Cdk5 in HD. Mutant Htt induces neurotoxicity by the release of toxic peptide fragments containing the poly-Q expansion after proteolytic cleavage. Remarkably, Cdk5 phosphorylation of Htt at S434 residue offers protection against Htt cleavage, aggregation, and subsequent toxicity [ 146 ]. Moreover, upon DNA damage, Cdk5 phosphorylates Htt at S1181 and S1201 [ 147 ]. Phospho-deficient mutants harbor exacerbated DNA damage-induced neurotoxicity, whereas phospho-mimic mutants prevent neuronal death in striatal neurons expressing mutant Htt [ 147 ]. These findings suggest the multifaceted modulation of Cdk5 in the development of HD.

HIV dementia

Infection with the human immunodeficiency virus (HIV) compromises the immune system and causes a number of physiological disruptions, including cognitive impairment [ 148 ]. Elevated levels of Cdk5 and hyperphosphorylation of tau have been observed in brain samples from patients with HIV encephalitis [ 149 ], a subset of HIV patients that exhibit more profound cognitive alterations and neurodegeneration. These findings indicate a possible role for Cdk5 hyperactivation in neurological disorders of HIV patients. In an in vitro model of HIV neurotoxicity, where cultured cortical neurons are exposed to supernatants from primary human HIV-infected macrophages, increased p25 generation and Cdk5 hyperactivity correlate with neurotoxicity [ 150 ]. Importantly, attenuating Cdk5 activity using the small molecule Cdk5 inhibitor roscovitine or the calpain inhibitor MDL28170 promotes neuronal viability in this model [ 150 ]. Furthermore, pharmacological inhibition of Cdk5 by roscovitine reduces tau phosphorylation, decreases neurodegeneration, and improves memory function in an HIV mouse model [ 149 , 151 ]. Together, these observations suggest Cdk5 hyperactivation may contribute to cognitive decline in HIV patients.

Diabetes-associated degeneration, insulin secretion, and insulin sensitivity

Recent reports strengthen the link between Cdk5 hyperactivation and diabetes-associated neurodegeneration. Diabetes mellitus is a prevalent metabolic disorder characterized by hyperglycemia, which results in insulin resistance and insufficient insulin secretion due to the failure of β-pancreatic cells [ 152 ]. High glucose damages a wide variety of tissue and organs, including the nervous system. Epidemiologic studies indicate that diabetes is also associated with higher rates of cognitive impairment [ 153 ]. High glucose exposure leads to p25 generation, Cdk5 hyperactivation, and tau hyperphosphorylation in cultured neurons [ 154 ]. Experimentally induced diabetes in animals through the administration of β-cytotoxic drugs such as streptozotocin (STZ) is well-characterized. Compared to the vehicle-treated control mice, STZ-induced diabetic mice show neuronal death and cognitive impairment [ 155 ]. Importantly, treatment of roscovitine, an inhibitor of Cdk5, reduces cell death and tau hyperphosphorylation in cultured cells exposed to STZ [ 125 ].

Interestingly, there is an emerging role for non-neuronal Cdk5 in insulin secretion and insulin sensitivity. Upon glucose stimulation, salt inducible kinase 2 (SIK2) phosphorylates p35 at Ser 91 residue, which promotes p35 protein degradation via the ubiquitin-proteosome pathway [ 156 ] and thus reduces Cdk5 activity. Blockade of Cdk5 function relieves inhibitory phosphorylation of L-type voltage-dependent calcium channel (L-VDCC) at Ser783, leading to calcium influx and insulin secretion in β cells [ 157 , 158 ].

On the other hand, Cdk5 modulates insulin sensitivity through peroxisome proliferator-activated receptor γ (PPARγ) in adipocytes. PPARγ is a ligand-activated transcription factor that belongs to nuclear receptor superfamily [ 159 ]. PPARγs highly expressed in adipose tissue and controls insulin sensitivity [ 159 ]. PPARγ is a Cdk5 target, and NCoR acts as an adaptor protein facilitating the ability of Cdk5 to associate with and phosphorylate PPARγ [ 160 , 161 ]. Cdk5-mediated phosphorylation of PPARγ at Ser 273 increases the interaction of PPARγ with thyroid hormone receptor-associated protein 3 (THRAP3), which downregulates the expression of adiponectin and adipsin [ 162 ]. Adiponectin and adipsin are key adipokines that enhance insulin sensitivity. Phosphorylation of PPARγ (Ser 273) results in an increase in adipocyte tissues of mice on high-fat diet, and a decrease in adiponectin and adipsin causes obesity-induced insulin resistance [ 161 ]. Thus, these findings reveal how Cdk5 and diabetes may be linked pathologically through its multifaceted functions in neurons, pancreatic β cells, and adipocytes.

Cdk5 as a target for disease treatment

Small molecule inhibitors of cdk5.

Modulating the aberrant activity of Cdk5 has attracted attention as a therapeutic target for neurodegenerative diseases. Multiple synthetic inhibitors of Cdk5 have been discovered, and most of them (roscovitine, olomoucine, and purvalanol-A) are purine derivates that share the basic ring structure of ATP [ 163 ]. Mechanistically, these small molecules compete with ATP for docking at the ATP-binding site of Cdk5. The lack of selectivity is a common issue for ATP competitive inhibitors, as the ATP-binding site is a conserved feature among Cdk members. Most Cdk5 inhibitors target a broad-range of Cdk members with various efficacies. Roscovitine shows increased selectivity for Cdk5 over other Cdks, and it is the most widely studied Cdk5 inhibitor in the field. The IC 50 of roscovitine for Cdk5/p25 complex is 0.16 μM, whereas a higher concentration is needed to reach the same inhibition on other Cdks [ 163 ]. Despite the lack of selectivity for Cdk5, roscovitine induces beneficial effects in various cellular and mouse models involving Cdk5 hyperactivation. Roscovitine attenuates Cdk5 hyperactivity and ameliorates p25-associated pathologies, such as DNA damage, cell cycle reentry, tau phosphorylation, and neuronal death [ 115 , 116 , 125 , 147 , 150 ]. Identifying compounds that more specifically target Cdk5 kinase activity without interfering with the ATP pocket of other Cdks is an alternative strategy to modulate Cdk5 function as a therapeutic intervention.

Peptide inhibitors of Cdk5

Peptide inhibitors are a promising approach for targeting Cdk5. Inhibitory peptides are typically derived from sequences of native proteins mediating protein–protein interactions, which contains a small number of key residues. Inhibitory peptides act as dominant-negative forms of the endogenous proteins, and have a greater efficacy and specificity than synthetic inhibitors [ 164 ].

CIP is the first identified peptide inhibitor of Cdk5. CIP is a 126 residue-long peptide that originally derived from the C-terminal of p35 (amino residues 154 to 279), within the region that is necessary for p35 to activate Cdk5 [ 165 , 166 ]. CIP inhibits recombinant Cdk5 kinase activity while histone H1 was used as the substrate. In addition, CIP has no effect on endogenous Cdc2 activity [ 165 ]. In HEK cells expressing Cdk5/p25, co-transfection of CIP reduces Cdk5 activity and tau hyperphosphorylation [ 165 ]. Moreover, CIP notably attenuates p25-associated pathological phenotypes in animals. In the p25 transgenic mouse model, overexpressing CIP attenuates Cdk5 hyperactivation and reduces the accumulation of Aβ and phosphorylated tau levels [ 167 ]. CIP overexpression also rescues neuronal loss and improves memory function in p25 transgenic mice [ 167 ]. Similarly, overexpressing CIP in the SOD1 G37R mouse model of ALS improves motor deficits and delays neurological pathology [ 141 ]. A 24-residue peptide called p5, spanning CIP residues Lys 245 –Ala 277 , inhibits Cdk5 comparably to CIP [ 168 ]. In Aβ 42 -treated cultured neurons, overexpression of P5 inhibits Cdk5 aberrant activity, phosphorylation of tau, and neuronal death [ 168 ]. Importantly, intraperitoneal delivery of a modified P5 (TFP5) rescues p25-associated pathologies in animal models [ 154 , 169 , 170 , 171 ].

A recent study revealed an inhibitory peptide targeting Cdk5 named Cdk5i 172 . Cdk5i is a 12-amino acid peptide derived from the T-loop of Cdk5 (A 148 RAFGIPVRCYS 159 ) [ 172 ], a critical region for its interaction with p25 conserved across species but distinct from the T-loop of other Cdks. Biochemical analysis demonstrates that Cdk5i binds to recombinant Cdk5/p25 complex and inhibits Cdk5/p25 kinase activity. Moreover, Cdk5i attenuates the activity of Cdk5 purified from tau P301S mouse brain. In the wild-type animals, Cdk5i has no effect on Cdk5 and Cdk2 activity [ 172 ]. In the same study, Cdk5i was modified to be conjugated with a FITC for microscopic visualization and a TAT sequence to increase the cell/brain penetration [ 172 ]. In the cellular models of tauopathy, this modified Cdk5i significantly decreased the phosphorylation of tau at several residues. Importantly, modified Cdk5i is brain-penetrant, evident by marked FITC signals in mouse brain after a single intraperitoneal injection. Treatment of modified Cdk5i ameliorated DNA damage and gliosis in CK-p25 mice after acute p25 induction [ 172 ]. Together, CIP/TFP5 and Cdk5i offer an exciting approach to attenuate aberrant Cdk5 activity observed in a number of neurological disorders and thereby alleviate pathologies and improve cognitive functions.

Non-neuronal functions of Cdk5

Despite the major role of Cdk5 in neurons, growing evidence suggests non-neuronal functions of Cdk5. In addition to insulin secretion and insulin sensitivity highlighted above, Cdk5’s role in other non-neuronal functions, especially its regulation in astrocyte activation, oligodendrocyte maturation/myelination, T cell activation, and cancer biology will be discussed below.

Astrocyte activation

Astrocytes regulate brain homeostasis and provide metabolic/structural support to neurons and synapses. Both p35 expression and Cdk5 activity increase in astrocytes injured by in vitro scratch assays, a condition that strongly activates astrocytes [ 173 ]. Impairment of Cdk5 function in astrocytes delays wound healing by inhibiting the reorganization of tubulin/GFAP and the extension of hypertrophic processes [ 173 ]. Furthermore, knockdown of Cdk5 (Cdk5-KD) in astrocytes elicits neuroprotection in astrocyte-neuron co-cultures. Cdk5-KD astrocytes secretes more BNDF and are resistant to glutamate-induced gliotoxicity [ 174 ]. In an animal model of cerebral ischemia, transplantation of Cdk5-KD astrocytes facilitates the recovery of neurovascular unit integrity and improves neurological performance [ 175 , 176 ].

Oligodendrocyte maturation/myelination

Oligodendrocyte precursor cells (OPCs) differentiate into oligodendrocytes to form myelin, which ensures the rapid propagation of action potentials. Cdk5 activity increases during OPC differentiation [ 177 ]. Overexpression of Cdk5 promotes OPC maturation and process outgrowth by phosphorylating the focal adhesion protein paxillin at Ser 244 residue [ 177 , 178 ]. Conversely, blocking Cdk5 function by genetic or pharmacological approach perturbs OPC maturation and myelination [ 177 , 178 , 179 ]. Additionally, Cdk5 regulates myelin repair capacity. Injection of lysolecithin (LPC) causes local demyelination and subsequent remyelination. Ablating Cdk5 in OPCs or localized pharmacological inhibition of Cdk5 diminishes myelin repair and the level of remyelinated axons in LPC-lesioned animal model [ 180 ].

T cell activation

Cdk5 regulates T cell activation. T cell receptor (TCR) stimulation induces T lymphocyte activation, which is accompanied by a rapid induction of Cdk5/p35 expression and Cdk5 activity [ 181 ]. Disruption of Cdk5 activity abrogates TCR-mediated lymphocyte activation [ 181 ]. Moreover, chimeric mice lacking Cdk5 gene expression in hematopoietic cells are resistant to the induction of experimental autoimmune encephalomyelitis (EAE), a T cell-mediated autoimmune pre-clinical model of multiple sclerosis [ 181 ]. In lymphocytes, Cdk5 phosphorylates an actin binding protein Coronin 1a at the Thr 418 residue, which facilitates actin polarization and the migration of lymphocytes to chemokine signals in the EAE model [ 181 ].

Cancer biology

Cdk5 and p35/p39 are expressed in clinical tumor specimens [ 182 ]. Patients with higher Cdk5 expression exhibit worse clinical outcomes whereas blocking Cdk5 activity confers protection in cellular and animal models of cancer [ 182 , 183 , 184 ]. Cdk5 contributes to tumorigenesis in various types of cancer. In CNS tumors, Cdk5 increases self-renewal of brain tumor stem cells through the activation of CREB1 [ 185 ]. Thus, inhibiting Cdk5 suppresses tumor growth and improves survival. It is known that binding of programmed death ligand-1 (PD-L1) to programmed cell death protein-1 (PD-1) hinders T-cell function and thus augments cancer immune evasion. Cdk5 promotes immune evasion of medulloblastoma by upregulating PD-L1 expression, and blocking Cdk5 enhances immune sensitivity [ 182 ]. In glioblastoma, Cdk5 phosphorylates a ubiquitin E3 ligase TRIM59 at Ser 308, which subsequently induces the degradation of macroH2A1, a tumor suppressive histone variant [ 186 ]. Moreover, Cdk5 promotes melanoma cell extravasation. Cdk5-mediated phosphorylation of vimentin at Ser 56 triggers depolymerization of vimentin filaments [ 184 ], thereby allowing cell migration. While Cdk5 plays a role in promoting tumorigenesis, there is also evidence supporting its anti-tumor effects. Cdk5 phosphorylates and activates the tumor suppressor DLC1, thus reducing the size of xenografted tumor in mice [ 187 ]. Nonetheless, DLC1 is down-regulated in a wide variety of tumors [ 187 ], which likely results in accentuating the pro-oncogenic activities of Cdk5.

Conclusions

Three decades ago, Cdk5 was discovered based on its high similarity to Cdc2 and was predicted to function in cell cycle regulation. With the remarkable progress through the years, significant insights have unveiled the role of Cdk5 in normal brain function and neurodegeneration. Hyperactivation of Cdk5 appears to be a common theme among different neurodegenerative diseases, and blocking Cdk5 hyperactivity attenuates disease progression. Single-cell transcriptomic analyses reveal the noticeable expression of Cdk5 and p35 in non-neuronal cell types, suggesting an exciting area of Cdk5 biology remains to be explored. Future studies will define more precisely how Cdk5 hyperactivation impairs brain function, will advance our understanding of Cdk5 in non-neuronal cell types, and will generate fundamentally novel therapeutic opportunities aimed at reining in Cdk5 aberrant activity implicated in many neurological disorders.

Availability of data and materials

Not applicable.

Meyerson M, et al. A family of human cdc2-related protein kinases. EMBO J. 1992. https://doi.org/10.1002/j.1460-2075.1992.tb05360.x .

Article   PubMed   PubMed Central   Google Scholar  

Lew J, Beaudette K, Litwin CME, Wang JH. Purification and characterization of a novel proline-directed protein kinase from bovine brain. J Biol Chem. 1992. https://doi.org/10.1016/s0021-9258(18)42222-3 .

Article   PubMed   Google Scholar  

Hellmich MR, Pant HC, Wada E, Battey JF. Neuronal cdc2-like kinase: a cdc2-related protein kinase with predominantly neuronal expression. Proc Natl Acad Sci USA. 1992. https://doi.org/10.1073/pnas.89.22.10867 .

Enserink JM, Kolodner RD. An overview of Cdk1-controlled targets and processes. Cell Div. 2010. https://doi.org/10.1186/1747-1028-5-11 .

Tsai LH, Takahashi T, Caviness VS, Harlow E. Activity and expression pattern of cyclin-dependent kinase 5 in the embryonic mouse nervous system. Development. 1993. https://doi.org/10.1242/dev.119.4.1029 .

Malumbres M, et al. Cyclin-dependent kinases: a family portrait. Nat Cell Biol. 2009. https://doi.org/10.1038/ncb1109-1275 .

Tsai LH, Delalle I, Caviness VS, Chae T, Harlow E. p35 is a neural-specific regulatory subunit of cyclin-dependent kinase 5. Nature. 1994. https://doi.org/10.1038/371419a0 .

Lew J, et al. A brain-specific activator of cyclin-dependent kinase 5. Nature. 1994. https://doi.org/10.1038/371423a0 .

Ishiguro K, et al. Identification of the 23 kDa subunit of tau protein kinase II as a putative activator of cdk5 in bovine brain. FEBS Lett. 1994. https://doi.org/10.1016/0014-5793(94)80501-6 .

Ximerakis M, et al. Single-cell transcriptomic profiling of the aging mouse brain. Nat Neurosci. 2019. https://doi.org/10.1038/s41593-019-0491-3 .

Tang D, et al. An isoform of the neuronal cyclin-dependent kinase 5 (Cdk5) activator. J Biol Chem. 1995. https://doi.org/10.1074/jbc.270.45.26897 .

Humbert S, Dhavan R, Tsai LH. p39 activates cdk5 in neurons, and is associated with the actin cytoskeleton. J Cell Sci. 2000. https://doi.org/10.1242/jcs.113.6.975 .

Ko J, et al. p35 and p39 are essential for cyclin-dependent kinase 5 function during neurodevelopment. J Neurosci. 2001. https://doi.org/10.1523/jneurosci.21-17-06758.2001 .

Takahashi S, Saito T, Hisanaga SI, Pant HC, Kulkarni AB. Tau phosphorylation by cyclin-dependent kinase 5/p39 during brain development reduces its affinity for microtubules. J Biol Chem. 2003. https://doi.org/10.1074/jbc.M211964200 .

Minegishi S, et al. Membrane association facilitates degradation and cleavage of the cyclin-dependent kinase 5 activators p35 and p39. Biochemistry. 2010. https://doi.org/10.1021/bi100631f .

Yamada M, et al. Cdk5-p39 is a labile complex with the similar substrate specificity to Cdk5-p35. J Neurochem. 2007. https://doi.org/10.1111/j.1471-4159.2007.04505.x .

Ouyang L, et al. p39-associated Cdk5 activity regulates dendritic morphogenesis. Sci Rep. 2020. https://doi.org/10.1038/s41598-020-75264-6 .

Ito Y, et al. Preferential targeting of p39-activated Cdk5 to Rac1-induced lamellipodia. Mol Cell Neurosci. 2014. https://doi.org/10.1016/j.mcn.2014.05.006 .

Niethammer M, et al. NUDEL is a novel Cdk5 substrate that associates with LIS1 and cytoplasmic dynein. Neuron. 2000. https://doi.org/10.1016/S0896-6273(00)00147-1 .

Patrick GN, et al. Conversion of p35 to p25 deregulates Cdk5 activity and promotes neurodegeneration. Nature. 1999. https://doi.org/10.1038/45159 .

Asada A, et al. Myristoylation of p39 and p35 is a determinant of cytoplasmic or nuclear localization of active cycline-dependent kinase 5 complexes. J Neurochem. 2008. https://doi.org/10.1111/j.1471-4159.2008.05500.x .

Zukerberg LR, et al. Cables links Cdk5 and c-Abl and facilitates Cdk5 tyrosine phosphorylation, kinase upregulation, and neurite outgrowth. Neuron. 2000. https://doi.org/10.1016/S0896-6273(00)81200-3 .

Matsuura I, Wang JH. Demonstration of cyclin-dependent kinase inhibitory serine/threonine kinase in bovine thymus. J Biol Chem. 1996. https://doi.org/10.1074/jbc.271.10.5443 .

Sharma P, Sharma M, Amin ND, Albers RW, Pant HC. Regulation of cyclin-dependent kinase 5 catalytic activity by phosphorylation. Proc Natl Acad Sci USA. 1999. https://doi.org/10.1073/pnas.96.20.11156 .

Lee J, et al. The acetylation of cyclin-dependent kinase 5 at lysine 33 regulates kinase activity and neurite length in hippocampal neurons. Sci Rep. 2018. https://doi.org/10.1038/s41598-018-31785-9 .

Zhang P, et al. S-nitrosylation of cyclin-dependent kinase 5 (Cdk5) regulates its kinase activity and dendrite growth during neuronal development. J Neurosci. 2010. https://doi.org/10.1523/JNEUROSCI.3899-10.2010 .

Zhang P, Fu WY, Fu AKY, Ip NY. S-nitrosylation-dependent proteasomal degradation restrains Cdk5 activity to regulate hippocampal synaptic strength. Nat Commun. 2015. https://doi.org/10.1038/ncomms9665 .

Dhavan R, Tsai LH. A decade of CDK5. Nat Rev Mol Cell Biol. 2001. https://doi.org/10.1038/35096019 .

Su SC, Tsai LH. Cyclin-dependent kinases in brain development and disease. Annu Rev Cell Dev Biol. 2011. https://doi.org/10.1146/annurev-cellbio-092910-154023 .

Lai K-O. Cdk5: a key player at neuronal synapse with diverse functions. Mini-Rev Med Chem. 2015. https://doi.org/10.2174/1389557515666150324122321 .

Lai KO, Ip NY. Recent advances in understanding the roles of Cdk5 in synaptic plasticity. Biochem Biophys Acta. 2009. https://doi.org/10.1016/j.bbadis.2009.05.001 .

Cheng K, Ip NY. Cdk5: a new player at synapses. Neurosignals. 2003. https://doi.org/10.1159/000074619 .

Cheung ZH, Fu AKY, Ip NY. Synaptic roles of Cdk 5: implications in higher cognitive functions and neurodegenerative diseases. Neuron. 2006. https://doi.org/10.1016/j.neuron.2006.02.024 .

Ye T, Fu AKY, Ip NY. Cyclin-dependent kinase 5 in axon growth and regeneration. Int Rev Neurobiol. 2012. https://doi.org/10.1016/B978-0-12-398309-1.00006-8 .

Cheung ZH, Ip NY. The roles of cyclin-dependent kinase 5 in dendtrite and synapse development. Biotechnol J. 2007. https://doi.org/10.1002/biot.200700056 .

Kesavapany S, Li BS, Pant HC. Cyclin-dependent kinase 5 in neurofilament function and regulation. Neurosignals. 2003. https://doi.org/10.1159/000074627 .

McLinden A. At the fulcrum in health and disease: Cdk5 and the balancing acts of neuronal structure and physiology. Brain Disord Ther. 2012. https://doi.org/10.4172/2168-975x.s1-001 .

Gilmore EC, Ohshima T, Goffinet AM, Kulkarni AB, Herrup K. Cyclin-dependent kinase 5-deficient mice demonstrate novel developmental arrest in cerebral cortex. J Neurosci. 1998. https://doi.org/10.1523/jneurosci.18-16-06370.1998 .

Ohshima T, et al. Migration defects of cdk5(−/−) neurons in the developing cerebellum is cell autonomous. J Neurosci. 1999. https://doi.org/10.1523/jneurosci.19-14-06017.1999 .

Ohshima T, et al. Targeted disruption of the cyclin-dependent kinase 5 gene results in abnormal corticogenesis, neuronal pathology and perinatal death. Proc Natl Acad Sci USA. 1996. https://doi.org/10.1073/pnas.93.20.11173 .

Kwon YT, Gupta A, Zhou Y, Nikolic M, Tsai LH. Regulation of N-cadherin-mediated adhesion by the p35-Cdk5 kinase. Curr Biol. 2000. https://doi.org/10.1016/S0960-9822(00)00411-5 .

Ye T, Ip JPK, Fu AKY, Ip NY. Cdk5-mediated phosphorylation of RapGEF2 controls neuronal migration in the developing cerebral cortex. Nat Commun. 2014. https://doi.org/10.1038/ncomms5826 .

Perlini LE, et al. Synapsin III acts downstream of semaphorin 3A/CDK5 signaling to regulate radial migration and orientation of pyramidal neurons in vivo. Cell Rep. 2015. https://doi.org/10.1016/j.celrep.2015.03.022 .

Sasaki S, et al. A LIS1/NUDEL/cytoplasmic dynein heavy chain complex in the developing and adult nervous system. Neuron. 2000. https://doi.org/10.1016/S0896-6273(00)00146-X .

Xie Z, Sanada K, Samuels BA, Shih H, Tsai LH. Serine 732 phosphorylation of FAK by Cdk5 is important for microtubule organization, nuclear movement, and neuronal migration. Cell. 2003. https://doi.org/10.1016/S0092-8674(03)00605-6 .

Ishizuka K, et al. DISC1-dependent switch from progenitor proliferation to migration in the developing cortex. Nature. 2011. https://doi.org/10.1038/nature09859 .

Tanaka T, et al. Cdk5 phosphorylation of doublecortin Ser297 regulates its effect on neuronal migration. Neuron. 2004. https://doi.org/10.1016/S0896-6273(03)00852-3 .

Singh KK, et al. Dixdc1 is a critical regulator of disc1 and embryonic cortical development. Neuron. 2010. https://doi.org/10.1016/j.neuron.2010.06.002 .

Nishimura YV, et al. Cdk5 and its substrates, Dcx and p27kip1, regulate cytoplasmic dilation formation and nuclear elongation in migrating neurons. Development. 2014. https://doi.org/10.1242/dev.111294 .

Nikolic M, Dudek H, Kwon YT, Ramos YFM, Tsai LH. The cdk5/p35 kinase is essential for neurite outgrowth during neuronal differentiation. Genes Dev. 1996. https://doi.org/10.1101/gad.10.7.816 .

Connell-Crowley L, Le Gall M, Vo DJ, Giniger E. The cyclin-dependent kinase Cdk5 controls multiple aspects of axon patterning in vivo. Curr Biol. 2000. https://doi.org/10.1016/S0960-9822(00)00487-5 .

Chae T, et al. Mice lacking p35, a neuronal specific activator of Cdk5, display cortical lamination defects, seizures, and adult lethality. Neuron. 1997. https://doi.org/10.1016/S0896-6273(01)80044-1 .

Nikolic M, Chou MM, Lu W, Mayer BJ, Tsai LH. The p35/Cdk5 kinase is a neuron-specific Rac effector that inhibits Pak1 activity. Nature. 1998. https://doi.org/10.1038/26034 .

Duhr F, et al. Cdk5 induces constitutive activation of 5-HT 6 receptors to promote neurite growth. Nat Chem Biol. 2014. https://doi.org/10.1038/nchembio.1547 .

Furusawa K, et al. Cdk5 regulation of the GRAB-mediated Rab8-Rab11 cascade in axon outgrowth. J Neurosci. 2017. https://doi.org/10.1523/JNEUROSCI.2197-16.2016 .

Paglini G, et al. Evidence for the participation of the neuron-specific CDK5 activator p35 during laminin-enhanced axonal growth. J Neurosci. 1998. https://doi.org/10.1523/jneurosci.18-23-09858.1998 .

Fang WQ, et al. Cdk5-mediated phosphorylation of Axin directs axon formation during cerebral cortex development. J Neurosci. 2011. https://doi.org/10.1523/JNEUROSCI.3120-11.2011 .

Ohshima T, et al. Impairment of hippocampal long-term depression and defective spatial learning and memory in p35-/- mice. J Neurochem. 2005. https://doi.org/10.1111/j.1471-4159.2005.03233.x .

Citri A, Malenka RC. Synaptic plasticity: multiple forms, functions, and mechanisms. Neuropsychopharmacology. 2008. https://doi.org/10.1038/sj.npp.1301559 .

Shuang R, et al. Regulation of Munc-18/syntaxin 1A interaction by cyclin-dependent kinase 5 in nerve endings. J Biol Chem. 1998. https://doi.org/10.1074/jbc.273.9.4957 .

Flute SR, et al. Amphiphysin 1 binds the cyclin-dependent kinase (cdk) 5 regulatory subunit p35 and is phosphorylated by cdk5 and cdc2. J Biol Chem. 2001. https://doi.org/10.1074/jbc.M008932200 .

Article   Google Scholar  

Tomizawa K, et al. Cophosphorylation of amphiphysin I and dynamin I by Cdk5 regulates clathrin-mediated endocytosis of synaptic vesicles. J Cell Biol. 2003. https://doi.org/10.1083/jcb.200308110 .

Su SC, et al. Regulation of N-type voltage-gated calcium channels and presynaptic function by cyclin-dependent kinase 5. Neuron. 2012. https://doi.org/10.1016/j.neuron.2012.06.023 .

Tomizawa K, et al. Cdk5/p35 regulates neurotransmitter release through phosphorylation and downregulation of P/Q-type voltage-dependent calcium channel activity. J Neurosci. 2002. https://doi.org/10.1523/jneurosci.22-07-02590.2002 .

Li BS, et al. Regulation of NMDA receptors by cyclin-dependent kinase-5. Proc Natl Acad Sci USA. 2001. https://doi.org/10.1073/pnas.211428098 .

Wang J, Liu SH, Fu YP, Wang JH, Lu YM. Cdk5 activation induces hippocampal CA1 cell death by directly phosphorylating NMDA receptors. Nat Neurosci. 2003. https://doi.org/10.1038/nn1119 .

Morabito MA, Sheng M, Tsai LH. Cyclin-dependent kinase 5 phosphorylates the N-terminal domain of the postsynaptic density protein PSD-95 in neurons. J Neurosci. 2004. https://doi.org/10.1523/JNEUROSCI.4582-03.2004 .

Bianchetta MJ, Lam TKT, Jones SN, Morabito MA. Cyclin-dependent kinase 5 regulates PSD-95 ubiquitination in neurons. J Neurosci. 2011. https://doi.org/10.1523/JNEUROSCI.2388-11.2011 .

Huang H, et al. Cdk5-dependent phosphorylation of liprinα1 mediates neuronal activity-dependent synapse development. Proc Natl Acad Sci USA. 2017. https://doi.org/10.1073/pnas.1708240114 .

Lai KO, et al. TrkB phosphorylation by Cdk5 is required for activity-dependent structural plasticity and spatial memory. Nat Neurosci. 2012. https://doi.org/10.1038/nn.3237 .

Liang Z, et al. The pseudokinase CaMKv is required for the activity-dependent maintenance of dendritic spines. Nat Commun. 2016. https://doi.org/10.1038/ncomms13282 .

Huber KM. Synaptic homeostasis: quality vs. quantity. Nat Neurosci. 2018. https://doi.org/10.1038/s41593-018-0159-4 .

Fernández-Alfonso T, Ryan TA. The efficiency of the synaptic vesicle cycle at central nervous system synapses. Trends Cell Biol. 2006. https://doi.org/10.1016/j.tcb.2006.06.007 .

Kim SH, Ryan TA. CDK5 serves as a major control point in neurotransmitter release. Neuron. 2010. https://doi.org/10.1016/j.neuron.2010.08.003 .

Seeburg DP, Feliu-Mojer M, Gaiottino J, Pak DTS, Sheng M. Critical role of CDK5 and polo-like kinase 2 in homeostatic synaptic plasticity during elevated activity. Neuron. 2008. https://doi.org/10.1016/j.neuron.2008.03.021 .

Seeburg DP, Pak D, Sheng M. Polo-like kinases in the nervous system. Oncogene. 2005. https://doi.org/10.1038/sj.onc.1208277 .

Takahashi JS. Transcriptional architecture of the mammalian circadian clock. Nat Rev Genet. 2017. https://doi.org/10.1038/nrg.2016.150 .

Videnovic A, Lazar AS, Barker RA, Overeem S. ‘The clocks that time us’ - Circadian rhythms in neurodegenerative disorders. Nat Rev Neurol. 2014. https://doi.org/10.1038/nrneurol.2014.206 .

Brenna A, et al. Cyclin dependent kinase 5 (Cdk5) regulates the circadian clock. Elife. 2019. https://doi.org/10.7554/eLife.50925 .

Kwak Y, et al. Cyclin-dependent kinase 5 (Cdk5) regulates the function of CLOCK protein by direct phosphorylation. J Biol Chem. 2013. https://doi.org/10.1074/jbc.M113.494856 .

Cheung ZH, Ip NY. Cdk5: a multifaceted kinase in neurodegenerative diseases. Trends Cell Biol. 2012. https://doi.org/10.1016/j.tcb.2011.11.003 .

Lee MS, et al. Neurotoxicity induces cleavage of p35 to p25 by calpain. Nature. 2000. https://doi.org/10.1038/35012636 .

Cheung ZH, Ip NY. Cdk5: mediator of neuronal death and survival. Neurosci Lett. 2004. https://doi.org/10.1016/j.neulet.2003.12.117 .

Ahlijanian MK, et al. Hyperphosphorylated tau and neurofilament and cytoskeletal disruptions in mice overexpressing human p25, an activator of cdk5. Proc Natl Acad Sci USA. 2000. https://doi.org/10.1073/pnas.040577797 .

Cruz JC, Tseng HC, Goldman JA, Shih H, Tsai LH. Aberrant Cdk5 activation by p25 triggers pathological events leading to neurodegeneration and neurofibrillary tangles. Neuron. 2003. https://doi.org/10.1016/S0896-6273(03)00627-5 .

Kim D, et al. Deregulation of HDAC1 by p25/Cdk5 in neurotoxicity. Neuron. 2008. https://doi.org/10.1016/j.neuron.2008.10.015 .

Yang Y, Geldmacher DS, Herrup K. DNA replication precedes neuronal cell death in Alzheimer’s disease. J Neurosci. 2001. https://doi.org/10.1523/jneurosci.21-08-02661.2001 .

Mathys H, et al. Temporal tracking of microglia activation in neurodegeneration at single-cell resolution. Cell Rep. 2017. https://doi.org/10.1016/j.celrep.2017.09.039 .

Sundaram JR, et al. Cdk5/p25-induced cytosolic PLA2-mediated lysophosphatidylcholine production regulates neuroinflammation and triggers neurodegeneration. J Neurosci. 2012. https://doi.org/10.1523/JNEUROSCI.5177-11.2012 .

Fischer A, Sananbenesi F, Pang PT, Lu B, Tsai LH. Opposing roles of transient and prolonged expression of p25 in synaptic plasticity and hippocampus-dependent memory. Neuron. 2005. https://doi.org/10.1016/j.neuron.2005.10.033 .

Shukla AK, Spurrier J, Kuzina I, Giniger E. Hyperactive innate immunity causes degeneration of dopamine neurons upon altering activity of Cdk5. Cell Rep. 2019. https://doi.org/10.1016/j.celrep.2018.12.025 .

Deture MA, Dickson DW. The neuropathological diagnosis of Alzheimer’s disease. Mol Neurodegener. 2019. https://doi.org/10.1186/s13024-019-0333-5 .

Shukla V, Skuntz S, Pant HC. Deregulated Cdk5 activity is involved in inducing Alzheimer’s disease. Arch Med Res. 2012. https://doi.org/10.1016/j.arcmed.2012.10.015 .

Tseng HC, Zhou Y, Shen Y, Tsai LH. A survey of Cdk5 activator p35 and p25 levels in Alzheimer’s disease brains. FEBS Lett. 2002. https://doi.org/10.1016/S0014-5793(02)02934-4 .

Alvarez A, Toro R, Cáceres A, Maccioni RB. Inhibition of tau phosphorylating protein kinase cdk5 prevents β-amyloid-induced neuronal death. FEBS Lett. 1999. https://doi.org/10.1016/S0014-5793(99)01279-X .

Oakley H, et al. Intraneuronal β-amyloid aggregates, neurodegeneration, and neuron loss in transgenic mice with five familial Alzheimer’s disease mutations: potential factors in amyloid plaque formation. J Neurosci. 2006. https://doi.org/10.1523/JNEUROSCI.1202-06.2006 .

Chow HM, et al. CDK5 activator protein p25 preferentially binds and activates GSK3β. Proc Natl Acad Sci USA. 2014. https://doi.org/10.1073/pnas.1402627111 .

Chow VW, Mattson MP, Wong PC, Gleichmann M. An overview of APP processing enzymes and products. NeuroMol Med. 2010. https://doi.org/10.1007/s12017-009-8104-z .

Selkoe DJ. Translating cell biology into therapeutic advances in Alzheimer’s disease. Nature. 1999. https://doi.org/10.1038/399a023 .

Cruz JC, et al. p25/cyclin-dependent kinase 5 induces production and intraneuronal accumulation of amyloid β in vivo. J Neurosci. 2006. https://doi.org/10.1523/JNEUROSCI.3133-06.2006 .

Wen Y, et al. Transcriptional regulation of β-secretase by p25/cdk5 leads to enhanced amyloidogenic processing. Neuron. 2008. https://doi.org/10.1016/j.neuron.2008.02.024 .

Fu AKY, et al. Cyclin-dependent kinase 5 phosphorylates signal transducer and activator of transcription 3 and regulates its transcriptional activity. Proc Natl Acad Sci USA. 2004. https://doi.org/10.1073/pnas.0307606100 .

Lee MS, et al. APP processing is regulated by cytoplasmic phosphorylation. J Cell Biol. 2003. https://doi.org/10.1083/jcb.200301115 .

Seo J, et al. Activity-dependent p25 generation regulates synaptic plasticity and aβ-induced cognitive impairment. Cell. 2014. https://doi.org/10.1016/j.cell.2014.01.065 .

Wang Y, Mandelkow E. Tau in physiology and pathology. Nat Rev Neurosci. 2016. https://doi.org/10.1038/nrn.2015.1 .

Ishiguro K, et al. Tau protein kinase I converts normal tau protein into A68-like component of paired helical filaments. J Biol Chem. 1992. https://doi.org/10.1016/s0021-9258(19)50102-8 .

Kimura T, Ishiguro K, Hisanaga SI. Physiological and pathological phosphorylation of tau by Cdk5. Front Mol Neurosci. 2014. https://doi.org/10.3389/fnmol.2014.00065 .

Neddens J, et al. Phosphorylation of different tau sites during progression of Alzheimer’s disease. Acta Neuropathol Commun. 2018. https://doi.org/10.1186/s40478-018-0557-6 .

Rao MV, et al. Specific calpain inhibition by calpastatin prevents tauopathy and neurodegeneration and restores normal lifespan in tau P301L Mice. J Neurosci. 2014. https://doi.org/10.1523/JNEUROSCI.1132-14.2014 .

Seo J, et al. Inhibition of p25/Cdk5 attenuates tauopathy in mouse and iPSC models of frontotemporal dementia. J Neurosci. 2017. https://doi.org/10.1523/JNEUROSCI.0621-17.2017 .

Adamec E, Vonsattel JP, Nixon RA. DNA strand breaks in Alzheimer’s disease. Brain Res. 1999. https://doi.org/10.1016/S0006-8993(99)02004-1 .

Miller KM, et al. Human HDAC1 and HDAC2 function in the DNA-damage response to promote DNA nonhomologous end-joining. Nat Struct Mol Biol. 2010. https://doi.org/10.1038/nsmb.1899 .

Madabhushi R, Pan L, Tsai LH. DNA damage and its links to neurodegeneration. Neuron. 2014. https://doi.org/10.1016/j.neuron.2014.06.034 .

Kruman II, et al. cell cycle activation linked to neuronal cell death initiated by DNA damage. Neuron. 2004. https://doi.org/10.1016/S0896-6273(04)00017-0 .

Tian B, Yang Q, Mao Z. Phosphorylation of ATM by Cdk5 mediates DNA damage signalling and regulates neuronal death. Nat Cell Biol. 2009. https://doi.org/10.1038/ncb1829 .

Lopes JP, Oliveira CR, Agostinho P. Cdk5 acts as a mediator of neuronal cell cycle re-entry triggered by amyloid-β and prion peptides. Cell Cycle. 2009. https://doi.org/10.4161/cc.8.1.7506 .

Chow HM, et al. Age-related hyperinsulinemia leads to insulin resistance in neurons and cell-cycle-induced senescence. Nat Neurosci. 2019. https://doi.org/10.1038/s41593-019-0505-1 .

Westermann B. Mitochondrial fusion and fission in cell life and death. Nat Rev Mol Cell Biol. 2010. https://doi.org/10.1038/nrm3013 .

Giacomello M, Pyakurel A, Glytsou C, Scorrano L. The cell biology of mitochondrial membrane dynamics. Nat Rev Mol Cell Biol. 2020. https://doi.org/10.1038/s41580-020-0210-7 .

Sun KH, De Pablo Y, Vincent F, Shah K. Deregulated Cdk5 promotes oxidative stress and mitochondrial dysfunction. J Neurochem. 2008. https://doi.org/10.1111/j.1471-4159.2008.05616.x .

Park J, et al. Abnormal mitochondria in a non-human primate model of MPTP-induced Parkinson’s disease: DRp1 and CDK5/p25 signaling. Exp Neurobiol. 2019. https://doi.org/10.5607/en.2019.28.3.414 .

Meuer K, et al. Cyclin-dependent kinase 5 is an upstream regulator of mitochondrial fission during neuronal apoptosis. Cell Death Differ. 2007. https://doi.org/10.1038/sj.cdd.4402087 .

Cherubini M, Puigdellívol M, Alberch J, Ginés S. Cdk5-mediated mitochondrial fission: a key player in dopaminergic toxicity in Huntington’s disease. Biochim Biophys Acta. 2015. https://doi.org/10.1016/j.bbadis.2015.06.025 .

Jahani-Asl A, et al. CDK5 phosphorylates DRP1 and drives mitochondrial defects in NMDA-induced neuronal death. Hum Mol Genet. 2015. https://doi.org/10.1093/hmg/ddv188 .

Park J, et al. Streptozotocin Induces Alzheimer’s disease-like pathology in hippocampal neuronal cells via CDK5/Drp1-mediated mitochondrial fragmentation. Front Cell Neurosci. 2020. https://doi.org/10.3389/fncel.2020.00235 .

Yang SH, Huang CY, Hsieh CY, Chuang JI. CDK4 and CDK5 inhibition have comparable mild hypothermia effects in preventing Drp1-dependent mitochondrial fission and neuron death induced by MPP+. Mol Neurobiol. 2020. https://doi.org/10.1007/s12035-020-02014-0 .

Rong R, et al. Cdk5-mediated Drp1 phosphorylation drives mitochondrial defects and neuronal apoptosis in radiation-induced optic neuropathy. Cell Death Dis. 2020. https://doi.org/10.1038/s41419-020-02922-y .

Poewe W, et al. Parkinson disease. Nat Rev Dis Prim. 2017. https://doi.org/10.1038/nrdp.2017.13 .

Alvira D, et al. Activation of the calpain/cdk5/p25 pathway in the girus cinguli in Parkinson’s disease. Park Relat Disord. 2008. https://doi.org/10.1016/j.parkreldis.2007.09.005 .

Beal MF. Experimental models of Parkinson’s disease. Nat Rev Neurosci. 2001. https://doi.org/10.1038/35072550 .

Smith PD, et al. Cyclin-dependent kinase 5 is a mediator of dopaminergic neuron loss in a mouse model of Parkinson’s disease. Proc Natl Acad Sci USA. 2003. https://doi.org/10.1073/pnas.2232515100 .

He R, et al. Cdk5 inhibitory peptide prevents loss of dopaminergic neurons and alleviates behavioral changes in an MPTP induced Parkinson’s disease mouse model. Front Aging Neurosci. 2018. https://doi.org/10.3389/fnagi.2018.00162 .

Tran J, et al. Therapeutic effects of TP5, a Cdk5/p25 inhibitor, in in vitro and in vivo models of Parkinson’s disease. Curr Res Neurobiol. 2021. https://doi.org/10.1016/j.crneur.2021.100006 .

Apel K, Hirt H. Reactive oxygen species: metabolism, oxidative stress, and signal transduction. Annu Rev Plant Biol. 2004. https://doi.org/10.1146/annurev.arplant.55.031903.141701 .

Qu D, et al. Role of Cdk5-mediated phosphorylation of Prx2 in MPTP toxicity and Parkinson’s disease. Neuron. 2007. https://doi.org/10.1016/j.neuron.2007.05.033 .

Zhang CW, Hang L, Yao TP, Lim KL. Parkin regulation and neurodegenerative disorders. Front Aging Neurosci. 2016. https://doi.org/10.3389/fnagi.2015.00248 .

Avraham E, Rott R, Liani E, Szargel R, Engelender S. Phosphorylation of Parkin by the cyclin-dependent kinase 5 at the linker region modulates its ubiquitin-ligase activity and aggregation. J Biol Chem. 2007. https://doi.org/10.1074/jbc.M608243200 .

Cleveland DW. From charcot to SOD1: mechanisms of selective motor neuron death in ALS. Neuron. 1999. https://doi.org/10.1016/S0896-6273(00)81108-3 .

Bajaj NPS, et al. Cyclin-dependent kinase-5 is associated with lipofuscin in motor neurones in amyotrophic lateral sclerosis. Neurosci Lett. 1998. https://doi.org/10.1016/S0304-3940(98)00176-1 .

Nguyen MD, Larivière RC, Julien JP. Deregulation of Cdk5 in a mouse model of ALS: toxicity alleviated by perikaryal neurofilament inclusions. Neuron. 2001. https://doi.org/10.1016/S0896-6273(01)00268-9 .

Bk B, et al. Overexpression of the Cdk5 inhibitory peptide in motor neurons rescue of amyotrophic lateral sclerosis phenotype in a mouse model. Hum Mol Genet. 2019. https://doi.org/10.1093/hmg/ddz118 .

Huntington T, et al. A novel gene containing a trinucleotide that is expanded and unstable on Huntington’s disease chromosomes. Cell. 1993;72:971.

Crespo-Biel N, et al. 3-Nitropropionic acid activates calpain/cdk5 pathway in rat striatum. Neurosci Lett. 2007. https://doi.org/10.1016/j.neulet.2007.05.038 .

Paoletti P, et al. Dopaminergic and glutamatergic signaling crosstalk in Huntington’s disease neurodegeneration: the role of p25/cyclin-dependent kinase 5. J Neurosci. 2008. https://doi.org/10.1523/JNEUROSCI.3237-08.2008 .

Brito V, et al. Cyclin-dependent kinase 5 dysfunction contributes to depressive-like behaviors in Huntington’s disease by altering the DARPP-32 phosphorylation status in the nucleus accumbens. Biol Psychiatry. 2019. https://doi.org/10.1016/j.biopsych.2019.03.001 .

Luo S, Vacher C, Davies JE, Rubinsztein DC. Cdk5 phosphorylation of huntingtin reduces its cleavage by caspases: implications for mutant Huntingtin toxicity. J Cell Biol. 2005. https://doi.org/10.1083/jcb.200412071 .

Anne SL, Saudou F, Humbert S. Phosphorylation of huntingtin by cyclin-dependent kinase 5 is induced by DNA damage and regulates wild-type and mutant huntingtin toxicity in neurons. J Neurosci. 2007. https://doi.org/10.1523/JNEUROSCI.1831-07.2007 .

Gendelman HE, et al. The neuropathogenesis of the AIDS dementia complex. AIDS (London, England). 1997;11:1.

Google Scholar  

Patrick C, et al. Increased CDK5 expression in HIV encephalitis contributes to neurodegeneration via tau phosphorylation and is reversed with roscovitine. Am J Pathol. 2011. https://doi.org/10.1016/j.ajpath.2010.12.033 .

Wang Y, et al. Activation of cyclin-dependent kinase 5 by calpains contributes to human immunodeficiency virus-induced neurotoxicity. J Neurochem. 2007. https://doi.org/10.1111/j.1471-4159.2007.04746.x .

Fields J, et al. Mechanisms of HIV-1 tat neurotoxicity via CDK5 translocation and hyper-activation: role in HIV-associated neurocognitive disorders. Curr HIV Res. 2015. https://doi.org/10.2174/1570162x13666150311164201 .

Höppener JWM, Ahrén B, Lips CJM. Islet amyloid and type 2 diabetes mellitus. N Engl J Med. 2000. https://doi.org/10.1056/nejm200008103430607 .

Munshi MN. Cognitive dysfunction in older adults with diabetes: what a clinician needs to know. Diabetes Care. 2017. https://doi.org/10.2337/dc16-1229 .

Bk B, et al. TFP5, a peptide derived from p35, a Cdk5 neuronal activator, rescues cortical neurons from glucose toxicity. J Alzheimer’s Dis. 2014. https://doi.org/10.3233/JAD-131784 .

Wu KK, Huan Y. Streptozotocin-induced diabetic models in mice and rats. Curr Protoc Pharmacol. 2008. https://doi.org/10.1002/0471141755.ph0547s40 .

Sakamaki JI, et al. Role of the SIK2-p35-PJA2 complex in pancreatic β-cell functional compensation. Nat Cell Biol. 2014. https://doi.org/10.1038/ncb2919 .

Wei FY, et al. Cdk5-dependent regulation of glucose-stimulated insulin secretion. Nat Med. 2005. https://doi.org/10.1038/nm1299 .

Zheng YL, et al. Cdk5 inhibitory peptide (CIP) inhibits Cdk5/p25 activity induced by high glucose in pancreatic beta cells and recovers insulin secretion from p25 damage. PLoS ONE. 2013. https://doi.org/10.1371/journal.pone.0063332 .

Tyagi S, Gupta P, Saini A, Kaushal C, Sharma S. The peroxisome proliferator-activated receptor: a family of nuclear receptors role in various diseases. J Adv Pharm Technol Res. 2011. https://doi.org/10.4103/2231-4040.90879 .

Li P, et al. Adipocyte NCoR knockout decreases PPARγ phosphorylation and enhances PPARγ activity and insulin sensitivity. Cell. 2011. https://doi.org/10.1016/j.cell.2011.09.050 .

Choi JH, et al. Anti-diabetic drugs inhibit obesity-linked phosphorylation of PPARγ 3 by Cdk5. Nature. 2010. https://doi.org/10.1038/nature09291 .

Choi JH, et al. Thrap 3 docks on phosphoserine 273 of PPAR γ and controls diabetic gene programming. Genes Dev. 2014. https://doi.org/10.1101/gad.249367.114 .

Meijer L, et al. Biochemical and cellular effects of roscovitine, a potent and selective inhibitor of the cyclin-dependent kinases cdc2, cdk2 and cdk5. Eur J Biochem. 1997. https://doi.org/10.1111/j.1432-1033.1997.t01-2-00527.x .

Eldar-Finkelman H, Eisenstein M. Peptide inhibitors targeting protein kinases. Curr Pharm Des. 2009. https://doi.org/10.2174/138161209788682253 .

Zheng YL, Li BS, Amin ND, Albers W, Pant HC. A peptide derived from cyclin-dependent kinase activator (p35) specifically inhibits Cdk5 activity and phosphorylation of tau protein in transfected cells. Eur J Biochem. 2002. https://doi.org/10.1046/j.1432-1033.2002.03133.x .

Amin ND, Albers W, Pant HC. Cyclin-dependent kinase 5 (cdk5) activation requires interaction with three domains of p35. J Neurosci Res. 2002. https://doi.org/10.1002/jnr.10116 .

Sundaram JR, et al. Specific inhibition of p25/Cdk5 activity by the Cdk5 inhibitory peptide reduces neurodegeneration in vivo. J Neurosci. 2013. https://doi.org/10.1523/JNEUROSCI.3593-12.2013 .

Zheng YL, et al. A 24-residue peptide (p5), derived from p35, the Cdk5 neuronal activator, specifically inhibits Cdk5-p25 hyperactivity and tau hyperphosphorylation. J Biol Chem. 2010. https://doi.org/10.1074/jbc.M110.134643 .

Shukla V, et al. TFP5, a peptide inhibitor of aberrant and hyperactive Cdk5/p25, attenuates pathological phenotypes and restores synaptic function in CK-p25Tg mice. J Alzheimer’s Dis. 2017. https://doi.org/10.3233/JAD-160916 .

Zhang QS, et al. TFP5 prevents 1-methyl-4-phenyl pyridine ion-induced neurotoxicity in mouse cortical neurons. Exp Ther Med. 2016. https://doi.org/10.3892/etm.2016.3658 .

Shukla V, et al. A truncated peptide from p35, a Cdk5 activator, prevents Alzheimer’s disease phenotypes in model mice. FASEB J. 2013. https://doi.org/10.1096/fj.12-217497 .

Seo J, Pao PC, Kritskiy O, Lee A, Patnaik D, Watson LA, Bula M, Barker SJ, Penney J, Silva MC, Haggarty SJ. A cyclin-dependent kinase 5-derived peptide inhibits Cdk5/p25 activity and improves neurodegenerative phenotypes. bioRxiv. 2020. https://doi.org/10.1101/2020.05.12.090472 .

He Y, et al. The presence of active Cdk5 associated with p35 in astrocytes and its important role in process elongation of scratched astrocyte. Glia. 2007. https://doi.org/10.1002/glia.20485 .

Posada-Duque RA, Palacio-Castañeda V, Cardona-Gómez GP. CDK5 knockdown in astrocytes provide neuroprotection as a trophic source via Rac1. Mol Cell Neurosci. 2015. https://doi.org/10.1016/j.mcn.2015.07.001 .

Becerra-Calixto A, Cardona-Gómez GP. Neuroprotection induced by transplanted CDK5 knockdown astrocytes in global cerebral ischemic rats. Mol Neurobiol. 2017. https://doi.org/10.1007/s12035-016-0162-2 .

Becerra-Calixto A, Posada-Duque R, Cardona-Gómez GP. Recovery of neurovascular unit integrity by CDK5-KD astrocyte transplantation in a global cerebral ischemia model. Mol Neurobiol. 2018. https://doi.org/10.1007/s12035-018-0992-1 .

Miyamoto Y, et al. Cdk5 regulates differentiation of oligodendrocyte precursor cells through the direct phosphorylation of paxillin. J Cell Sci. 2007. https://doi.org/10.1242/jcs.018218 .

Yang Y, et al. Cyclin dependent kinase 5 is required for the normal development of oligodendrocytes and myelin formation. Dev Biol. 2013. https://doi.org/10.1016/j.ydbio.2013.03.023 .

Luo F, Zhang J, Burke K, Miller RH, Yang Y. The activators of cyclin-dependent kinase 5 p35 and p39 are essential for oligodendrocyte maturation, process formation, and myelination. J Neurosci. 2016. https://doi.org/10.1523/JNEUROSCI.2250-15.2016 .

Luo FC, Burke K, Kantor C, Miller RH, Yang Y. Cyclin-dependent kinase 5 mediates adult OPC maturation and myelin repair through modulation of Akt and GsK-3β signaling. J Neurosci. 2014. https://doi.org/10.1523/JNEUROSCI.0710-14.2014 .

Pareek TK, et al. Cyclin-dependent kinase 5 activity is required for T cell activation and induction of experimental autoimmune encephalomyelitis. J Exp Med. 2010. https://doi.org/10.1084/jem.20100876 .

Dorand RD, et al. Cdk5 disruption attenuates tumor PD-L1 expression and promotes antitumor immunity. Science. 2016. https://doi.org/10.1126/science.aae0477 .

Feldmann G, et al. Inhibiting the cyclin-dependent kinase CDK5 blocks pancreatic cancer formation and progression through the suppression of Ras-Ral signaling. Cancer Res. 2010. https://doi.org/10.1158/0008-5472.CAN-09-1107 .

Sharma S, et al. Targeting the cyclin-dependent kinase 5 in metastatic melanoma. Proc Natl Acad Sci USA. 2020. https://doi.org/10.1073/pnas.1912617117 .

Mukherjee S, et al. CDK5 inhibition resolves PKA/cAMP-independent activation of CREB1 signaling in glioma stem cells. Cell Rep. 2018. https://doi.org/10.1016/j.celrep.2018.04.016 .

Sang Y, et al. CDK5-dependent phosphorylation and nuclear translocation of TRIM59 promotes macroH2A1 ubiquitination and tumorigenicity. Nat Commun. 2019. https://doi.org/10.1038/s41467-019-12001-2 .

Tripathi BK, et al. CDK5 is a major regulator of the tumor suppressor DLC1. J Cell Biol. 2014. https://doi.org/10.1083/jcb.201405105 .

Download references

Acknowledgements

We thank Hugh P. Cam, Mitchell Murdock, Jay Penney, Matheus Victor, Lorenzo Bozzelli, and members of Tsai lab for discussion and valuable comments on manuscript.

This work was supported by NIH Grant (R37NS051874), Robert A. and Renee E. Belfer Family Foundation, and Belfer Neurodegeneration Consortium to L.-H.T.

Author information

Authors and affiliations.

Picower Institute for Learning and Memory, Massachusetts Institute of Technology, Cambridge, MA, 02139, USA

Ping-Chieh Pao & Li-Huei Tsai

Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, MA, 02139, USA

You can also search for this author in PubMed   Google Scholar

Contributions

P.-C.P. and L.-H.T. conceptualized and wrote the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Li-Huei Tsai .

Ethics declarations

Ethics approval and consent to participate, consent for publication, competing interests.

L.-H.T. has licensed intellectual property related to Cdk5i. The remaining author declares no competing interests.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Pao, PC., Tsai, LH. Three decades of Cdk5. J Biomed Sci 28 , 79 (2021). https://doi.org/10.1186/s12929-021-00774-y

Download citation

Received : 11 August 2021

Accepted : 04 November 2021

Published : 23 November 2021

DOI : https://doi.org/10.1186/s12929-021-00774-y

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Neurodegeneration
• Alzheimer’s disease

Journal of Biomedical Science

ISSN: 1423-0127

• Submission enquiries: Access here and click Contact Us
• General enquiries: [email protected]

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings
• My Bibliography
• Collections
• Citation manager

Save citation to file

Email citation, add to collections.

• Create a new collection
• Add to an existing collection

Add to My Bibliography

Your saved search, create a file for external citation management software, your rss feed.

• Search in PubMed
• Search in NLM Catalog
• Add to Search

Protocols for Characterization of Cdk5 Kinase Activity

Affiliations.

• 1 National Institute of Dental and Craniofacial Research, National Institutes of Health, Bethesda, Maryland.
• 2 National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, Maryland.
• 3 CSIR-Institute of Genomics and Integrative Biology (IGIB), New Delhi, India.
• 4 Department of Biology, Universidad de Chile, Santiago, Chile.
• 5 Celloram Inc., Beachwood, Ohio.
• PMID: 34679246
• PMCID: PMC8555461
• DOI: 10.1002/cpz1.276

Cyclin-dependent kinases (Cdks) are generally known to be involved in controlling the cell cycle, but Cdk5 is a unique member of this protein family for being most active in post-mitotic neurons. Cdk5 is developmentally important in regulating neuronal migration, neurite outgrowth, and axon guidance. Cdk5 is enriched in synaptic membranes and is known to modulate synaptic activity. Postnatally, Cdk5 can also affect neuronal processes such as dopaminergic signaling and pain sensitivity. Dysregulated Cdk5, in contrast, has been linked to neurodegenerative disorders such as Alzheimer's disease (AD), Parkinson's disease (PD), and amyotrophic lateral sclerosis (ALS). Despite primarily being implicated in neuronal development and activity, Cdk5 has lately been linked to non-neuronal functions including cancer cell growth, immune responses, and diabetes. Since Cdk5 activity is tightly regulated, a method for measuring its kinase activity is needed to fully understand the precise role of Cdk5 in developmental and disease processes. This article includes methods for detecting Cdk5 kinase activity in cultured cells or tissues, identifying new substrates, and screening for new kinase inhibitors. Furthermore, since Cdk5 shares homology and substrate specificity with Cdk1 and Cdk2, the Cdk5 kinase assay can be used, with modification, to measure the activity of other Cdks as well. © 2021 Wiley Periodicals LLC. This article has been contributed to by US Government employees and their work is in the public domain in the USA. Basic Protocol 1: Measuring Cdk5 activity from protein lysates Support Protocol 1: Immunoprecipitation of Cdk5 using Dynabeads Alternate Protocol: Non-radioactive protocols to measure Cdk5 kinase activity Support Protocol 2: Western blot analysis for the detection of Cdk5, p35, and p39 Support Protocol 3: Immunodetection analysis for Cdk5, p35, and p39 Support Protocol 4: Genetically engineered mice (+ and - controls) Basic Protocol 2: Identifying new Cdk5 substrates and kinase inhibitors.

Keywords: Cdk5; kinase assay; p25; p35; p39.

© 2021 Wiley Periodicals LLC. This article has been contributed to by US Government employees and their work is in the public domain in the USA.

PubMed Disclaimer

Conflict of interest statement

Conflict of Interest

The authors declare no conflict of interest.

Neuronal and non-neuronal functions of…

Neuronal and non-neuronal functions of Cdk5. Cdk5 and its regulatory subunits p35 and…

Depiction of Cdk5 radioactive assay.…

Depiction of Cdk5 radioactive assay. A) Structure of the Cdk5/p25 complex (van der…

Cdk5 activity was measured in…

Cdk5 activity was measured in wild-type (WT) and transgenic p35 overexpressing (Tgp35) mice.…

Cdk5 assay was performed using…

Cdk5 assay was performed using Dynabeads™ with anti-Cdk5 antibody (~2 μg) from different…

Cdk5 was immunoprecipitated from wild…

Cdk5 was immunoprecipitated from wild type (WT) and transgenic p35 overexpressing (Tgp35) mice…

Specificity and linearity of the…

Specificity and linearity of the ADPGlo Kinase Assay. Graphs A and B depict…

Western blot for Cdk5 and…

Western blot for Cdk5 and its activator p35 along with Beta-actin as a…

Representative IF for p35 (red)…

Representative IF for p35 (red) in transfected COS7 cells by using three diferent…

Representative ICC images for p35…

Representative ICC images for p35 (red) in COS7 cells transfected with a p35…

Representative IF against Cdk5 (left…

Representative IF against Cdk5 (left panel), p35 (middle panel) and p39 (right panel)…

Representative IF against P2X2 (red),…

Representative IF against P2X2 (red), Cdk5 (green), βIII-tubulin (white), and DAPI (cyan) from…

PCR results for Tgp35 and…

PCR results for Tgp35 and WT mice. After the SpeI digest, mice hemizygous…

PCR was used to detect…

PCR was used to detect p35−/− mice. One of the primers binds to…

Cdk5 wildtype and knockout mice…

Cdk5 wildtype and knockout mice can be used to confirm Cdk5- mediated phosphorylation…

PCR to identify the targeted…

PCR to identify the targeted Cdk5 null allele. Ablation of Cdk5 results in…

A peptide kinase assay can…

A peptide kinase assay can be used to pinpoint the most probable Cdk5…

p5, a 24 a.a. peptide…

p5, a 24 a.a. peptide derived from p25 was identified as an inhibitor…

p5 equally inhibits both Cdk5-p25…

p5 equally inhibits both Cdk5-p25 and Cdk5-p35 activities. Increasing concentrations of p5 were…

Densitometric analysis of Cdk5 activity.…

Densitometric analysis of Cdk5 activity. If desired, the Cdk5 activity detected using autoradiography…

Similar articles

• p39 Is Responsible for Increasing Cdk5 Activity during Postnatal Neuron Differentiation and Governs Neuronal Network Formation and Epileptic Responses. Li W, Allen ME, Rui Y, Ku L, Liu G, Bankston AN, Zheng JQ, Feng Y. Li W, et al. J Neurosci. 2016 Nov 2;36(44):11283-11294. doi: 10.1523/JNEUROSCI.1155-16.2016. J Neurosci. 2016. PMID: 27807169 Free PMC article.
• Phosphorylation of cyclin-dependent kinase 5 (Cdk5) at Tyr-15 is inhibited by Cdk5 activators and does not contribute to the activation of Cdk5. Kobayashi H, Saito T, Sato K, Furusawa K, Hosokawa T, Tsutsumi K, Asada A, Kamada S, Ohshima T, Hisanaga S. Kobayashi H, et al. J Biol Chem. 2014 Jul 11;289(28):19627-36. doi: 10.1074/jbc.M113.501148. Epub 2014 May 28. J Biol Chem. 2014. PMID: 24872417 Free PMC article.
• The regulation of cyclin-dependent kinase 5 activity through the metabolism of p35 or p39 Cdk5 activator. Hisanaga S, Saito T. Hisanaga S, et al. Neurosignals. 2003 Sep-Oct;12(4-5):221-9. doi: 10.1159/000074624. Neurosignals. 2003. PMID: 14673209 Review.
• Physiological and Pathological Roles of Cdk5: Potential Directions for Therapeutic Targeting in Neurodegenerative Disease. Allnutt AB, Waters AK, Kesari S, Yenugonda VM. Allnutt AB, et al. ACS Chem Neurosci. 2020 May 6;11(9):1218-1230. doi: 10.1021/acschemneuro.0c00096. Epub 2020 Apr 29. ACS Chem Neurosci. 2020. PMID: 32286796 Review.
• The protein kinase Cdk5. Structural aspects, roles in neurogenesis and involvement in Alzheimer's pathology. Maccioni RB, Otth C, Concha II, Muñoz JP. Maccioni RB, et al. Eur J Biochem. 2001 Mar;268(6):1518-27. doi: 10.1046/j.1432-1033.2001.02024.x. Eur J Biochem. 2001. PMID: 11248668 Review.
• A Retrospective Tribute to Dr. Harish Pant (1938-2023) and His Seminal Work on Cyclin Dependent Kinase 5. Hall B, Amin N, Veeranna, Hisanaga SI, Kulkarni AB. Hall B, et al. Neurochem Res. 2024 Sep 5. doi: 10.1007/s11064-024-04234-5. Online ahead of print. Neurochem Res. 2024. PMID: 39235580 Review.
• TFP5, a Peptide Derived from Cdk5 Activator p35, Protects Pancreatic β Cells from Glucose Toxicity. Liu SY, Cao SL, Luo HY, Bao L, E J, Li B, Lan XM, Zhang GQ, Bao X, Zheng YL. Liu SY, et al. Bull Exp Biol Med. 2023 Nov;176(1):19-25. doi: 10.1007/s10517-023-05959-z. Epub 2023 Dec 13. Bull Exp Biol Med. 2023. PMID: 38087140
• Activation of cyclin-dependent kinase 5 broadens action potentials in human sensory neurons. Tiwari MN, Hall BE, Ton AT, Ghetti R, Terse A, Amin N, Chung MK, Kulkarni AB. Tiwari MN, et al. Mol Pain. 2023 Jan-Dec;19:17448069231218353. doi: 10.1177/17448069231218353. Mol Pain. 2023. PMID: 37982142 Free PMC article.
• Amin ND, Albers W, Pant HC (2002). Cyclin-dependent kinase 5 (cdk5) activation requires interaction with three domains of p35. J Neurosci Res. 67(3), 354–62. doi: 10.1002/jnr.10116. - DOI - PubMed
• Appelmans O, Kashyap RS, Gilles P, De Borggraeve WM, Voet A, & Van Lint J (2021) LSA-50 paper: An alternative to P81 phosphocellulose paper for radiometric protein kinase assays. Anal Biochem. 630,114313. doi: 10.1016/j.ab.2021.114313. - DOI - PubMed
• Asada A, Yamamoto N, Godha M, Saito T, Hayashi N, & Hisanaga S (2008). Myristoylation of p39 and p35 is a determinant of cytoplasmic or nuclear localization of active cyclin-dependent kinase 5 complexes. J Neurochem, 106(3), 1325–36. doi: 10.1111/j.1471-4159.2008.05500.x. - DOI - PubMed
• Bankston AN, Li W, Zhang H, Ku L, Liu G, Papa F, Zhao L, Bibb JA, Cambi F, Tiwari-Woodruff SK, & Feng Y (2013). p39, the primary activator for cyclin-dependent kinase 5 (Cdk5) in oligodendroglia, is essential for oligodendroglia differentiation and myelin repair. J Biol Chem, 288(25), 18047–57. doi: 10.1074/jbc.M113.453688. - DOI - PMC - PubMed
• Bankston AN, Ku L, & Feng Y (2017). Active Cdk5 Immunoprecipitation and Kinase Assay. Bio Protoc. 7(13), e2363. doi: 10.21769/BioProtoc.2363. - DOI - PMC - PubMed
• Search in MeSH

Related information

Grants and funding.

• DE/NIDCR NIH HHS/United States
• Science and Engineering Research Board
• NH/NIH HHS/United States
• Council of Scientific & Industrial Research
• Z01 DE000664/ImNIH/Intramural NIH HHS/United States

LinkOut - more resources

Full text sources.

• Europe PubMed Central
• PubMed Central

Other Literature Sources

• Bio-protocol Exchange

Research Materials

• NCI CPTC Antibody Characterization Program

Miscellaneous

• NCI CPTAC Assay Portal

• Citation Manager

NCBI Literature Resources

MeSH PMC Bookshelf Disclaimer

The PubMed wordmark and PubMed logo are registered trademarks of the U.S. Department of Health and Human Services (HHS). Unauthorized use of these marks is strictly prohibited.

• Advanced search

Advanced Search

Cdk5 Regulates Activity-Dependent Gene Expression and Dendrite Development

• Find this author on Google Scholar
• Find this author on PubMed
• Search for this author on this site
• Figures & Data
• Info & Metrics

This article has a correction. Please see:

• Correction: Liang et al., Cdk5 Regulates Activity-Dependent Gene Expression and Dendrite Development - January 06, 2016

The proper growth and arborization of dendrites in response to sensory experience are essential for neural connectivity and information processing in the brain. Although neuronal activity is important for sculpting dendrite morphology, the underlying molecular mechanisms are not well understood. Here, we report that cyclin-dependent kinase 5 (Cdk5)-mediated transcriptional regulation is a key mechanism that controls activity-dependent dendrite development in cultured rat neurons. During membrane depolarization, Cdk5 accumulates in the nucleus to regulate the expression of a subset of genes, including that of the neurotrophin brain-derived neurotrophic factor, for subsequent dendritic growth. Furthermore, Cdk5 function is mediated through the phosphorylation of methyl-CpG-binding protein 2, a key transcriptional repressor that is mutated in the mental disorder Rett syndrome. These findings collectively suggest that the nuclear import of Cdk5 is crucial for activity-dependent dendrite development by regulating neuronal gene transcription during neural development.

SIGNIFICANCE STATEMENT Neural activity directs dendrite development through the regulation of gene transcription. However, how molecular signals link extracellular stimuli to the transcriptional program in the nucleus remains unclear. Here, we demonstrate that neuronal activity stimulates the translocation of the kinase Cdk5 from the cytoplasmic compartment into the nucleus; furthermore, the nuclear localization of Cdk5 is required for dendrite development in cultured neurons. Genome-wide transcriptome analysis shows that Cdk5 deficiency specifically disrupts activity-dependent gene transcription of bdnf . The action of Cdk5 is mediated through the modulation of the transcriptional repressor methyl-CpG-binding protein 2. Therefore, this study elucidates the role of nuclear Cdk5 in the regulation of activity-dependent gene transcription and dendritic growth.

• nuclear translocation
• transcription
• neurotrophin
• Introduction

Neuronal information is received, integrated, and processed at the dendrites, the functioning of which depends on their proper growth and patterning. Although the early establishment of the specific patterns of dendritic trees in different types of neurons is determined intrinsically by genetic programs, the growth and refinement of dendritic branches are modulated by neuronal activity ( Katz and Shatz, 1996 ; Konur and Ghosh, 2005 ). Blocking neuronal activity by tetrodotoxin in cultured neurons or intact animals leads to dendritic growth deficits. In contrast, membrane depolarization of cultured neurons by KCl or increasing synaptic activity by exposing experimental animals to an enriched environment promotes dendritic growth ( McAllister, 2000 ). Activity-dependent dendritic growth requires gene transcription. Among activity-dependent genes, bdnf , which encodes brain-derived neurotrophic factor (BNDF), is one of the most extensively studied. BDNF plays important roles in dendrite development and synaptic plasticity ( Huang and Reichardt, 2001 ) and its transcription is substantially upregulated by neuronal activity in cultured neurons in vitro and by sensory experience in vivo ( Barth, 2007 ).

Although the importance of activity-dependent gene transcription in dendrite development is well documented, how signals generated by neuronal activity are transduced into the nucleus to regulate transcription has only recently begun to be elucidated. Several transcription factors are implicated in mediating activity-dependent gene transcription crucial for dendritic growth. The cAMP response element-binding protein (CREB) plays a central role in long-term synaptic plasticity and memory in different animals including Drosophila , Aplysia , and mice ( Alberini, 2009 ). CREB is also implicated in mediating the activity-dependent expression of BDNF during cortical development ( Hong et al., 2008 ). In addition, the calcium-dependent transcription factors NeuroD and CREST are crucial for membrane depolarization-induced gene expression and dendritic growth during early development ( Aizawa et al., 2004 ; Gaudillière et al., 2004 ). Recent studies also provide important insights into how these transcription factors are activated by neuronal activity, which involves posttranslational modifications such as phosphorylation. For example, CREB is activated upon phosphorylation by the kinase CaMKIV after neuronal depolarization. CREB is also activated by a slower and more persistent mechanism that involves the nuclear translocation of MAPK and PKA ( Deisseroth et al., 2003 ).

Cortical pyramidal neurons of mice lacking cyclin-dependent kinase 5 (Cdk5) exhibit abnormal dendritic morphology ( Nikolic et al., 1996 ; Ohshima et al., 2007 ). We have reported previously that Cdk5 mediates BDNF-induced dendritic growth in hippocampal neurons through the phosphorylation of BDNF receptor (TrkB) and the regulation of actin dynamics ( Cheung et al., 2007 ). Although Cdk5 plays an essential role in BDNF-induced dendrite development, whether and how Cdk5 participates in activity-dependent brain development remains unclear. In addition, given the nuclear localization of Cdk5 in neurons ( Ino and Chiba, 1996 ) and the various transcriptional regulators (e.g., STAT3, MEF2, and mSds3) as Cdk5 substrates ( Su and Tsai, 2011 ), it is tempting to speculate that Cdk5 regulates dendrite development through transcriptional regulation in conjunction with cytoskeletal reorganization. Accordingly, the present study determines whether and how nuclear Cdk5 promotes neuronal activity-dependent dendrite development in neurons.

Here, we show that Cdk5 translocates into the nucleus after neuronal depolarization. Moreover, loss of Cdk5 significantly reduces dendritic growth and activity-induced bdnf transcription and the dendritic defects can be rescued by wild-type Cdk5, but not the nuclear-localization-deficient mutant. Intriguingly, we found that Cdk5 function is mediated through the regulation of the phosphorylation—and thus the activity of—methyl-CpG-binding protein 2 (MeCP2), a key transcriptional repressor. Therefore, the present study reveals a novel mechanism underlying activity-dependent dendritic growth that involves the nuclear translocation of Cdk5 and its subsequent regulation of the transcriptional program.

• Materials and Methods

Antibodies and constructs.

Antibodies against Cdk5 (C-8, DC-17) and Lamin B (C-7) were from Santa Cruz Biotechnology; antibodies against actin (A3853), FLAG (M2), and MeCP2 (M6818) were from Sigma-Aldrich; MeCP2 antibody (ab2828) was from Abcam; proline-directed phosphoserine antibody was purchased from Cell Signaling Technology; and phospho-S421-MeCP2 antibody was a kind gift from Prof. Michael Greenberg (Harvard Medical School). The target sequence of rat Cdk5 shRNA used in this study was 5′-CCGGGAGATCTGTCTACTC-3′. The GFP-Cdk5 and NES-GFP-Cdk5 constructs were kind gifts from Prof. Karl Herrup (The Hong Kong University of Science and Technology). The nuclear export signal (NES)-Cdk5 was generated by PCR using the MAPKK NES (ALQKKLEELELD).

Protein extraction and fractionation.

The visual cortices of mice of either sex were collected as described previously ( Yoshii et al., 2013 ). Cytosolic and nuclear fractionation was performed using the Nuclear/Cytosol Extraction Kit (BioVision) and the Nuclear Complex Co-IP Kit (Active Motif).

RNA extraction, ChIP, and quantitative real-time PCR.

RNA was extracted using the QIAGEN RNA extraction kit according to the manufacturer's instructions. For ChIP, cells were lysed and fragmented with a Covaris S220 focused ultrasonicator, followed by immunoprecipitation with rabbit polyclonal MeCP2 antibody. Quantitative real-time PCR was performed with fast-standard SYBR green dye using an AB7500 real-time PCR machine as described previously ( Ng et al., 2013 ). The following primers were used for real-time PCR: bdnf exon IV forward 5′-CTGCCTTGATGTTTACTTTGACAAG-3′, bdnf exon IV reverse 5′-ACCATAGTAAGGAAAAGGATGGTGAT-3′; bdnf forward 5′-TTGAGCACGTGATCGAAGAG-3′, bdnf reverse 5′-CCAGCAGAAGAGCAGAGGA-3′; bdnf exon IV (promotor) forward 5′-GCGCGGAATTCTGATTCTGGTAA T-3′; bdnf exon IV (promotor) reverse 5′-GAGAGGGCTCCACGCTGCCTTGAC G-3′; hprt1 mRNA (endogenous control) forward 5′-TGACACTGGTAAAACAATGCA-3′, reverse 5′-GGTCCTTTTCACCAGCAAGCT-3′.

Cell culture and transfection.

Primary hippocampal or cortical neurons were prepared from embryonic day 18 (E18) rats or transgenic mice and maintained in neurobasal medium with 2% B27 supplement. Primary neurons were transfected using the calcium phosphate method as described previously ( Goff et al., 2007 ) or were transfected with the Lipofectamine 2000 transfection kit according to the manufacturer's instructions.

For KCl depolarization, cells were depolarized with 50 m m KCl by adding 31% depolarization buffer (210 m m KCl, 2 m m CaCl 2 , 1 m m MgCl 2 , 10 m m HEPES, pH 7.4) to the culture medium ( Flavell et al., 2006 ). For the control groups, KCl was substituted with the same concentration of NaCl. For the gene expression experiments, cells were pretreated with 1 μ m TTX and 100 μ m APV for 2 h.

Microarray and data analysis.

Cortical neurons at 10 d in vitro (DIV) were cultured from E18 Cdk5-knock-out embryos and littermates. RNA was analyzed with the GeneChip WT PLUS Reagent Kit (WT PLUS Kit; Affymetrix) according to the manufacturer's instructions. The microarray data were analyzed using R language ( http://www.r-project.org/ ). Genes with p values <0.05 and changes of expression exceeding 1.5-fold between the experimental and control conditions (i.e., NaCl vs KCl or cdk5 +/+ vs cdk5 −/− ) were selected to generate the heat map. Gene list analysis was performed using the PANTHER classification system ( http://www.pantherdb.org/ ).

Confocal imaging and quantification.

Images were captured with Nikon A1 confocal microscopes with 40× oil-immersion lenses. The number and length of dendrites were quantified using ImageJ with the NeuronJ plugin ( Meijering et al., 2004 ). Sholl analysis, which measures the number of intersections of neuronal dendrites crossing a series of concentric circles from the cell body, was performed using ImageJ embedded with the Sholl analysis plugin (A. Ghosh, University of California San Diego). Approximately 20–40 neurons from three independent experiments were measured.

Cdk5 nuclear translocation is induced by neuronal activity

To determine whether Cdk5 is involved in activity-dependent dendritic development, we initially investigated whether increased neuronal activity regulates the subcellular localization of Cdk5. To this end, we determined the subcellular localization of Cdk5 in the mouse visual cortex before and after eye opening, when the animals receive visual stimulation. Interestingly, accumulation of Cdk5 in the nuclear fraction of the visual cortex was detected after eye opening ( Fig. 1 A , B ), indicating that Cdk5 translocates into the nucleus after increased neuronal activity. To verify the nuclear localization of endogenous Cdk5 in neurons, hippocampal neurons derived from cdk5 −/− mice and their wild-type littermates were stained with antibody against Cdk5. The results showed that Cdk5 immunoreactivity was colocalized with the dendritic marker MAP2 and nuclear marker DAPI in cdk5 +/+ neurons, whereas the signal was absent in the cdk5 −/− neurons ( Fig. 1 C ). These results indicate that Cdk5 is localized in both the cytoplasm and nucleus of neurons. Cdk5 increasingly accumulated in the nuclei of cultured hippocampal neurons from 3 to 10 DIV, which corresponds to the period critical for dendritic outgrowth and maturation ( Fig. 1 D ). Furthermore, Cdk5 was enriched in the nuclear fraction of the cultured neurons after KCl-induced neuronal depolarization ( Fig. 1 E , F ), whereas the total level of Cdk5 did not change during treatment (data not shown). Notably, time-lapse confocal imaging demonstrated that the nuclear translocation of GFP-Cdk5 in hippocampal neurons was specifically induced by KCl-induced neuronal depolarization but not the control (NaCl) ( Fig. 1 G , H ). These findings collectively indicate that neuronal activity promotes the nuclear translocation of Cdk5.

• Download figure
• Open in new tab
• Download powerpoint

Cdk5 translocates into the nucleus as a result of visual stimulation during neural development. A , B , Nuclear levels of Cdk5 were increased in the mouse visual cortex after eye opening. A , Western blot analysis. P, Postnatal day. B , Quantification: nuclear and total Cdk5 levels were normalized to those of Lamin B and actin, respectively (mean ± SEM; n = 4 from 3 experiments; Student's t test). C , Cdk5 is localized in both the nucleus and cytoplasm. Cultured hippocampal neurons derived from cdk5 +/+ and cdk5 −/− mice were stained with the nuclear marker DAPI (blue), as well as antibodies against Cdk5 (green) and the dendritic marker MAP2 (red). Scale bar, 10 μm. D , Cdk5 nuclear accumulation increased during neuronal maturation (mean ± SEM; n = 15 from two experiments; one-way ANOVA with Tukey's post hoc test, ** p < 0.01; *** p < 0.001). E , F , Cdk5 translocated into the nucleus by neuronal depolarization. Cortical neurons were treated with 50 m m KCl for different times and the nuclear fractions were collected. E , Western blot analysis; Lamin B was used as a loading control. F , Quantification (mean ± SEM; n = 3 from 3 experiments; one-way ANOVA with Tukey's post hoc test, * p < 0.05). G , H , Confocal time-lapse imaging analysis of hippocampal neurons expressing GFP-Cdk5 after KCl treatment. Hippocampal neurons (2 DIV) were transfected with GFP-Cdk5. Two days after transfection, neurons were treated with 50 m m KCl or NaCl. G , Representative images showing Cdk5 translocated into the nucleus after 50 m m KCl treatment. H , Quantification of the average intensity signal (mean ± SEM; n = 4; two-way ANOVA with Bonferroni post hoc test, treatment × time F (6,36) = 24.52, p < 0.001, treatment F (1,6) = 11.69, p < 0.05, time F (6,36) = 29.95, p < 0.001, *** p < 0.001). Scale bar, 10 μm.

Inhibition of Cdk5 attenuates activity-dependent dendrite development

Neuronal activity induces dendritic growth ( Wong and Ghosh, 2002 ). Therefore, we investigated whether Cdk5 activity is required for neuronal depolarization-induced dendritic growth. Although depolarization by KCl increased the number of dendrites in cultured hippocampal neurons, treatment with the Cdk5 inhibitor roscovitine (Ros) significantly attenuated the induced number of dendrites ( Fig. 2 A , B ). To confirm the essential role of Cdk5 in activity-dependent dendrite development, hippocampal neurons derived from cdk5 +/+ and cdk5 −/− mice were treated with KCl and activity-induced dendritic growth was examined. Neuronal depolarization significantly increased dendritic number, whereas the induction was abolished in cdk5 −/− neurons ( Fig. 2 C , D ). Therefore, Cdk5 is critical for the growth of dendritic arbors induced by neuronal depolarization.

Cdk5 inhibition attenuates activity-dependent dendrite development. A , B , The increase in dendrite arborization after 24 h of KCl depolarization was attenuated by Ros. Cultured hippocampal neurons expressing GFP at 3 DIV were cotreated with 25 μ m Ros. A , Representative images showing neuron morphology. Scale bar, 20 μm. B , Quantification of dendrite number (mean ± SEM; n = 10; two-way ANOVA with Bonferroni post hoc test, treatment × drug F (1,18) = 2.41, p = 0.1382, treatment F (1,18) = 1.93, p = 0.1821, drug F (1,18) = 11.48, p < 0.01, ** p < 0.01). C , D , The KCl-induced dendritic growth was abolished in cdk5 −/− neurons. C , Cultured hippocampal neurons from cdk5 −/− mice and their littermates were treated with KCl and stained with dendritic marker MAP2 (green). Scale bar, 10 μm. D , Quantification of the dendrite (mean ± SEM; n = 6; two-way ANOVA with Bonferroni post hoc test, treatment × genotype F (1,10) = 14.09, p < 0.001, treatment F (1,10) = 12.02, p < 0.01, genotype F (1,10) = 5.25, p < 0.05, *** p < 0.001). E – H , Cdk5 nuclear expression is important for dendrite development. E , Subcellular expressions of WT-Cdk5 and NES-Cdk5 in cultured hippocampal neurons. Cultured hippocampal neurons transfected with Cdk5 or its nuclear localization-deficient mutant (NES-Cdk5) were stained with Cdk5 antibody. Scale bar, 10 μm. F – H , Hippocampal neurons (7 DIV) were transfected with Cdk5 shRNA together with RNAi-resistant Cdk5 constructs. F , Sholl analysis of transfected neurons with Cdk5 shRNA and Cdk5 constructs. G , H , Knock-down of Cdk5 expression significantly reduced dendrite number and length. Coexpression of wild-type Cdk5 but not NES-Cdk5 rescued the defective dendritic arborization (mean ± SEM; n = 30 from three experiments; one-way ANOVA with Tukey's post hoc test, ** p < 0.01, *** p < 0.001).

Nuclear localization of Cdk5 is required for dendrite development in hippocampal neurons

Given that Cdk5 is required for activity-induced dendritic growth and accumulates in the nucleus as a result of increased neuronal activity, we hypothesized that the nuclear localization of Cdk5 plays an important regulatory role during dendrite development. To test this hypothesis, a nuclear-localization-deficient Cdk5 mutant was constructed by adding an NES ( Zhang et al., 2010 ), which reduced the nuclear expression of Cdk5 ( Fig. 2 E ). Cdk5 knock-down in hippocampal neurons by RNAi significantly reduced dendrite number and length. Importantly, coexpression of wild-type Cdk5 partially rescued the defective dendritic arborization caused by the Cdk5 RNAi, but the nuclear localization-deficient mutant (NES-Cdk5) failed to rescue the dendritic defects ( Fig. 2 F–H ). These results suggest that the nuclear localization of Cdk5 is required for its function in dendrite development.

Genome-wide analysis of Cdk5-regulated activity-dependent genes

To investigate how Cdk5 promotes activity-dependent dendrite development, an unbiased microarray analysis was performed to identify activity-dependent genes differentially regulated in cortical neurons derived from wild-type versus Cdk5-knock-out embryos. First, we identified that, upon KCl treatment, 2973 genes (including c-Fos , Arc , bdnf , and Npas4 ) were upregulated and 3230 genes were downregulated (KCl vs NaCl, data not shown); many of these genes have been identified in previous studies, suggesting robust methodology ( Lin et al., 2008 ). Among the activity-regulated genes, 332 were significantly altered in cdk5 −/− neurons compared with cdk5 +/+ neurons after KCl treatment ( cdk5 −/− vs cdk5 +/+ under KCl condition; Fig. 3 A , B ; Gene Expression Omnibus accession number GSE68320 ). Among these, 135 and 139 genes were downregulated and upregulated in cdk5 −/− neurons, respectively ( Fig. 3 A , B ). These results indicate that Cdk5 can regulate transcription during membrane depolarization in a bidirectional manner, as a transcriptional activator, or as a transcriptional repressor.

Cdk5 is essential for activity-dependent transcriptional regulation. A , Heat map showing the clustering of 388 activity-dependent probe sets (representing 332 unique genes) that are differentially expressed between cdk5 +/+ and cdk5 −/− cortical neurons upon KCl treatment ( n = 3 mice per group). Cortical neurons (10 DIV) derived from cdk5 +/+ and cdk5 −/− mice were treated with 50 m m KCl or NaCl for 6 h to examine the mRNA expression. The expression level of each probe set is normalized to a mean of 0 and SD of 1 (log 2 ). Expression values are displayed within the range [−2, 2], with levels above and below the mean displayed in yellow and blue, respectively. B , List of the top 20 genes showing the differentially expressed genes (upregulated and downregulated genes are shown in red and blue, respectively) in cdk5 −/− neurons relative to cdk5 +/+ neurons after KCl depolarization. C , Biological functions of putative Cdk5-regulated genes on the basis of gene ontology information provided by the PANTHER classification system. D , Downregulated and upregulated genes that may function in dendrite development on the basis of the DAVID database ( https://david.ncifcrf.gov/ ). E , Comparison of the expression levels of each probe set between cdk5 +/+ and cdk5 −/− neurons ( x -axis) after KCl depolarization ( y -axis). Two probe sets for cdk5 and bdnf are denoted in red, respectively.

Gene function analysis ( Mi et al., 2013 ) demonstrated that Cdk5 regulates a wide variety of genes such as those encoding cell surface receptors and their ligands and transporters, enzymes such as kinases and phosphatases, transcriptional factors and cofactors, and genes involved in cytoskeleton-modulating pathways ( Fig. 3 C ). Although some Cdk5-regulated genes (e.g., bdnf , Cdh1 , and Cacnb2 ) are implicated in neuronal development ( Van Aelst and Cline, 2004 ; Tan et al., 2010 ; Breitenkamp et al., 2014 ), the functions of most of the genes identified in this analysis in dendrite development are uncharacterized. Interestingly, among the activity-induced genes, bdnf was one of the most significantly downregulated in cdk5 −/− neurons ( Fig. 3 D , E ).

Cdk5 regulates activity-dependent gene expression through the modulation of MeCP2 transcriptional activity

The activity-dependent gene bdnf promotes dendritic growth both in vitro and in vivo ( Van Aelst and Cline, 2004 ). Notably, the bdnf transcript that contains exon IV promoter (induced by activity) was consistently decreased in the absence of Cdk5 in the KCl condition in microarray analysis (>2-fold; Fig. 4 A ). Therefore, we examined the activity-induced changes of bdnf transcripts in cdk5 +/+ and cdk5 −/− cortical neurons triggered by membrane depolarization using quantitative real-time PCR. Although treatment with KCl markedly upregulated the expressions of all bdnf transcripts and the exon IV transcript, the increase of their respective mRNA levels was attenuated in cdk5 −/− neurons compared with cdk5 +/+ neurons ( Fig. 4 B , C ). These results suggest that Cdk5 specifically regulates activity-dependent bdnf expression.

Cdk5 regulates MeCP2 phosphorylation and DNA-binding activity. A , Schematic diagram of mouse bdnf ; exons and introns are represented by boxes and lines, respectively. The probes used are denoted as a–k . The log 2 values of the normalized intensity of individual probe set are plotted (mean ± SEM from three independent experiments; *** p < 0.005, cdk5 −/− vs cdk5 +/+ in the KCl condition, one-way ANOVA with Tukey's post hoc test). B , C , Fold changes of all bdnf transcripts and activity-induced exon IV transcript in cdk5 +/+ and cdk5 −/− cortical neurons triggered by membrane depolarization determined by quantitative real-time PCR (mean ± SEM; n = 3; one-way ANOVA with Tukey's post hoc test, *** p < 0.001). Cortical neurons (10 DIV) derived from cdk5 +/+ and cdk5 −/− mice were treated with 50 m m KCl or NaCl for 6 h to examine bdnf mRNA expression. D , MeCP2 was phosphorylated by Cdk5/p35 at Ser-Pro site(s) in HEK293T cells. FLAG-tagged MeCP2 was immunoprecipitated by anti-FLAG antibody, followed by Western blotting with antibody against proline-directed phosphoserine antibody. E , F , Regulation of KCl-induced Ser421 MeCP2 phosphorylation by Cdk5. E , Cultured cortical neurons were pretreated with 25 μ m DMSO control or Ros for 1 h, followed by 50 m m KCl or NaCl for 30 min. The experiment was repeated at least three times. F , KCl-induced Ser421 MeCP2 phosphorylation was attenuated in cdk5 −/− neurons. G , ChIP assay of MeCP2 followed by real-time PCR using bdnf exon IV promoter-specific primers (data were normalized to those of the Control+NaCl group; mean ± SEM; n = 5; one-way ANOVA with Tukey's post hoc test, * p < 0.05, ** p < 0.01). Cultured cortical neurons were pretreated with 25 μ m DMSO control or Ros for 1 h, followed by 50 m m KCl or NaCl for 90 min.

Activity-dependent bdnf transcription was recently reported to be controlled by MeCP2, mutations of which cause the neurological disorder Rett syndrome ( Chen et al., 2001 ; Guy et al., 2001 ; Shahbazian et al., 2002 ). Because our previous mass spectrometry screening identified MeCP2 as a potential substrate of Cdk5 and because the activity of nuclear Cdk5 was elevated by membrane depolarization (data not shown), we investigated whether Cdk5 can phosphorylate MeCP2 and regulate its transcriptional repressor activity. Using antibodies that specifically recognize proline-directed phosphoserine residues, we found that Cdk5/p35 coexpression specifically induced MeCP2 serine phosphorylation ( Fig. 4 D ), but not threonine phosphorylation (data not shown). MeCP2 is reported to be phosphorylated at Ser421 by CaMKII in response to neuronal activity and its specific phosphorylation is critical for activity-dependent bdnf transcription and dendrite development ( Zhou et al., 2006 ). Intriguingly, the KCl-induced Ser421 phosphorylation of MeCP2 was attenuated by Ros pretreatment and reduced in cdk5 −/− neurons ( Fig. 4 E , F ). Therefore, the results indicate that Cdk5 is required for the activity-dependent phosphorylation of MeCP2 at Ser421. We subsequently investigated whether Cdk5 modulates the transcriptional repressor activity of MeCP2. ChIP assay revealed that membrane depolarization reduced the association of MeCP2 with bdnf exon IV by ∼50% in control neurons upon KCl stimulation ( Fig. 4 G ). In contrast, this release of MeCP2 from bdnf exon IV was significantly attenuated after Ros treatment. These results suggest that Cdk5 activity is required for the effective release of MeCP2 from the bdnf promoter and the subsequent transcription of bdnf upon neuronal activity.

Proper dendrite growth and branching are crucial for neural circuit formation and nervous system function. Although defects in dendrite development are associated with various human mental disorders such as autism and schizophrenia ( Kaufmann and Moser, 2000 ), relatively little is known about the molecular mechanisms that control dendrite development. The present study identified Cdk5-MeCP2-dependent transcriptional regulation as an important signaling axis in activity-dependent dendrite development. Although Cdk5 regulates transcription and dendritic growth through the phosphorylation, thus modulating the activity of MeCP2, a key transcriptional repressor for activity-regulated genes, this regulatory mechanism is precisely controlled by the activity-dependent nuclear translocation of Cdk5. The present findings advance the current understanding of how neuronal activity shapes dendritic arborization patterns and connectivity during nervous system development.

The formation of dendritic arbors is tightly regulated by neuronal activity during neural development and one of the major underlying mechanisms involves the expressions of activity-dependent genes such as that of the neurotrophin BDNF ( Katz and Shatz, 1996 ; Konur and Ghosh, 2005 ). Although gene transcription can be specifically regulated by the phosphorylation of transcriptional regulators, how protein kinases propagate signals generated by neuronal activity to the transcriptional machinery remains poorly understood. Despite being studied extensively as a key kinase that regulates cytoskeleton in brain development and synaptic plasticity ( Su and Tsai, 2011 ), the roles of Cdk5 in regulating transcriptional machinery have only begun to be elucidated. The present findings show that Cdk5 transduces the extracellular signal to the nucleus to regulate activity-induced gene transcription for the growth of dendritic arbors, which may serve as an important mechanism that molds the neural circuit in response to sensory experience during brain development.

Various transcriptional regulators, including MEF2 and mSds3, are substrates of Cdk5. Cdk5 phosphorylates and consequently inhibits MEF2 transcriptional activity, suggesting that Cdk5 might regulate the development of dendritic spines and excitatory synapses via the regulation of MEF2-mediated gene expression ( Flavell et al., 2006 ; Pulipparacharuvil et al., 2008 ). Moreover, we reported previously that Cdk5 phosphorylates mSds3, a corepressor of mSin3-histone deacetylase (HDAC), so this phosphorylation is crucial for its repressive activity ( Li et al., 2004 ). Notably, HDAC is implicated in the activity-regulated expression of bdnf ( Zhang et al., 2007 ). Furthermore, other transcription factors that mediate activity-dependent gene transcription, such as NeuroD, SRF, ELK, NFAT, CREST, and NF-κB, also contain putative Cdk5 phosphorylation sites (Z. Liang and N. Y. Ip, unpublished observations); however, whether they are regulated by Cdk5 has not been investigated. In addition to the phosphorylation of transcription factors, Cdk5 may regulate the transcriptional machinery through the modulation of different signaling pathways such as the cAMP/CREB pathway ( Guan et al., 2011 ). Our finding of the nuclear translocation of Cdk5 upon activity stimulation provides a molecular basis for how this kinase modulates transcription.

The results of our microarray analysis are the first to demonstrate that Cdk5 is required for the activation and repression of distinct subsets of activity-regulated genes during membrane depolarization ( Fig. 3 A ). Among the genes upregulated by neuronal activity, the expression of bdnf exon IV promoter-specific transcript was markedly attenuated in Cdk5-deficient neurons ( Fig. 4 A–C ). Indeed, knock-in mice with a mutation in bdnf exon IV exhibit defects in dendrite morphology ( Hong et al., 2008 ). These lines of evidence collectively highlight the significance of Cdk5 in activity-induced bdnf transcription, which is critical for dendrite development.

How does Cdk5 regulate activity-dependent bdnf gene transcription? BDNF is encoded by a complex gene with multiple distinct promoters that give rise to at least nine transcript variants ( Aid et al., 2007 ). In the absence of neuronal activity, MeCP2, a key contributor to Rett syndrome, inhibits bdnf gene transcription by selectively binding to its exon IV promoter. Membrane depolarization triggers the calcium-dependent Ser421 phosphorylation of MeCP2, thereby releasing MeCP2 from bdnf exon IV and activating bdnf gene transcription ( Chen et al., 2003 ; Zhou et al., 2006 ). In the present study, the results of a phosphorylation assay identified MeCP2 as a novel substrate of Cdk5. Inhibition of Cdk5 attenuated activity-induced Ser421 MeCP2 phosphorylation in cortical neurons and impaired its effective release from bdnf . Notably, MeCP2 Ser421 phosphorylation may facilitate a genome-wide response of chromatin to neuronal activity ( Cohen et al., 2011 ). Given that the structure of chromatin modulates the access of condensed genomic DNA to the regulatory transcription machinery proteins, thereby controlling gene expression, it would be of interest to further characterize the physiological functions of the Cdk5-dependent regulation of MeCP2 in experience-dependent dendrite development and remodeling; the results of such a study may shed light on the molecular basis of normal brain development and the etiology of neurological disorders associated with dendrite abnormalities, such as autism and schizophrenia.

We are grateful to Prof. Michael Greenberg (Harvard Medical School) for sharing MeCP2 antibodies and Prof. Karl Herrup (The Hong Kong University of Science and Technology) for the GFP-Cdk5 and NES-GFP-Cdk5 constructs. We also thank Cara Kwong, Busma Butt, William Chau, Renfei Wu, Yilei Cai, and Dr. Maggie Chu for their excellent technical assistance as well as Prof. Tom Cheung and other members of the Ip laboratory for helpful discussions. This study was supported in part by the Research Grants Council of Hong Kong SAR (HKUST660810, HKUST661111, HKUST661212, and HKUST660213), the National Key Basic Research Program of China (2013CB530900), Hong Kong Research Grants Council Theme-based Research Scheme (T13-607/12R), and the SH Ho Foundation.

The authors declare no competing financial interests.

• Correspondence should be addressed to Prof. Nancy Y. Ip, Division of Life Science, Molecular Neuroscience Center and State Key Laboratory of Molecular Neuroscience, The Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong, China. boip{at}ust.hk
• Kazantseva A ,
• Piirsoo M ,
• Balakrishnan K ,
• Gurevich I ,
• Alberini CM
• Breitenkamp AF ,
• Matthes J ,
• Lehmkuhl G ,
• Nürnberg P ,
• Akbarian S ,
• Meissner A ,
• Griffith EC ,
• Jaenisch R ,
• Greenberg ME
• Cheung ZH ,
• Hemberg M ,
• Hutchinson AN ,
• Sadacca LA ,
• Harmin DA ,
• Greenberg RS ,
• Verdine VK ,
• Wetsel WC ,
• Deisseroth K ,
• Mermelstein PG ,
• Flavell SW ,
• Paradis S ,
• Gaudillière B ,
• Konishi Y ,
• de la Iglesia N ,
• Johnson RN ,
• McDonald HR ,
• Jumper JM ,
• Clauser KR ,
• Hendrich B ,
• Martin JE ,
• McCord AE ,
• Reichardt LF
• Kaufmann WE ,
• DePinho RA ,
• Bloodgood BL ,
• Hauser JL ,
• McAllister AK
• Meijering E ,
• Sarria JC ,
• Steiner P ,
• Hirling H ,
• Muruganujan A ,
• Casagrande JT ,
• Nikolic M ,
• Ohshima T ,
• Hirasawa M ,
• Iwasato T ,
• Itohara S ,
• Hashimoto M ,
• Nakajima K ,
• Kulkarni AB ,
• Mikoshiba K
• Pulipparacharuvil S ,
• Renthal W ,
• Taniguchi M ,
• Greengard P ,
• Nestler EJ ,
• Shahbazian M ,
• Yuva-Paylor L ,
• Spencer C ,
• Antalffy B ,
• Noebels J ,
• Armstrong D ,
• Van Aelst L ,
• Pandian S ,
• van Zundert B ,
• Constantine-Paton M
• Kohlhaas M ,
• Phillips W ,
• Dybkova N ,
• Schmidt L ,

In this issue

• Table of Contents
• Table of Contents (PDF)
• About the Cover
• Index by author
• Advertising (PDF)
• Ed Board (PDF)

Thank you for sharing this Journal of Neuroscience article.

NOTE: We request your email address only to inform the recipient that it was you who recommended this article, and that it is not junk mail. We do not retain these email addresses.

Citation Manager Formats

• EndNote (tagged)
• EndNote 8 (xml)
• RefWorks Tagged
• Ref Manager

• Tweet Widget
• Facebook Like
• Google Plus One

Jump to section

Responses to this article, jump to comment:, related articles, cited by..., more in this toc section.

• Memory Retrieval Has a Dynamic Influence on the Maintenance Mechanisms That Are Sensitive to ζ-Inhibitory Peptide (ZIP)
• Neurophysiological Evidence for a Cortical Contribution to the Wakefulness-Related Drive to Breathe Explaining Hypocapnia-Resistant Ventilation in Humans
• Monomeric Alpha-Synuclein Exerts a Physiological Role on Brain ATP Synthase

Development/Plasticity/Repair

• Blockade of GluN2B-Containing NMDA Receptors Prevents Potentiation and Depression of Responses during Ocular Dominance Plasticity
• Cortical Tagged Synaptic Long-Term Depression in the Anterior Cingulate Cortex of Adult Mice
• Neuritin controls axonal branching in serotonin neurons: A possible mediator involved in the regulation of depressive and anxiety behaviors via FGF signaling.

• Research Article
• Biochemistry and Chemical Biology
• Neuroscience

Cyclin-dependent kinase 5 (CDK5) regulates the circadian clock

• Andrea Brenna
• Iwona Olejniczak
• Rohit Chavan
• Jürgen A Ripperger
• Sonja Langmesser
• Elisabetta Cameroni
• Claudio De Virgilio
• Jörn Dengjel

• University of Fribourg, Switzerland ;
• Open access
• Copyright information

Share this article

Cite this article.

• Urs Albrecht
• Copy to clipboard
• Download BibTeX
• Download .RIS

eLife digest

Introduction, materials and methods, data availability, article and author information.

Circadian oscillations emerge from transcriptional and post-translational feedback loops. An important step in generating rhythmicity is the translocation of clock components into the nucleus, which is regulated in many cases by kinases. In mammals, the kinase promoting the nuclear import of the key clock component Period 2 (PER2) is unknown. Here, we show that the cyclin-dependent kinase 5 (CDK5) regulates the mammalian circadian clock involving phosphorylation of PER2. Knock-down of Cdk5 in the suprachiasmatic nuclei (SCN), the main coordinator site of the mammalian circadian system, shortened the free-running period in mice. CDK5 phosphorylated PER2 at serine residue 394 (S394) in a diurnal fashion. This phosphorylation facilitated interaction with Cryptochrome 1 (CRY1) and nuclear entry of the PER2-CRY1 complex. Taken together, we found that CDK5 drives nuclear entry of PER2, which is critical for establishing an adequate circadian period of the molecular circadian cycle. Of note is that CDK5 may not exclusively phosphorylate PER2, but in addition may regulate other proteins that are involved in the clock mechanism. Taken together, it appears that CDK5 is critically involved in the regulation of the circadian clock and may represent a link to various diseases affected by a derailed circadian clock.

Anyone who has crossed multiple time zones on a long flight will be familiar with jet lag, and that feeling of wanting to sleep at lunchtime and eat in the middle of the night. Many bodily processes, including appetite and wakefulness, roughly follow a 24-hour cycle. These cycles are known as circadian rhythms, from the Latin ‘circa diem’ meaning about a day. An area of the brain called the suprachiasmatic nucleus (SCN) coordinates circadian rhythms. It acts as a master clock by generating a 24-hour signal for the rest of the body to follow. Jet lag occurs when this internal circadian rhythm becomes out of sync with the local day-night cycle.

Although jet lag can be uncomfortable, it tends to disappear over the course of a few days. This is because exposure to daylight in our new location resets the SCN master clock, enabling us to adapt to a new time zone. But evidence suggests that long-term disruption of circadian rhythms, for example as a result of shift work, may have lasting harmful effects. These include an increased risk of degenerative brain disorders such as Parkinson's disease and Alzheimer's disease.

Brenna et al. now identify a molecular mechanism that could explain this link. A key component of the SCN master clock is a protein called Period2 (PER2). Levels of PER2 rise and fall over each 24-hour period, helping the brain keep track of time. Brenna et al. show that PER2 interacts with CDK5, a protein that helps regulate brain development and that has been implicated in Parkinson's disease and Alzheimer's disease. Reducing CDK5 levels in mice shortened their circadian rhythms by several hours. It also altered the animals’ behavioral patterns over a 24-hour period. Deleting the gene for PER2 had a similar effect, suggesting that CDK5 helps regulate PER2.

Future studies should investigate the molecular links between CDK5, circadian rhythms and processes such as neurodegeneration. The results would provide clues to whether manipulating the circadian clock could help prevent or treat neurological disorders.

The circadian clock, prevalent in most organisms, is an evolutionary adaptation to the daily light-dark cycle generated by the sun and the earth’s rotation around its own axis ( Rosbash, 2009 ). This clock allows organisms to organize physiology and behavior over the 24 hr time scale in order to adapt and thus optimize, body function to predictably recurring daily events. Malfunctioning or disruption of the circadian clock in humans results in various pathologies including obesity, cancer, and neurological disorders ( Roenneberg and Merrow, 2016 ). In order to maintain phase synchronicity with the environmental light-dark cycle, the suprachiasmatic nuclei (SCN), a bipartite brain structure located in the ventral part of the hypothalamus above the optic chiasm, receive light information from the retina. The SCN convert this information into humoral and neuronal signals to set the phase of all circadian oscillators in the body ( Dibner et al., 2010 ).

In order to measure the length of one day, organisms have developed cell-based molecular mechanisms relying on feedback loops involving a set of clock genes. The existence of such loops was suggested by the analysis of Drosophila having various mutations in their period ( per ) gene ( Hardin et al., 1990 ). Further studies completed the picture of intertwined transcriptional feedback loops at the heart of the Drosophila circadian oscillator ( Darlington et al., 1998 ). Every day, per accumulates to a certain concentration upon which it enters into the nucleus together with timeless (tim). This protein complex inhibits transcriptional activation mediated by dClock and cycle acting on the expression of per and tim . After the degradation of the inhibitor complex, the repression is relieved and a new circadian cycle starts.

To fine-tune the period of the circadian oscillator, kinases regulate the accumulation and nuclear entry of per and tim. The kinase double-time (dbt) phosphorylates per to destabilize it and to prevent its transport into the nucleus ( Kloss et al., 1998 ; Price et al., 1998 ). On the other hand, the kinase shaggy (shg) phosphorylates tim to stabilize the heterodimer and to promote its nuclear translocation ( Martinek et al., 2001 ). Many other kinases and phosphatases are necessary to complete the Drosophila circadian cycle and to adjust its phase to the external light-dark rhythm ( Garbe et al., 2013 ).

The circadian oscillator of mammals is arranged very similarly to the one of Drosophila , with some modifications ( Dibner et al., 2010 ; Takahashi, 2017 ). For instance, the function of Drosophila tim to escort per into the nucleus was replaced by the Cryptochromes (Cry) in the mammalian system ( van der Horst et al., 1999 ). Furthermore, the first mutation to affect the mammalian circadian oscillator, Tau , was later mapped to Casein kinase Iε (CK1ε), which is the Drosophila dbt orthologue ( Lowrey et al., 2000 ). One of the sites phosphorylated by CK1ε within human PER2 is mutated in the Familial Advanced Sleep Phase Syndrome (FASPS) ( Toh et al., 2001 ). This mutation and also the Tau mutation were subsequently introduced into the mouse genome to prove their functional relevance ( Meng et al., 2008 ; Xu et al., 2007 ). However, a kinase similar to the function of shg in Drosophila , which stabilizes and promotes the import of PER proteins into the nucleus of mammals ( Hirano et al., 2017 ), has not been identified. Interestingly, PER2 contains over 20 potential phosphorylation sites ( Vanselow et al., 2006 ), indicating that mammalian PER and specifically PER2 are highly regulated at the post-translational level. This degree of phosphorylation is probably contributing to the precise rhythmicity of PER2, which stands out as a crucial feature of the core clock ( Chong et al., 2012 ).

Among the plethora of kinases identified that phosphorylate mammalian clock proteins, cyclin-dependent kinase 5 (CDK5) was found to target CLOCK ( Kwak et al., 2013 ). CDK5 is a proline-directed serine-threonine kinase belonging to the Cdc2/Cdk1 family that is controlled by the neural specific activators p35, p39 ( Tang et al., 1995 ; Tsai et al., 1994 ), and cyclin I ( Brinkkoetter et al., 2009 ). CDK5 regulates various neuronal processes such as neurogenesis, neuronal migration, and axon guidance ( Kawauchi, 2014 ). Outside of the nervous system CDK5 regulates vesicular transport, apoptosis, cell adhesion, and migration in many cell types ( Contreras-Vallejos et al., 2012 ). It has been proposed that CDK5 modulates the brain reward system ( Benavides et al., 2007 ; Bibb et al., 2001 ) and that it is consequently linked to psychiatric diseases ( Engmann et al., 2011 ; Zhu et al., 2012 ). Interestingly, the clock components PER2 and CLOCK have been associated with the same processes ( Abarca et al., 2002 ; Hampp et al., 2008 ; Roybal et al., 2007 ). However, it is unknown whether CDK5 plays an important role in the central oscillator of the circadian clock.

In this study, we wanted to identify proteins promoting the nuclear transport of PER2 with focus on kinase(s) acting similarly to shg. Using a genetic synthetic lethal dosage screen in yeast, we observed a genetic interaction between Per2 and PHO85 , which encodes a cyclin-dependent protein kinase that is orthologous to CDK5 in mammals. Subsequent experiments in mice demonstrated that silencing of Cdk5 in the SCN shortened the clock period. Our study identified CDK5 as a critical protein kinase in the regulation of the circadian clock and in particular as an important regulator of the crucial clock component PER2.

Genetic interaction between Per2 and CDK5 in yeast and diurnal activity of CDK5

In order to gain insight into the regulation of PER2 function in mice, we initially tried to identify genes that genetically interact with Per2 in budding yeast by using a variation of the Synthetic Genetic Array (SGA) method ( Tong, 2001 ). To this end, we carried out a synthetic dosage lethality (SDL) screen, which is based on the concept that a high dosage of a given protein (i.e. PER2 in this case) may have negligible effect on growth in wild-type cells (as we found to be the case for PER2; Figure 1A ), but may compromise growth in mutants that have defects in pathway components or in functionally related processes ( Measday et al., 2005 ; Sopko et al., 2006 ). Of note, SDL screens have been instrumental in the past to specifically predict the relationship between protein kinases and their targets ( Sharifpoor et al., 2012 ). Our search in a yeast knockout collection (encompassing 4857 individual deletion strains) for mutants that exhibited significantly reduced growth when combined with increased dosage of PER2 (see Materials and methods for further details) allowed us to isolate three mutants, namely eap1∆ , gnd1∆ , and pho85∆ ( Figure 1A ). Among these, the strain lacking the cyclin-dependent protein kinase Pho85 was most dramatically compromised for growth in the presence of high doses of PER2. Hence, Pho85 antagonizes the growth-inhibitory effect of PER2 in yeast, which indicates that the Pho85-orthologous CDK5 may potentially act upstream of PER2 in mammalian cells.

CDK5 intersects with PER2 and has diurnal activity in the SCN.

( A ) Loss of Eap1, Gnd1, or Pho85 compromises growth of PER2-overproducing yeast cells. The yeast mutants eap1∆ , gnd1∆ , and pho85∆ were identified in a synthetic dosage lethal screen as detailed under Methods. Wild-type (BY4741) as well as eap1∆ , gnd1∆ , and pho85∆ mutant cells carrying the control plasmid (YCpIF2) or the YPpIF2- mPer2 plasmid (that drives expression of mouse PER2 from a galactose-inducible promoter) were pre-grown on glucose-containing SD-Leu media (to an OD 600 of 2.0), spotted (in 10-fold serial dilutions) on raffinose and galactose-containing SD-Raf/Gal-Leu plates, and grown for 3 days at 30°C. ( B ) Immunoblot was performed on SCN extracts around the clock. SCN from seven animals were pooled at each indicated ZT between ZT0-20. Protein levels of CDK5, CRY1, and HSP90 were analyzed by western blot. ( C ) Diurnal activity of CDK5 was measured by an in vitro kinase assay. CDK5 was immunoprecipitated at each same time point between ZT0 and ZT20, and half of the immunoprecipitated material was used for performing an in vitro kinase assay using histone H1 (autoradiography, middle panel), whereas the other half was used to quantify the immunoprecipitated CDK5 (upper panel). Coomassie staining shows loading of the substrate (H1). Bottom panel: Quantification of three independent experiments (mean ± SEM). One-way ANOVA with Bonferroni’s post-test, *: p<0.001. ( D ) The in vitro kinase assay was performed with SCN extracts at ZT12, and either LiCl (GSK3β inhibitor) or 34 μM roscovitine (CDK5 inhibitor). Histone H1 phosphorylation could not be detected with roscovitine treatment, showing the specificity of H1 phosphorylation by CDK5.

The protein kinase CDK5 is mostly expressed in the brain and has previously been implicated in phosphorylation of mammalian CLOCK ( Kwak et al., 2013 ). However, the functional relevance of CDK5 for the clock mechanism has never been tested. Therefore, we investigated whether CDK5 affected the functioning of the circadian clock. First, we assessed whether CDK5 displayed time of day-dependent expression and activity in the SCN, the master clock of the circadian system. We collected SCN samples every 4 hr starting from ZT0 until ZT20 (ZT0 = light on, ZT12 = light off), and performed western blots on total extracts using specific antibodies ( Figure 1B ). The immunoblot against CRY1 showed a diurnal profile of this protein with a peak during the late-night phase, confirming that the mice were entrained properly to the light-dark cycle. In contrast, the CDK5 accumulation profile seemed to be unaffected by the time of day ( Figure 1B ). Next, we investigated whether CDK5 kinase activity displayed a diurnal profile. While CDK5 levels did not change significantly over one day ( Figure 1B ), we observed that histone-H1, a known CDK5 target ( Peterson et al., 2010 ), was phosphorylated by this kinase in a time of day-dependent manner, with the highest levels of CDK5 activity observed at ZT12 to ZT20, that is during the dark phase ( Figure 1C ). Phosphorylation of histone-H1 was specifically blocked by roscovitine, a CDK5 inhibitor ( Hsu et al., 2013 ), whereas LiCl, a Gsk3β inhibitor, did not affect this phosphorylation ( Figure 1D ), suggesting a CDK5-specific phosphorylation. Altogether, these data demonstrated that CDK5 kinase activity (but not protein accumulation) was diurnal in the SCN.

CDK5 regulates the circadian clock

Since CDK5 activity displayed a diurnal profile in the SCN, we tested whether knock-down of CDK5 in the master clock of the SCN changed circadian behavior in mice. To this end, we tested various shRNAs against Cdk5 in NIH 3T3 fibroblast cells ( Figure 2—figure supplement 1 ) and subsequently injected into the SCN region adeno-associated viral particles containing expression vectors for either a scrambled set of shRNA or a Cdk5 -specific shRNA (variant D, Figure 2—figure supplement 1 ). After recovery from the procedure the animals were transferred into cages containing a running-wheel in order to assess their activity profiles. The control animals expressing the scrambled set of shRNA displayed normal activity in the light-dark (LD) cycle with precise onset of activity at the beginning of the dark phase (ZT12). This onset of activity was less precise in mice with a Cdk5 knock-down (shCdk5) but comparable to animals with a deletion mutation in the clock gene Per2 , designated as Per2 Brdm1 ( Figure 2A , Figure 2—figure supplement 2 ). In constant darkness (DD), χ 2 -periodogram analysis revealed a normal average free-running period for the scramble control mice, whereas for shCdk5 and Per2 Brdm1 , the period was significantly shortened ( Figure 2B ). In one case, the shCdk5 animals became arrhythmic ( Figure 2C ), again comparable to Per2 Brdm1 mice that eventually became arrhythmic in DD as well ( Zheng et al., 1999 ). The total wheel-running activity was significantly reduced in shCdk5 and Per2 Brdm1 mice under DD as well as under LD conditions when compared with the scrambled control animals ( Figure 2—figure supplement 3 ). The reduction of activity in the mutants under LD conditions is confined to the dark phase, but comparable between all three genotypes in the light phase ( Figure 2—figure supplement 4 ). These results indicate that the period of the clock is affected by the lack of Cdk5 expression in the SCN.

CDK5 affects the circadian clock.

( A ) Wheel-running activity of mice (black bins) infected with AAV expressing scrambled control shRNA, or shCdk5, and an animal with a deletion in the Per2 gene ( Per2 Brdm1 ). The actograms are double plotted displaying in one row and below 2 consecutive days. The locomotor activity was confined to the dark period (shaded in gray), while under LD the mice displayed low activity during the light phase (white area). Under DD (continuous gray shaded area) the shCdk5 and Per2 Brdm1 animals show earlier onset of activity each day compared with the control animals. The χ 2 -periodogram analysis for each of the animals is shown below the corresponding actogram to determine the period length (τ). ( B ) Quantification of the circadian period: 23.3 ± 0.1 hr for the control mice (n = 6, black bar), 22.5 ± 0.2 hr for shCdk5 injected mice (n = 6, gray bar), and 22.4 ± 0.1 hr for Per2 Brdm1 mice (n = 4, white bar), (mean ± SEM). One-way ANOVA with Bonferroni’s post-test, **p<0.01. ( C ) In some cases, mice in which Cdk5 was silenced in the SCN became arrhythmic. ( D ) Wheel-running activity (black bins) of Per2 Brdm1 mice infected with AAV expressing scrambled control shRNA (scr), or shRNA against Cdk5 (shCdk5). The actograms are double plotted displaying in one row and below 2 consecutive days. The dark shaded area indicates darkness during which the free-running period was determined. To the right of each actogram the corresponding χ 2 -periodogram is shown. The number in each periodogram indicates the period of the animal. ( E ) Quantification of the circadian period: 22.35 ± 0.03 hr for the scrambled Per2 Brdm1 (n = 3, black bar) and 21.77 ± 0.03 hr for the shCdk5 injected Per2 Brdm1 mice (n = 5, gray bar). Values are the mean ± SEM, t-test, ***p<0.0001. ( F ) 1-way ANOVA test on wild-type and Per2 Brdm1 animals infected with AAV expressing scrambled control shRNA (scr), or shRNA against Cdk5 (shCdk5). N = 3–6 animals, error bars are the mean ± SEM, Bonferroni multiple comparisons test, ***p<0.001.

Interestingly, period in Per2 Brdm1 mutant and wild-type shCdk5 knocked-down mice was not significantly different ( Figure 2B ), suggesting that CDK5 activity is linked to PER2 as indicated by our SDL screen ( Figure 1 ). In order to test the contribution of Cdk5 , we knocked down Cdk5 in Per2 Brdm1 mutant mice. This even further shortened period in Per2 Brdm1 mutant animals compared to scramble control Per2 Brdm1 animals ( Figure 2D,E , Figure 2—figure supplement 5 ). This effect, however, was not simply additive (Δ wt versus wt KD ≈ 0.8 hr; Δ Per2 Brdm1 versus Per2 Brdm1 KD ≈ 0.6 hr, Figure 2F ). Additionally, Δ wt KD versus Per2 Brdm1 KD ≈ 0.8 hr, indicating that the difference in genetic background plays an important role. Overall, our observations suggest that Cdk5 may affect period partially via PER2 but also via additional factors (e.g. CLOCK, Kwak et al., 2013 ). Taken together, it appears that CDK5 is a main regulator of the circadian clock mechanism.

In order to confirm that the different phenotypes were associated with the accumulation levels of CDK5 in control and Cdk5 -silenced mice, we performed immunofluorescence assays on coronal sections of the SCN. Sections were stained with DAPI (blue) in order to label nuclei, with GFP antibody (green) in order to show cells infected by the virus, and with CDK5 antibody (red) in order to compare protein accumulation between the two strains. Scramble as well as shCdk5 mice expressed GFP in the SCN, indicating that the two different viruses infected cells in this brain region ( Figure 3A , Figure 3—figure supplements 1 – 2 ). The expression of Cdk5 was efficiently suppressed in the SCN by the shCdk5 but not by the scrambled shRNA ( Figure 3A , Figure 3—figure supplements 1 – 2 ), indicating that the behavioral phenotypes observed are due to efficient knock-down of Cdk5. The Cdk5 shRNAs was expressed in the SCN (the injection site) and to some extent also dorsal to the SCN but not in distant brain regions (i.e. the piriform cortex) as confirmed by lack of the GFP signal outside of the targeted region ( Figure 3A ).

Immunohistochemistry in the SCN of control and shCdk5 silenced wild type and Per2 Brdm1 mice.

( A ) Representative sections of the SCN region after injection of AAVs carrying either scrambled shRNA, or shCdk5. Slices were stained with DAPI (blue), or anti-GFP (green) and anti-CDK5 (red) antibodies. GFP was used as marker for those cells infected by the virus. CDK5 was efficiently down-regulated in the SCN by shCdk5 (red panels) but not by scrambled shRNA, which was as efficiently delivered as shCDK5. As control, the non-infected piriform cortex from the same animal in which Cdk5 was silenced is shown. Scale bar: 200 µm. ( B ) Analysis of PER2 expression in sections of the SCN of scrambled shRNA, shCdk5 and Per2 Brdm1 mice. Silencing of Cdk5 leads to lack of PER2 (red) compared with control at ZT12, which almost resembles the situation observed in Per2 Brdm1 animals. Blue color: DAPI staining for cell nuclei. Scale bar: 200 µm.

Surprisingly, the phenotypes of shCdk5 and Per2 Brdm1 mice showed considerable similarity, implicating that the levels of PER2 accumulation might be similar in these two different mouse strains. In order to test whether Cdk5 knock-down affected PER2, we stained with DAPI (blue) and immunostained with anti-PER2 (red) SCN sections obtained from control, shCdk5 and Per2 Brdm1 mice perfused at ZT12. PER2 was observed in the SCN of scramble controls, but was strongly reduced in shCdk5 and almost absent in Per2 Brdm1 animals ( Figure 3B , Figure 3—figure supplements 3 – 4 ). These data suggested that CDK5 is a main regulator of the core circadian clock in the SCN and may alter PER2 accumulation and potentially other proteins involved in clock regulation.

CDK5 interacts with PER2 protein in a temporal fashion

A study in Drosophila has shown that several kinases, including cyclin-dependent kinases, phosphorylate specific sites on per to maintain the circadian period ( Garbe et al., 2013 ). Therefore, we aimed to understand whether a molecular interaction exists between CDK5 and PER2, as suggested by our SDL screen ( Figure 1 ). We transfected cells with Per2 and Cdk5 expression vectors and tested whether the two proteins co-immunoprecipitated. We observed that immunoprecipitation with an anti-CDK5 antibody pulled down PER2 protein in two different cell lines ( Figure 4A , Figure 4—figure supplement 1 ). Similar interactions were observed when cells were transfected with expression constructs resulting in PER2 and CDK5 proteins fused to short amino-acid tags of viral protein 5 (V5) and haemaglutinine (HA) fused to them, respectively ( Figure 4B ). Interestingly, interaction between PER2-V5 and CDK5-HA was reduced when roscovitine, which inhibits interaction of CDK5 with its targets ( Hsu et al., 2013 ), was added to the cells ( Figure 4B ). This suggested that active CDK5 protein interacted better with PER2 than CDK5 in its inhibited form.

PER2 interacts with CDK5 in a temporal fashion in the cytoplasm.

( A ) Overexpression of PER2 and CDK5-HA in NIH 3T3 cells and subsequent immunoprecipitation (IP) using an anti-CDK5 antibody. The left panel shows 5% of the input and the right panel co-precipitation of PER2 with CDK5. ( B ) Overexpression of PER2-V5 and CDK5-HA in HEK293 cells in presence or absence of 34 μM roscovitine (CDK5 inhibitor) and DMSO (solvent). Left panel shows 5% of the input and the right panel the immunoprecipitation with anti-CDK5 or without antibody. ( C ) Immunoprecipitation (IP) of PER2 and CDK5 from total mouse brain extract collected at ZT12. Left panel shows the input. The right panel depicts co-immunoprecipitation of PER2 and CDK5 using either anti-CDK5 antibody or anti-GST antibody for precipitation. The middle lane shows PER2-CDK5 co-immunoprecipitation in control animals ( Per2 +/+ ) but not in Per2 -/- mice illustrating the specificity of the PER2-CDK5 interaction. The * in the blot indicates unspecific signal. ( D ) Temporal profile of the PER2-CDK5 interaction in total extracts from SCN tissue around the clock. Input was analyzed by immunoblot using anti-CDK5, anti-PER2, and anti-HSP90 antibodies (left panel). CDK5 co-immunoprecipitated PER2 in a diurnal fashion with a peak between ZT12 and ZT16. The statistical analysis of the PER2/CDK5 signal around the clock is shown below (one-way ANOVA with Bonferroni’s post-test, n = 3, *p<0.0001, values are mean ± SEM). * in the blot indicates unspecific signal. ( E ) Immunoprecipitation of PER2 with CDK5 from cytoplasmic and nuclear brain extracts collected at ZT12. The left panel shows the input and the right panel co-IP of PER2 and CDK5, which occurs only in the cytoplasm but not in the nucleus. The smaller band detected by the anti-PER2 antibody depicts an unspecific band that is smaller than PER2. * in the blot indicates unspecific signal. ( F ) Slices from the SCN obtained at ZT12 were immunostained with PER2 antibody (green), CDK5 (red), and nuclei were marked with DAPI (blue). Co-localization of the two proteins results in the yellow color. Scale bar: 10 µm. The z-stacks right and below the micrograph confirm co-localization of PER2 and CDK5 (yellow). ( G ) Purification of the N-terminal half of PER2 (1–576) or the C-terminal half (577–1256) (left panel, coomassie). CDK5-His was pulled down by both recombinant PER2 attached to the glutathione resin, but only the C-terminal was able to retain CDK5 (immunoblot using anti-His antibody, right panel).

In order to test whether this interaction could be observed in tissue, we prepared total brain extracts at ZT12, when kinase activity of CDK5 was high ( Figure 1C ). At two different salt concentrations, we could pull-down PER2 and CDK5 using either anti-CDK5 or anti-PER2 antibodies ( Figure 4—figure supplement 2 ). The specificity of the signals was confirmed by using brain extracts from Per2 -/- mice ( Chavan et al., 2016 ) that completely lack PER2 protein ( Figure 4C ). Next, we wanted to investigate whether the interaction between the two proteins is time of day-dependent in the SCN. Total extracts of SCN tissue at ZT0, 4, 8, 12, 16 and 20 were prepared and immunoprecipitation with an anti-CDK5 antibody pulled down PER2 at ZT8, 12, and 16, with the strongest signals at ZT12 and ZT16 ( Figure 4D ). Taken together, these observations suggested that the interaction between CDK5 and PER2 can occur in brain tissue and that in the SCN this interaction was time of day-dependent. This observation was confirmed on SCN tissue sections, where we observed PER2 expression at ZT12 but less at ZT0 with co-localization of CDK5 restricted to ZT12 ( Figure 4—figure supplement 3 ).

Next, we tested in which subcellular compartment the interaction between CDK5 and PER2 takes place. We prepared nuclear and cytoplasmic extracts from total brain tissue and performed immunoprecipitation using an anti-CDK5 antibody. PER2 could only be observed in the cytoplasmic but not the nuclear fraction ( Figure 4E ). This was supported by the observation that the two proteins were co-localized only in the cytoplasm in SCN tissue ( Figure 4F , yellow color).

Furthermore, we evaluated with which part of PER2 the CDK5 protein interacts. We tested whether deletions in the PAS-domain of PER2, a known domain for protein interactions ( Ponting and Aravind, 1997 ), influenced CDK5 binding. No significant effect of deletions of the PAS-A and PAS-B domains on the interaction was observed ( Figure 4—figure supplement 4 ). Next, we generated expression vectors coding either for the N-terminal (1-576) or the C-terminal part (577–1257) of PER2 fused to GST ( Figure 4—figure supplement 5 ). The recombinant forms of PER2 and histidine-tagged CDK5 were produced in bacteria. A pull-down assay with these proteins showed that the C-terminal but not the N-terminal half of the PER2 protein was pulled-down by CDK5, suggesting that CDK5 binds to the C-terminal part of PER2 ( Figure 4G ). This does, however, not exclude weak interactions of the CDK5 protein with the N-terminal half in vivo. Taken together, our data suggest a physical interaction of PER2 and CDK5 in the cytoplasm.

CDK5 phosphorylates PER2 at serine 394

In order to understand whether CDK5 phosphorylates the PER2 protein we overexpressed the N-terminal and C-terminal parts of PER2 fused to GST in bacteria ( Figure 5—figure supplement 1 ) and performed an in vitro kinase assay with the recombinant proteins. Recombinant CDK5/p35 protein complex along with γ- 32 P labeled ATP resulted in phosphorylation of the N-terminal part of the PER2 protein with a main signal at around 120 kD ( Figure 5A , Figure 5—figure supplement 2 , 32 P panels). Addition of roscovitine abolished phosphorylation of PER2 whereas LiCl had no effect ( Figure 5—figure supplement 3 ). Interestingly, no phosphorylation of the C-terminal part of PER2 was observed, only a signal corresponding to the auto-phosphorylation of CDK5 was detected at around 60 kD ( Figure 5A , 32 P panel).

CDK5 phosphorylates PER2 at S394.

( A ) An in vitro kinase assay was performed using recombinant CDK5/p35 and either GST-PER2 1–576 or GST-PER2 577–1256 as substrate. The samples were subjected to 10% SDS page (Coomassie, left panel) and the phosphorylation of PER2 was detected by autoradiography in order to visualize 32 P-labeled proteins (right panel). CDK5 phosphorylates the N-terminal half (1-576) of a GST-PER2 fusion protein whereas the C-terminal half (577–1257) is not phosphorylated. The signal for CDK5/p35 alone indicates CDK5 auto-phosphorylation seen in all lanes when CDK5 is present. ( B ) Annotated mass spectrum of the tryptic peptide PER2 383-397 ILQAGGQPFDYpSPIR containing the phosphorylated residue S394. The red color depicts the y-ion series (1-12) and blue the b-ion series (2–7, a2); y5-98, y8-98, y11-98 show the de-phosphorylated ions. ( C ) In vitro kinase assay was performed as in ( A ). The putative phosphorylation site was mutated to aspartic acid (S394D) or glycine (S394G). Both mutations abrogated the CDK5-mediated phosphorylation. Coomassie staining reveals equal expression of the GST-PER2 fragments. The bar diagram at the right shows the quantification of three experiments. One-way-ANOVA with Bonferroni’s post-test, *: p<0.001 ( D ) The monoclonal antibody produced against P-S394-PER2 does recognizes PER2 (1–576) S394 phosphorylation mediated by CDK5/p35 in presence but not in absence of the kinase or when CDK5 is inactivated by roscovitine. This antibody does not recognize the S394G mutated form even in presence of CDK5/p35. ( E ) Temporal profile of P-S394-PER2 and total PER2 in SCN tissue. Upper panels show western blots of the corresponding proteins indicated on the right. Below the quantification of three experiments is shown, in which the value at ZT12 of PER2 has been set to 1. The data were double plotted. Values are the mean ± SEM. Two-way ANOVA with Bonferroni’s multiple comparisons revealed that the two curves are significantly different with p<0.0001, F = 93.65, DFn = 1, DFd = 48. ( F ) Subcellular localization of P-S394-PER2. Total wild-type mouse brain extracts were separated into cytoplasmic (HSP90 positive) and nuclear (laminB positive) fractions. Phosphorylated PER2 was predominantly detected in the cytoplasm with the P-S394-PER2 antibody, whereas the general PER2 antibody detected PER2 in both compartments with higher amounts in the nuclear fraction.

Next, we aimed to identify the phosphorylation site(s) in the N-terminal part of PER2 using the recombinant protein, which was phosphorylated by CDK5/p35 in vitro. Mass spectrometry revealed several phosphorylation sites at serine and threonine residues, respectively ( Supplementary file 1 ). One of the serine residues of PER2 was located within a CDK5 consensus sequence and had the highest probability score for being phosphorylated ( Figure 5B ). The serine residue at position 394 (S394) of PER2 is located at the end of the PAS domain and within the deletion of the mutated PER2 of Per2 Brdm1 mice ( Zheng et al., 1999 ). This suggested that CDK5/p35 phosphorylates S394 and that this phosphorylation is of functional relevance. Mutations of this serine to aspartic acid (S394D) or glycine (S394G) reduced phosphorylation by CDK5/p35 significantly ( Figure 5C ), confirming that CDK5/p35 phosphorylated S394. Next, we produced a monoclonal antibody against the phosphorylated serine at 394 of PER2 (P-S394-PER2) ( Figure 5—figure supplements 4 – 6 ). With this antibody we detected the phosphorylated N-terminal fragment of PER2 in presence of CDK5/p35 but not when S394 was mutated to glycine (S394G) or when CDK5 was inhibited by roscovitine ( Figure 5D ), confirming S394 phosphorylation by CDK5/p35.

In order to determine whether PER2 phosphorylation at S394 is time of day-dependent, we collected SCN tissue every 4 hr. The P-S394-PER2 specific antibody detected highest phosphorylation at ZT12 with weaker or no phosphorylation at other time points indicating that S394 is phosphorylated in a time of day-dependent manner ( Figure 5E ). Fractionation of wild-type brain cellular extracts prepared at ZT12 into nuclear and cytoplasmic parts showed phosphorylated S394 predominantly in the cytoplasm with little or no signal in the nucleus when labeled with the P-S394-PER2 antibody ( Figure 5F ). Total PER2 was observed in both cellular compartments with higher levels in the nucleus ( Figure 5F ). This suggested that phosphorylation of S394 of PER2 happens predominantly in the cytoplasm and that this phosphorylation is either removed or occluded when PER2 enters the nucleus.

CDK5 affects stability and nuclear localization of PER2

To evaluate the function of CDK5-driven PER2 phosphorylation, we wanted to determine whether CDK5 affects PER2 stability. We treated NIH 3T3 cells with roscovitine and DMSO as control and determined endogenous levels of PER2. We observed that roscovitine treatment of cells reduced PER2 levels, suggesting that CDK5 can affect protein stability ( Figure 6A ). In order to challenge this observation, we deleted Cdk5 in NIH 3T3 cells using the CRISPR/Cas9 method ( Figure 6—figure supplements 1 – 3 ). We observed that deletion of Cdk5 led to reduced amounts of PER2 ( Figure 6B ), consistent with the data shown in Figure 6A . These observations support the notion that phosphorylation by CDK5 affects PER2 abundance. In order to monitor PER2 stability, we knocked down Cdk5 using the shRNA D ( Figure 2—figure supplement 1 ). We observed that increasing amounts of shCdk5 dampened PER2 levels proportionally to the decreasing CDK5 levels ( Figure 6C ).

CDK5 affects PER2 stability and nuclear localization.

( A ) Western blot of NIH 3T3 cell extracts with and without roscovitine treatment. When roscovitine inhibited CDK5, less PER2 protein was detected in cell extracts. The bar diagram below shows values (mean ± SEM) of three experiments with significant differences between roscovitine treated and untreated cells, t-test, *p<0.001. ( B ) CRISPR/Cas9-mediated knockout of Cdk5 in NIH 3T3 cells. Western blot shows absence of PER2 in cells when Cdk5 is deleted. ( C ) Titration of CDK5 knock-down as revealed by Western blotting. PER2 levels decreased proportionally to increasing amounts of shCdk5. 2.7 μM of shCdk5 (red) was used for subsequent experiments. The value without shCdk5 was set to 1. One-way ANOVA with Bonferroni post-test, n = 4, ***p<0.001, ****p<0.0001, mean ± SD. The * in the blot indicates unspecific signal. ( D ) Temporal profile of protein abundance in NIH 3T3 cells 0, 3 and 6 hr after inhibition of protein synthesis by 100 μM cycloheximide (CHX) in presence of scrambled shRNA, or shCdk5, respectively (2.7 μM of the respective shRNA was used). The diagram below shows quantification of PER2 protein over time. Linear regression with 95% confidence intervals (hatched lines) indicates that knock-down of Cdk5 leads to less stable PER2 (shCdk5 t 1/2 =4h, scr t 1/2 =11h). Two-way ANOVA with Bonferroni’s post-test revealed that the two curves are significantly different, n = 3, p<0.01, F = 24.53, DFn = 1, DFd = 4. ( E ) Inhibition of the proteasome by epoxomycin in cells with shCdk5 leads to amounts of PER2 that are higher compared with the levels without epoxomycin treatment and are comparable to the levels observed in cells without Cdk5 knockdown. Diagram below displays the quantification of three experiments. Scrambled shRNA values were set to 1. One-way ANOVA with Bonferroni’s post-test shows no significant reduction of PER2 in shCdk5 cells in presence of epoxomycin, but significantly lower values in absence of epoxomycin when compared with scrambled shRNA treatment. One-way ANOVA with Bonferroni’s post-test, n = 3, p<0.001. ( F ) PER2 abundance in nuclear extracts of NIH 3T3 cells. Knockdown of Cdk5 reduces PER2 levels in the nucleus as revealed by Western blotting. HSP90 = cytosolic marker, LaminB = nuclear maker. ( G ) Immunofluorescence of PER2 (red) at ZT12 in mouse SCN sections after infection with AAV (green) expressing scrambled shRNA (left panel), or shCdk5 (right panel). Nuclei are visualized by DAPI staining (blue). PER2 can only be observed in the nucleus in presence (white arrow heads) but not in absence of CDK5 (white arrow). Scale bar = 7.5 µm. ( H ) Co-immunoprecipitation of CRY1 by PER2 in NIH 3T3 cells. Substitution of S394 to G in PER2 reduces the levels of co-precipitated CRY1 (right panel). The left panel shows the input. The bar diagram on the right displays the quantification of three experiments, where the amount of precipitated CRY1 by PER2 is set to 1. Paired t-test reveals a significant difference between the amounts of CRY1 precipitated by PER2 and the S394G PER2 mutation, n = 3, *p<0.05, mean ± SD.

In order to determine whether CDK5 modulates degradation of PER2, we blocked protein synthesis using cycloheximide. Under conditions that partially knocked down Cdk5 (at a concentration of 2.7 µM of shCdk5, Figure 6C ), we measured PER2 and CDK5 protein levels over 6 hr after cycloheximide treatment. We found that degradation of PER2 was faster when Cdk5 was knocked down compared with unspecific shRNA treatment (shCdk5 t 1/2 =4 h, scr t 1/2 =11 h) ( Figure 6D ), indicating that reduction of Cdk5 accelerated PER2 degradation. Next, we investigated whether PER2 degradation involved the proteasome. Cells were treated with epoxomycin, a proteasome inhibitor, or with the solvent DMSO. In line with our previous experiments, shCdk5 treatment efficiently knocked down CDK5 and reduced PER2 levels compared with scrambled shRNA treatment. Addition of epoxomycin, but not DMSO, significantly increased PER2 levels despite absence of CDK5 ( Figure 6E ), indicating that PER2 degradation involved the proteasome. Residual amounts of CDK5 in the cells still may phosphorylate PER2 and direct it into the nucleus. Therefore, we wanted to see whether PER2 could be detected in nuclear extracts of shCdk5 knocked down cells. In line with our previous observations we did not detect PER2 in nuclear extract ( Figure 6F ), supporting the idea that PER2 needed to be phosphorylated by CDK5 in order to enter the nucleus. Data from immunofluorescence experiments on SCN sections were in line with this hypothesis. PER2 was only detected in nuclei when CDK5 was available ( Figure 6G , arrowheads, Figure 6—figure supplement 4 ), but not when shCdk5 was expressed in SCN cells ( Figure 6G , white arrow, Figure 6—figure supplement 4 ).

It has been described that nuclear entry of PER2 involves CRY1 ( Kume et al., 1999 ; Ollinger et al., 2014 ). In addition, CRY1-mediated hetero-dimerization stabilizes PER2 by inhibiting its own ubiquitination ( Yagita et al., 2000 ). Therefore, we tested the interaction potential of wild-type PER2 and the S394G PER2 mutation with CRY1 by overexpressing the two PER variants in NIH 3T3 cells. Immunoprecipitation of wild-type PER2 pulled down CRY1; however, the S394G PER2 mutation was significantly less efficient in doing so ( Figure 6H ). The small amounts of CRY1 detected may be bound to endogenous PER2 that is present in the cells. In summary, these experiments suggested that CDK5 affects PER2 stability, interaction with CRY1, and nuclear localization.

Not only do kinases play a crucial role in signal transduction in response to extracellular stimuli, but they also regulate cycling processes such as the cell cycle and circadian rhythms. Most cyclin dependent kinases (CDKs) regulate the cell cycle, with few exceptions such as the cyclin dependent kinase 5 (CDK5). This kinase is ubiquitously expressed and its function is vital in post-mitotic neurons, where other CDKs are not active. Although CDK5 is not implicated in cell cycle progression, it can aberrantly activate components of the cell cycle when it is dysregulated in post-mitotic neurons, leading to cell death ( Chang et al., 2012 ). Interestingly, cell death is affected by the clock component PER2 as well ( Magnone et al., 2014 ), suggesting that both, CDK5 and PER2 act in the same pathway, or that their pathways cross at a critical point during the regulation of cell death. The synthetic dosage lethal screen that we performed in yeast supports this notion, as expression of PER2 in yeast lacking Cdk5 strongly and significantly compromised growth ( Figure 1A ).

The kinase CDK5 displays many effects that ensure proper brain function and development. Mice deficient for Cdk5 are perinatal lethal ( Gilmore et al., 1998 ; Ohshima et al., 1996 ). CDK5 influences cortical neuron migration, cerebellar development, synapse formation and plasticity ( Kawauchi, 2014 ). Here, we identified a new role for this kinase, that is the regulation of the circadian clock in vivo. Previously, CDK5 had been identified to phosphorylate CLOCK and thereby regulate CLOCK stability and cellular distribution in cells ( Kwak et al., 2013 ). In the SCN, however, NPAS2 may replace the function of CLOCK ( Debruyne et al., 2006 ; DeBruyne et al., 2007 ) and therefore phosphorylation of CLOCK by CDK5 may play a minor role in the SCN. Hence, to unravel the novel function of CDK5 in the circadian oscillator, we had to restrict ourselves to the use of SCN tissue and whole animals.

CDK5 activity, but not its protein accumulation, displays a diurnal profile in the SCN with high activity during the night and low activity during the day ( Figure 1C ). The activity displayed a typical on/off profile similar to other CDKs. This finding raises the question how this diurnal activity of CDK5 may be achieved. On one hand, ATP accumulation, which is required for phosphorylation, peaks during the night in the SCN ( Yamazaki et al., 1994 ). On the other hand, CDK5 activity is regulated by cofactors. Depending on its cofactor, CDK5 in the brain phosphorylates targets involved in neurodegenerative diseases (e.g. Tau, MAP1B), neuronal migration (e.g. DCX), and synaptic signaling (e.g. Ca v 2.2, Dynamin1, NR2A, DARPP-32) ( Kawauchi, 2014 ). The most obvious candidates to regulate its time-dependent activity would be cyclins D1 and E, which inhibit CDK5, or cyclin I, which activates it. Alternatively, other known CDK5 regulators such as p35 may be involved ( Shah and Lahiri, 2014 ). Most likely, positive and negative feedback loops of other kinases and phosphatases are necessary to generate the on/off profile, although the components involved in this mechanism are probably different from the ones known for CDKs that regulate the cell cycle. Interestingly, CK1 phosphorylates and activates CDK5 in vitro ( Sharma et al., 1999 ) and CDK5 is thought to phosphorylate and inhibit CK1δ in vitro ( Ianes et al., 2016 ; Eng et al., 2017 ) potentially establishing a feedback loop between the two kinases. However, additional research is needed to determine the precise mechanism of diurnal on/off activation of CDK5.

As evidenced in Figure 2 , Cdk5 knock-down affects circadian clock period at the behavioral level. The shortening of period in mice with knocked-down Cdk5 is comparable to mice containing a mutation of the Per2 gene ( Per2 Brdm1 mutant mice, Zheng et al., 1999 ). Interestingly, however, knock-down of Cdk5 in Per2 Brdm1 mutant mice leads to further shortening of circadian period. This suggests that Cdk5 may affect period either independently of Per2 or, while PER2 may still be important, CDK5 regulates other proteins important for clock function. Since the difference between the period in control versus Cdk5 knock-down (0.8 hr, black and gray bars, Figure 2F ) is not the same as in Per2 Brdm1 mutant versus its Cdk5 knock-down (0.6 hr, red and rose bars, Figure 2F ) the second possibility is more likely. Moreover, wt KD versus Per2 Brdm1 KD show a difference in period ( Figure 2F , gray and rose bars), suggesting that the difference in genotype plays an important role. From a dynamic perspective, it is possible that lack of PER2 protein will unmask Cdk5 targets that otherwise would be phosphorylated less efficiently or not at all. For example, the PER2 site that is phosphorylated by CDK5 (PFD Y S PIR) is very similar in PER1 (PFD H S PIR). If PER1 would be phosphorylated by CDK5 at this site at the same rate as PER2, then PER1 as well as PER2 should be absent in the nucleus of SCN cells. This would correspond to a PER1/PER2 double knock-out, which become immediately arrhythmic when subjected to constant darkness ( Zheng et al., 2001 ). This is not the phenotype we observe in the Cdk5 knock-down mice and hence it is unlikely that CDK5 affects PER1 in the same manner as it affects PER2. However, in the absence of PER2 the dynamics may change and PER1 may become a better target for CDK5 and influence period. This view is consistent with the observation that knock-down of Cdk5 in Per2 Brdm1 mutant mice can shorten period ( Figure 2D,E ).

CDK5 binds to the C-terminal half of PER2 ( Figure 4G ) and phosphorylates it at S394 ( Figure 5 ), which is located in the PAC domain of the N-terminal half of the protein. Hence, the binding and phosphorylation sites are far apart, suggesting a structure of PER2 allowing proximity of the CDK5 binding and phosphorylation domains. We cannot exclude weak binding of CDK5 to the N-terminal half of PER2, because phosphorylation at S394 occurs in vitro even in the absence of the C-terminal half of the PER2 protein ( Figure 5A ). This may be due to the fact that the N-terminal half is overexpressed in vitro, which strongly increases the probability of phosphorylation by CDK5 even in the absence of the C-terminal binding domain. It is also known that p35 (which is used in the in vitro kinase assay to activate CDK5) can increase the interaction between CDK5 and its targets ( Hsu et al., 2013 ).

In SCN tissue PER2 phosphorylation at S394 appears to be time of day-dependent, with highest levels at ZT12 and ZT16 ( Figure 5E ) when CDK5 activity is high ( Figure 1C ). Compared with total PER2 protein the S394 phosphorylated form appears to be slightly advanced in its phase. The difference in phase is probably even larger than it appears here, because the polyclonal antibody that detects total PER2 also detects the phosphorylated S394 PER2 variant. This is especially important in the rise of the signal detected, which appears to be identical in Figure 5E . Probably the steep increase between ZT8 and ZT12 represents the S394 phosphorylated forms in both curves. In contrast, the decrease in PER2 levels differs between total PER2 and P-S394-PER2 form. Consistent with previous studies total PER2 peaks in the nucleus at ZT16 in the SCN ( Nam et al., 2014 ) when P-S394-PER2 is not detected anymore. This highlights that additional post-translational modifications of PER2 exist ( Toh et al., 2001 ; Vanselow et al., 2006 ) and that P-S394-PER2 disappears faster compared with other modified forms. Probably, P-S394-PER2 plays a role in PER2 dynamics in terms of shuttling from the cytoplasm to the nucleus, because P-S394-PER2 can only be observed in the cytoplasmic and not the nuclear fraction ( Figure 5F ). The phosphorylation of PER2 by CDK5 may therefore be critical for the assembly of a macromolecular complex in the cytoplasm ( Aryal et al., 2017 ), which then enters the nucleus.

The difference in the decline between PER2 and its S394 phosphorylated form in the SCN may suggest a role of the S394 phosphorylation not only for nuclear transport but also for PER2 protein stability. The earlier decline of the P-S394-PER2 signal compared with total PER2 ( Figure 5F ) might suggest that the S394 phosphorylated form is less stable. Apparently, the opposite is the case, as shown in Figure 6 . Pharmacological inhibition of CDK5 ( Figure 6A ), CRISPR/Cas9-mediated knock-out of Cdk5 ( Figure 6B ), and shRNA-mediated knock-down of Cdk5 ( Figure 6C ) all led to reduced levels of PER2 in cells. The half-life of PER2 is clearly increased in the presence of CDK5, rising from about 4 hr to 11 hr, indicating that phosphorylation at S394 has a stabilizing function. This is in accordance with previous results that described almost absent levels of PER2 in the Per2 Brdm1 mutant mice ( Zheng et al., 1999 ). This mouse strain expresses a PER2 lacking 87 amino acids in the PAS and PAC domains, where the S394 and the CDK5 consensus sequence are localized. CDK5 cannot phosphorylate this mutant PER2 and therefore the protein is less stable. As a consequence, the formation of the macromolecular complex responsible for nuclear transport of PER2 is disturbed. This results in a temporal change of BMAL1/CLOCK/NPAS2 activity, shortening the clock period. Accordingly, Per2 Brdm1 mutant mice display a short period or no circadian period in constant darkness ( Zheng et al., 1999 ), similar to the phenotype observed for the CDK5 knock-down mice ( Figure 2B ).

PER2 stability is affected by CK1δ/ε, which phosphorylate PER2 at several sites and regulate degradation of PER2 via the proteasome ( Eide et al., 2005 ; Xu et al., 2007 ; Narasimamurthy et al., 2018 ). This effect is similar to the action of dbt on Drosophila per. Interestingly, CDK5 can phosphorylate CK1δ to reduce its activity ( Ianes et al., 2016 ). This phosphorylation could cross-regulate the activities of both kinds of kinases to fine-tune the amount of PER2. This is evidenced by the observation, that knock-down of Cdk5 in Per2 Brdm1 mutant mice further shortens period in these animals ( Figure 2D,E ). The mammalian orthologue of shg, Gsk3β, does not phosphorylate the mammalian Tim but the nuclear receptor NR1D1 ( Mukherji et al., 2015 ). This change in substrate may be related to the shift in function of the CRYs to replace Tim in the mammalian circadian oscillator. Similar to shg, CDK5 phosphorylation of PER2 increases its half-life ( Figure 6D ). Lack of CDK5, and therefore lack of phosphorylation at S394 of PER2, leads to proteasomal degradation of PER2 as evidenced by epoxomycin treatment, which inhibits the proteasome and reduces the decline of PER2 levels in the cell ( Figure 6E ). This is consistent with a recent report that describes the ubiquitin ligase MDM2 as controlling PER2 degradation via the proteasome ( Liu et al., 2018 ). However, it is not clear whether it is the phosphorylation at S394 per se or the capacity to participate in a macromolecular complex to enter the nucleus that stabilizes PER2. In any case, this phosphorylation appears to be essential for nuclear entry of PER2 ( Figure 6F,G ).

A recent report showed that mammalian PER represses and de-represses transcription by displacing BMAL1-CLOCK from promoters in a CRY-dependent manner ( Chiou et al., 2016 ). Our data support these findings. PER2 containing a S394G mutation, which abolishes CDK5-mediated phosphorylation, displayed reduced interaction potential with CRY1 ( Figure 6H ). Because CRY1 is involved in nuclear transport of PER2 ( Kume et al., 1999 ; Ollinger et al., 2014 ; Yagita et al., 2000 ), lack of interaction with the S394G mutant form of PER2 leaves this protein in the cytoplasm, unable to enter the nucleus ( Figure 6G ). The present data are also in agreement with previous experiments in which we investigated the role of protein phosphatase 1 (PP1) and its effects on the circadian clock ( Schmutz et al., 2011 ). Expression of a specific PP1 inhibitor in the brain lengthened circadian period and increased PER2 levels and its nuclear accumulation in neurons. These effects are all opposite to what we observe when PER2 is not phosphorylated at S394. Therefore, it could be speculated that PP1 is involved in the dephosphorylation of P-S394, thereby counterbalancing phosphorylation of this site by CDK5.

Taken together, our results indicate that CDK5 potentially affects several proteins that regulate circadian clock period. In particular, we find that CDK5 phosphorylates PER2 at S394. This phosphorylation appears to be important for PER2 to bind efficiently to CRY1 in order to allow entry of PER2 into the nucleus. Inhibition of CDK5 in cells leads to degradation of PER2 in the proteasome ( Figure 7 ). Inhibition of CDK5 in vivo inhibits nuclear entry of PER2 and shortens period to a similar extent as observed in Per2 Brdm1 mutant mice, which express a barely detectable level of protein lacking 87 amino acids including S394. Taken together, CDK5 regulates the circadian clock and influences PER2 nuclear transport via phosphorylation. Because PER2 is involved in several physiologically relevant pathways in addition to clock regulation ( Albrecht et al., 2007 ), PER2 may mediate several biological functions that were previously linked to CDK5, such as the regulation of the brain reward system ( Benavides et al., 2007 ; Bibb et al., 2001 ) and psychiatric diseases ( Engmann et al., 2011 ; Zhu et al., 2012 ).

Model illustrating the regulation of PER2 by CDK5.

The upper row illustrates phosphorylation of PER2 at S394 by CDK5 that subsequently favors interaction with CRY1 and leads to transport into the nucleus, where the PER2/CRY1 complex inhibits BMAL1/NPAS2 (or in the periphery CLOCK)-driven transcriptional activation. Of note is that CDK5 potentially phosphorylates other clock relevant components, such as CLOCK, PER1 and CKI. The lower part illustrates that inhibition of CDK5 leads to a lack of S394 PER2 phosphorylation, which renders the PER2 protein more prone to degradation by the proteasome. CRY1 does not form a complex with PER2 and hence PER2 is not transported into the nucleus. CRY1 enters the nucleus independently and can inhibit the BMAL1:NPAS2 (or in the periphery CLOCK) transcriptional complex. This model is consistent with the dual modulation of transcriptional inhibition ( Ye et al., 2014 ; Xu et al., 2015 ). Transcriptional inhibition is modulated in an intricate unknown manner by various additional factors (gray) ( Aryal et al., 2017 ) that may be cell type specific.

Animals and housing

All mice were housed with food and water ad libidum in transparent plastic cages (267 mm long ×207 mm wide ×140 mm high; Techniplast Makrolon type 2 1264C001) with a stainless-steel wire lid (Techniplast 1264C116), kept in light- and soundproof ventilated chambers. All mice were entrained to a 12:12 hr LD cycle, and the time of day was expressed as Zeitgeber time (ZT; ZT0 lights on, ZT12 lights off). Two- to four-month-old males were used for the experiments. Housing as well as experimental procedures were performed in accordance with the guidelines of the Schweizer Tierschutzgesetz and the declaration of Helsinki. The state veterinarian of the Canton of Fribourg approved the protocol. The floxed Per2 mice ( Chavan et al., 2016 ) are available at the European Mouse Mutant Archive (EMMA) strain ID EM:10599, B6;129P2-Per2 tm1Ual /Biat.

Synthetic dosage lethal (SDL) screen

The SDL screen was essentially performed as described earlier ( Measday et al., 2005 ; Tong, 2001 ). Briefly, the bait strain Y2454 (MATα mfa1Δ::MFA1pr-HIS3, can1Δ, his3Δ1, leu2Δ0, ura3∆0, lys2Δ0 ) carrying the plasmid YCplF2- mPer2 (that drives expression of PER2 from the galactose-inducible GAL1 promoter) was inoculated into 50 mL glucose-containing synthetic dropout medium lacking leucine (SD-Leu) and grown at 30°C overnight with shaking. Cells were then centrifuged, resuspended in 20 mL of the supernatant, poured into a sterile rectangular petri dish, spotted in a 96-well format on rectangular SD-Leu plates (coined ‘bait plates’ hereafter) using a Biomek 2000 robot (Beckman Coulter, USA), and then grown at 30°C for 3 days. In parallel, the gene deletion array in the strain BY4741 (MATa his3∆1, leu2∆0, met15∆0, ura3∆0 ) was spotted from the storage plates onto fresh G418-containing YPD plates (96-well format) and also grown at 30°C for 3 days. For the mating procedure (overnight at 30°C), colonies from bait plates were (robotically) spotted onto plates containing YPD (plus adenine) and the colonies from the gene deletion array plates were (each separately and in duplicate) spotted on top of them. The next day, the colonies were transferred to G418-containing SD plates lacking lysine, methionine, and leucine (SD-Lys/Met/Leu/+G418) to select for diploids that harbour the YCplF2- mPer2 plasmid. Diploids were then spotted onto plates containing sporulation medium (10 g L −1 potassium acetate, 1 g L −1 yeast extract, 0.1 g L −1 glucose, 2% w/v agar, supplemented with uracil, histidine, and G418) and incubated at 24°C. After 9 days, tetrads were observed and the colonies were transferred to canavanine-containing SD plates lacking arginine and histidine (SD-Arg/His/+canavanine) to select for MATa haploids. Following growth at 30°C for three days, a second haploid selection was carried out by spotting the colonies on SD-Arg/His/Leu/+canavanine plates (to select for MATa haploids containing the YCplF2- mPer2 plasmid). Following growth at 30°C for 2 days, a third haploid selection was carried out by spotting cells on SD-Arg/His/Leu/+canavanine/+G418 plates (to select for MATa haploids containing the YCplF2- mPer2 plasmid as well as the respective gene deletions of the yeast knockout collection). Following incubation at 30°C for 5 days, colonies were then spotted in parallel onto SD-Arg/His/Leu/+G418 plates and on SD-Raf/Gal-Arg/His/Leu/+G418 plates (containing 1% raffinose and 2% galactose as carbon sources) to induce expression of PER2. Both types of plates were incubated at 30°C for 4 days and photographed every day. Strains that grew significantly less well on SD-Raf/Gal-Arg/His/Leu/+G418 than on SD-Arg/His/Leu/+G418 included eap1∆ , gnd1∆ , and pho85∆ . In control experiments, the respective original yeast knockout collection mutants were transformed in parallel with the YCplF2- mPer2 or the empty YCplF2 plasmid ( Foreman and Davis, 1994 ), selected on SD-Leu plates, grown overnight in liquid SD-Leu, spotted (10-fold serial dilutions) on SD-Raf/Gal-Leu plates, and grown for 3 days at 30° ( Figure 1A ). Please note that all media containing G418 were made with glutamate (1 g L −1 ) instead of ammonium sulfate as nitrogen source, as recommended in Tong (2001) .

Adeno Associate Virus (AAV) production and stereotaxic injections

Adeno Associate Viruses (AAVs) were produced in the Viral Vector Facility (ETH Zurich). Plasmids used for the production are available on the VVF web site. Two constructs were produced. ssAAV-9/2-hSyn1-chI[mouse(shCdk5)]-EGFP-WPRE-SV40p(A) carried the shRNA against Cdk5 (shD, see Figure 2—figure supplement 1 and Supplementary file 2 ) which knocked down only neuronal Cdk5 . ssAAV-9/2-hSyn1-chI[1x(shNS)]-EGFP-WPRE-SV40p(A) was the scrambled control.

Stereotaxic injections were performed on 8-week-old mice under isofluorene anaesthesia using a stereotaxic apparatus (Stoelting). The brain was exposed by craniotomy and the Bregma was used as reference point for all coordinates. AAVs were injected bilaterally into the SCN (Bregma: anterior-posterior (AP) − 0.40 mm; medial-lateral (ML) ±0.00 mm; dorsal-ventral (DV) – 5.5 mm, angle + /- 3°) using a hydraulic manipulator (Narishige: MO-10 one-axis oil hydraulic micromanipulator, http://products.narishige-group.com/group1/MO-10/electro/english.html ) at a rate of 40 nL/min through a pulled glass pipette (Drummond, 10 µl glass micropipet; Cat number: 5-000-1001-X10). The pipette was first raised 0.1 mm to allow spread of the AAVs, and later withdrawn 5 min after the end of the injection. After surgery, mice were allowed to recover for 2 weeks and entrained to LD 12:12 prior to behavior and molecular investigations.

Locomotor activity monitoring

Analysis of locomotor activity parameters was done by monitoring wheel-running activity, as described in Jud et al. (2005) , and calculated using the ClockLab software (Actimetrics). Briefly, for the analysis of free-running rhythms, animals were entrained to LD 12:12 and subsequently released into constant darkness (DD). Internal period length (τ) was determined from a regression line drawn through the activity onsets of ten days of stable rhythmicity under constant conditions. Total and daytime activity, as well as activity distribution profiles, was calculated using the respective inbuilt functions of the ClockLab software (Acquisition Version 3.208, Analysis version 6.0.36). Numbers of animals used in the behavioral studies are indicated in the corresponding figure legends.

Immunofluorescence

Animals used for the immunohistochemistry were killed at appropriate ZTs indicated in the corresponding figure legends. Brains were perfused with 0.9% NaCl and 4% PFA. Perfused brains were cryoprotected by 30% sucrose solution and sectioned (40 µm, coronal) using a cryostat. Sections chosen for staining were placed in 24-well plates (two sections per well), washed three times with 1x TBS (0.1 M Tris/0.15 M NaCl) and 2x SSC (0.3 M NaCl/0.03 M tri-Na-citrate pH 7). Antigen retrieval was performed with 50% formamide/2x SSC by heating to 65°C for 50 min. Then, sections were washed twice in 2x SSC and three times in 1x TBS pH 7.5, before blocking them for 1.5 hr in 10% fetal bovine serum (Gibco)/0.1% Triton X-100/1x TBS at RT. After the blocking, the primary antibodies, rabbit anti-PER2-1 1:200 (Alpha Diagnostic, Lot numb. 869900A1.2-L), mouse anti-Cdk5 clone 2H6 1:20 (Origene, Lot numb. A001), and rabbit anti-GFP 1:500 (abcam ab6556) diluted in 1% FBS/0.1% Triton X-100/1x TBS, were added to the sections and incubated overnight at 4°C. The next day, sections were washed with 1x TBS and incubated with the appropriate fluorescent secondary antibodies diluted 1:500 in 1% FBS/0.1% Triton X-100/1x TBS for 3 hr at RT. (Alexa Fluor 488-AffiniPure Donkey Anti-Rabbit IgG (H+L) no. 711–545–152, Lot: 132876, Alexa Fluor647-AffiniPure Donkey Anti-Mouse IgG (H+L) no. 715–605–150, Lot: 131725, Alexa Fluor647-AffiniPure Donkey Anti-Rabbit IgG (H+L) no. 711–602–152, Lot: 136317 and all from Jackson Immuno Research). Tissue sections were stained with DAPI (1:5000 in PBS; Roche) for 15 min. Finally, the tissue sections were washed again twice in 1x TBS and mounted on glass microscope slides. Fluorescent images were taken by using a confocal microscope (Leica TCS SP5), and images were taken with a magnification of 40x or 63x. Images were processed with the Leica Application Suite Advanced Fluorescence 2.7.3.9723 according to the study by Schnell et al. (2014) .

Immunostained sections were quantified using ImageJ version 1.49. Background was subtracted and the detected signal was divided by the area of measurement. An average value obtained from three independent areas for every section was used. The signal coming from sections obtained from silenced mice was quantified as relative amount to the scramble, which was set to 1. Statistical analysis was performed on three animals per treatment.

Cell culture

NIH3T3 mouse fibroblast cells (ATCCRCRL-1658) were maintained in Dulbecco's modified Eagle's medium (DMEM), containing 10% fetal calf serum (FCS) and 100 U/mL penicillin-streptomycin at 37°C in a humidified atmosphere containing 5% CO2. Cdk5 KO cells were produced applying the CRISPR/Cas9 technique according to the manufacturer’s protocol of the company (Origene, SKU # KN303042).

The following plasmids used were previously described: pSCT-1, pSCT-1mPer2, pSCT-1 mPer-V5, pSCT1 ΔPasA mPer2 -V5, pSCT1 ΔPasB mPer2 -V5 ( Langmesser et al., 2008 ) ( Schmutz et al., 2010 ). pSCT-1 Cdk5-HA, pet-15b Cdk5-HIS, Gex-4T Per2 1–576, pGex-4T Per2 577–1256 were produced for this paper. The full-length cDNA (or partial fragments) encoding PER2 and the full-length Cdk5 were previously sub-cloned in the TOPO vector according to the manufacturer’s protocol (Catalog numbers pCR2.1-TOPO vector: K4500-01). They were subsequently transferred into the plasmid pSCT-1 using appropriate restriction sites. pGex-4T Per2 1–576 S394G, S394D, pSCT-1 mPer2 S394G were obtained using site-specific mutagenesis according to the manufacturer’s protocol on the requested codon carrying the interested amino acid of interest (Agilent Catalog # 200518). For accession numbers, vectors, mutations, and primers sources, see Supplementary file 2 .

Transfection and co-immunoprecipitation of overexpressed proteins

NIH 3T3 cells were transfected in 10 cm dishes at about 70% of their total confluency using linear polyethylenimine (LINPEI25; Polysciences Europe). The amounts of expression vectors were adjusted to yield comparable levels of expressed protein. Thirty hours after transfection, protein extracts were prepared. With regard to immunoprecipitation, each antibody mentioned in the paper was used in the conditions indicated by the respective manufacturer. The next day, samples were captured with 50 µL at 50% (w/v) of protein-A agarose beads (Roche) at 50% (w/v) and the reaction was kept at 4°C for 3 hr on a rotary shaker. Prior to use, beads were washed three times with the appropriate protein buffer and resuspended in the same buffer (50% w/v). The beads were collected by centrifugation and washed three times with NP-40 buffer (100 mM Tris-HCl pH7.5, 150 mM NaCl, 2 mM EDTA, 0.1% NP-40). After the final wash, beads were resuspendend in 2% SDS, 10% glycerol, 63 mM Trish-HCL pH 6.8 and proteins were eluted for 15 min at RT. Laemmli buffer was finally added, samples were boiled for 5 min at 95° C and finally loaded onto 10% SDS-PAGE gels ( Laemmli, 1970 ).

Total protein extraction from cells (Ripa method)

Medium was aspirated from cell plates, which were washed two times with 1x PBS (137 mM NaCl, 7.97 mM Na 2 HPO 4 × 12 H 2 O, 2.68 mM KCl, 1.47 mM KH 2 PO 4 ). Then PBS was added again and plates were kept 5 min at 37°C. NHI3T3 or HEK cells were detached and collected in tubes and washed two times with 1x PBS. After the last washing, pellets were frozen in liquid N 2 , resuspended in Ripa buffer (50 mM Tris-HCl pH7.4, 1% NP-40, 0.5% Na-deoxycholate, 0.1% SDS, 150 mM NaCl, 2 mM EDTA, 50 mM NaF) with freshly added protease or phosphatase inhibitors, and homogenized by using a pellet pestle. After that samples were centrifuged for 15 min at 16,100 g at 4°C. The supernatant was collected in new tubes and pellet discarded.

Total protein extraction from brain tissue

Total brain or isolated SCN tissue was frozen in liquid N 2 , and resuspended in lysis buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.25% SDS, 0.25% sodium deoxycholate, 1 mM EDTA) and homogenized by using a pellet pestle. Subsequently, samples were kept on ice for 30 min and vortexed five times for 30 s each time. The samples were centrifuged for 20 min at 12,000 rpm at 4°C. The supernatant was collected in new tubes and the pellet discarded.

Nuclear/cytoplasmic fractionation

Tissues or cells were resuspended in 100 mM Tris-HCl pH 8.8/10 mM DTT and homogenized with a disposable pellet pestle. After 10 min incubation on ice, the samples were centrifuged at 2500 g for 2 min at 4°C and the supernatant discarded. After adding 90 μL of completed cytoplasmic lysis buffer (10 mM EDTA, 1 mM EGTA, 10 mM Hepes pH 6.8, 0.2% Triton X-100, protease inhibitor cocktail (Roche), NaF, PMSF, ß-glycerophosphate), the pellet was resuspended by vortexing, followed by centrifugation at 5200 rpm for 2 min at 4°C. The supernatant transferred into a fresh 1.5 mL tube was the CYTOPLASMIC EXTRACT. The pellet was washed three times with cytoplasmic lysis buffer and resuspended in 45 μL 1x NDB (20% glycerol, 20 mM Hepes pH 7.6, 0.2 mM EDTA, 2 mM DTT) containing 2x proteinase and phosphatase inhibitors followed by adding 1 vol of 2x NUN (2 M Urea, 600 mM NaCl, 2% NP-40, 50 mM Hepes pH 7.6). After vortexing the samples were incubated 30 min on ice, centrifuged 30 min at 13,000 rpm at 4°C and the supernatant that was transferred into a fresh tube was the NUCLEAR EXTRACT.

Immunoprecipitation using brain tissue extracts

A protein amount corresponding to between 400 and 800 µg of total extract was diluted with the appropriate protein lysis buffer in a final volume of 250 µL and immunoprecipitated using the indicated antibody (ratio 1:50) and the reaction was kept at 4°C overnight on a rotary shaker. The day after, samples were captured with 50 µL of 50% (w/v) protein-A agarose beads (Roche) and the reaction was kept at 4°C for 3 hr on a rotary shaker. Prior to use, beads were washed three times with the appropriate protein buffer and resuspended in the same buffer (50% w/v). The beads were collected by centrifugation and washed three times with NP-40 buffer (100 mM Tris-HCl pH7.5, 150 mM NaCl, 2 mM EDTA, 0.1% NP-40). After the final wash, beads were resuspendend in 2% SDS 10%, glycerol, 63 mM Trish-HCL pH 6.8 and proteins were eluted for 15 min at RT. Laemmli buffer was finally added, samples were boiled 5 min at 95° C and loaded onto 10% SDS-PAGE gels.

Pull-down assay with GST-Per2 fragments

GST-fused recombinant Per2 proteins were expressed in the E. coli Rosetta strain [plasmids: GST-Per2 N-M (1-576), GST-Per2 M-C (577-1256)]. Proteins were induced with 1 mM IPTG (Sigma-Aldrich) for 3 hr at 30°C. Subsequently, proteins were extracted in an appropriate GST lysis buffer (50 mM Tris-Cl pH 7.5, 150 mM NaCl, 5% glycerol) adjusted to 0.1% Triton X-100 and purified to homogeneity with glutathione-agarose beads for 2 hr at 4°C. The beads were then incubated overnight at 4°C and washed with GST lysis buffer adjusted with 1 mM DTT. Subsequently, elution with 10 mM reduced glutathione took place for 15 min at room temperature. Elution was stopped by adding Laemmli buffer and samples were loaded onto the gel after 5 min at 95°C and WB was performed using anti-GST (Sigma no. 06–332) and anti-HA antibodies (Roche no. 11867423001) for immunoblotting.

CRISPR/Cas9 Cdk5 knock-out cell line

The CRISPR/Cas9 Cdk5 cell line was produced starting from NIH3T3 cells using a kit provided by Origene ( www.origene.com ). The knock-out cell line was produced according to the manufacturer’s protocol. Briefly, cells at 80% of confluency were co-transfected with a donor vector containing the homologous arms and functional cassette, and the guide vector containing the sequence that targets the Cdk5 gene. In parallel, a scrambled negative guide was also co-transfected with a donor vector. 48 hr after transfection the cells were split 1:10 and grown for 3 days. Cells were split another seven times (this time is necessary to eliminate the episomal form of donor vector, in order to have only integrated forms). Then, single colonies were produced and clones were analyzed by PCR in order to find those clones that did not express Cdk5 RNA. Positive clones were re-amplified.

PCR primers for genomic Cdk5:

FW: 5’- tgtgagtaccacctcctctgcaa -3’

RW: 5’- ttaaacaggccaggcccc -3’

Knockdown of Cdk5

About 24 hr after seeding cells, different shRNA Cdk5 plasmids (Origene TL515615 A/B/C/D Cdk5 shRNA) were transfected to knock down Cdk5 according to the manufacturer’s instructions. The knock-down efficiency was assessed at 48 hr post transduction by western blotting. Scrambled shRNA plasmid (Origene TR30021) was used as a negative control.

Cycloheximide treatment

NIH3T3 cells were treated with 100 µM cycloheximide 48 hr after transfection with the indicated vectors, and cells were harvested 0, 3, and 6 hr after treatment.

Proteasome inhibitor treatment

About 48 hr after transfection with either scrambled or shCdk5, cells where Cdk5 was silenced were treated for 12 hr with either DMSO (vehicle) or epoxomicin (Sigma-Aldrich) at a final concentration of 0.2 µM. Samples were collected, and proteins extracted followed by western blotting.

In vitro kinase assay

Recombinant GST-fused PER2 protein fragments were expressed and purified from the BL21 Rosetta strain of E. coli according to the manufacturer’s protocol described before, using glutathione-sepharose affinity chromatography (GE Healthcare). Each purified protein (1 µg) was incubated in the presence or absence of recombinant Cdk5/p35 (the purified recombinant N-terminal His6-tagged human Cdk5 and N-terminal GST-tagged human p25 were purchased from Millipore). Reactions were carried out in a reaction buffer (30 mM Hepes, pH 7.2, 10 mM MgCl2, and 1 mM DTT) containing [γ- 32 P] ATP (10 Ci) at room temperature for 1 hr and then terminated by adding SDS sample buffer and boiling for 10 min. Samples were subjected to SDS-PAGE, stained by Coomassie Brilliant Blue, and dried, and then phosphorylated proteins were detected by autoradiography.

In vitro kinase assay using immunoprecipitated Cdk5 from SCN

CDK5 was immunoprecipitated from SCN samples at different ZTs (circa 600 µg of protein extract) ( Figure 8 ). After immunoprecipitation at 4°C overnight with 2x Protein A agarose (Sigma-Aldrich), samples were diluted in washing buffer and split in two halves. One half of the IP was used for an in vitro kinase assay. Briefly, 1 µg of histone H1 (Sigma-Aldrich) was added to the immunoprecipitated CDK5 and assays were carried out in reaction buffer (30 mM Hepes, pH 7.2, 10 mM MgCl 2 , and 1 mM DTT) containing [γ- 32 P] ATP (10 Ci) at room temperature for 1 hr and then terminated by adding SDS sample buffer and boiling for 5 min. Samples were subjected to 15% SDS-PAGE, stained by Coomassie Brilliant Blue, and dried, and then phosphorylated histone H1 was detected by autoradiography. The other half of the IP was used for Western blotting to determine the total amount of CDK5 immunoprecipitated from the SCN samples. To quantify the kinase activity at each time point, the following formula was used: ([ 32 P] H1/total H1 for each reaction)/CDK5 IP protein.

Workflow of the in vitro kinase assay.

Workflow of the in vitro kinase assay performed using immunoprecipitated CDK5 from SCN protein extracts is schematized here. Seven mice were sacrificed, SCN tissues were isolated and pooled together every 4 hr starting from ZT 0 (lights on) until ZT20 (ZT12 lights off). Total protein was obtained from each pool of tissues, the quality of the extracts was checked by WB, and subsequently CDK5 was immunoprecipitated at each time point. Agarose beads detained the immunoprecipitation and one half of the precipitate was used for an in vitro kinase assay using as substrate commercial histone H1 as substrate. The other half was analyzed by WB in order to quantify the amount of protein immunoprecipitated, which was used for the kinase assay. Kinase activity around the clock was quantified using the following formula: ( 32 P-H1/total H1)/amount of immunoprecipitated CDK5.

Filter-aided in vitro kinase assay, phosphopeptide enrichment and mass spectrometry analyses

Filter-aided in vitro kinase assays and mass spectrometry analyses were performed essentially as described ( Hatakeyama et al., 2019 ). Briefly, recombinant Cdk5/p35 (Millipore) was incubated with the GST-fused PER2 protein fragment. On 10 kDa MW-cutoff filters (PALL) samples were incubated in kinase buffer containing 50 mM Hepes, pH 7.4, 150 mM NaCl, 0.625 mM DTT, Phostop tablets (Roche), 6.25 mM MgCl 2 , and 1.8 mM ATP at 30°C for 1 hr. Samples without ATP were used as negative control. Assays were quenched by 8 M urea and 1 mM DTT. Protein digestion for MS analysis was performed overnight ( Wiśniewski et al., 2009 ). Phosphopeptides were enriched by metal oxide affinity enrichment using titanium dioxide (GL Sciences Inc, Tokyo, Japan) ( Zarei et al., 2016 ).

LC-MS/MS measurements were performed on a QExactive Plus mass spectrometer coupled to an EasyLC 1000 nanoflow-HPLC. Peptides were separated on fused silica HPLC-column tip (I.D. 75 µm, New Objective, self-packed with ReproSil-Pur 120 C18-AQ, 1.9 µm [Dr. Maisch, Ammerbuch, Germany] to a length of 20 cm) using a gradient of A (0.1% formic acid in water) and B (0.1% formic acid in 80% acetonitrile in water): loading of sample with 0% B with a flow rate of 600 nL min-1; separation ramp from 5–30% B within 85 min with a flow rate of 250 nL min-1. NanoESI spray voltage was set to 2.3 kV and ion-transfer tube temperature to 250°C; no sheath and auxiliary gas was used. Mass spectrometers were operated in the data-dependent mode; after each MS scan (mass range m/z = 370–1750; resolution: 70,000) a maximum of 10 MS/MS scans were performed using a normalized collision energy of 25%, a target value of 1000 and a resolution of 17,500. The MS raw files were analyzed using MaxQuant Software version 1.4.1.2 ( Cox and Mann, 2008 ) for peak detection, quantification and peptide identification using a full-length Uniprot Mouse database (April, 2016) and common contaminants such as keratins and enzymes used for digestion as reference. Carbamidomethylcysteine was set as fixed modification and protein amino-terminal acetylation, serine-, threonine- and tyrosine-phosphorylation, and oxidation of methionine were set as variable modifications. The MS/MS tolerance was set to 20 ppm and three missed cleavages were allowed using trypsin/P as enzyme specificity. Peptide, site and protein FDR based on a forwards-reverse database were set to 0.01, minimum peptide length was set to 7, and minimum number of peptides for identification of proteins was set to one, which must be unique. The ‘match-between-run’ option was used with a time window of 1 min. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD012068 (project name: Cyclin dependent kinase 5 (CDK5) regulates the circadian clock; project accession: PXD012068).

Generation of an antibody against phospho-serine 394

We raised in mouse a specific monoclonal antibody recognizing the phosphorylated form of serine 394 of PER2 in collaboration with GenScript Company. The sequence used for the immunogen preparation was: FDY {pSer} PIRFRTRNGEC. 3 Balb/c mice and 3 C57 mice were immunized with conventional strategies and antisera obtained from those animals were used for the first control experiment performed by in vitro kinase assay ( Figure 5—figure supplement 3 ). The positive antiserum was used for the cell fusions. Subsequently, a screening with 16 96-well plates (from 2 × 10E4 clones) was performed by indirect ELISA, primary screening with phospho-peptide, then counter-screening with non-phospho-peptide. The obtained supernatants were tested by in vitro kinase assay in order to screen which one was better recognized the phospho-form of PER2 S394 ( Figure 5—figure supplement 4 ). Finally, five selected positive primary clones selected were subcloned by limiting dilution and tested as final antibody ( Figure 5—figure supplement 5 ).

Statistical analysis

Statistical analysis of all experiments was performed using GraphPad Prism6 software. Depending on the type of data, either an unpaired t-test, or one- or two-way ANOVA with Bonferroni or Tukey’s post-hoc test was performed. Values considered significantly different are highlighted. [p<0.05 (*), p<0.01 (**), or p<0.001 (***)].

Data supporting the findings of this work are available within the paper and its supplementary files, and on Dryad ( https://doi.org/10.5061/dryad.4067r78 ). Non-commercial biological materials are provided upon request to the corresponding author. Proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD012068. The Per2Brdm1 mutant mouse strain is available at the Jackson Laboratory Stock No: 003819 (B6.Cg-Per2 tm1Brd Tyr c-Brd). The floxed Per2 mice are available at the European Mouse Mutant Archive (EMMA) strain ID EM:10599, B6;129P2-Per2tm1Ual/Biat.

• Olejniczak I
• Ripperger J
• Langmesser S
• Virgilio CD
• Google Scholar
• Benavides DR
• Svenningsson P
• Greengard P
• Brinkkoetter PT
• Henderson S
• Hockenbery D
• Shankland SJ
• Ripperger JA
• Contreras-Vallejos E
• Gonzalez-Billault C
• Darlington TK
• Wager-Smith K
• Takahashi JS
• Debruyne JP
• DeBruyne JP
• Vielhaber EL
• Hortobágyi T
• Bernstein HG
• Goffinet AM
• Kulkarni AB
• Perreau-Lenz S
• Ahnert-Hilger G
• Hatakeyama R
• Péli-Gulli MP
• Jaquenoud M
• Garcia Osuna GM
• De Virgilio C
• Henne-Bruns D
• Knippschild U
• Rothenfluh A
• Shearman LP
• Hastings MH
• Mikoshiba K
• Finkielstein CV
• Shimomura K
• Zemenides PD
• Manoukian AS
• Yamaguchi-Iwai Y
• Semikhodskii AS
• Glossop NRJ
• Bechtold DA
• Boot-Handford RP
• Narasimamurthy R
• Longenecker G
• Peterson DW
• Dahlquist FW
• Roenneberg T
• Chakravarty S
• Vitaterna MH
• Carlezon WA
• Baeriswyl-Aebischer S
• Descombes P
• Sharifpoor S
• Baryshnikova A
• VanderSluis B
• Matsushita M
• Caviness VS
• van der Horst GT
• Muijtjens M
• Kobayashi K
• van Leenen D
• Hoeijmakers JH
• Vanselow JT
• Westermark PO
• Schlosser A
• Wiśniewski JR
• Gustafson CL
• Ramanathan C
• Yamaguchi S
• van Der Horst GT
• Cagampang FR
• Ozkan-Dagliyan I
• Gaddameedhi S
• Rackiewicz M

Author details

Contribution, competing interests.

Present address

For correspondence, fondazione cenci bolognetti, instituto pasteur, schweizerischer nationalfonds zur förderung der wissenschaftlichen forschung (31003a_166682), schweizerischer nationalfonds zur förderung der wissenschaftlichen forschung (310030_166474/1), schweizerischer nationalfonds zur förderung der wissenschaftlichen forschung (316030_177088).

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Stéphanie Aebischer, Antoinette Hayoz, Cressida Harvey, Naila Ben Fredi, Jean-Charles Paterna (Viral Vector Facility, University of Zürich) and the Bioimage platform (University of Fribourg) for technical support. Funding from the Swiss National Science Foundation (31003A_166682, 310030_166474/1 and 316030_177088) is acknowledged. AB was supported by a fellowship from the Fondazione Cenci Bolognetti, Instituto Pasteur.

Animal experimentation: This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the Swiss Legislation and the declaration of Helsinki. The protocols were approved by the state veterinarian of the State of Fribourg (Permit Number: 2015-33).

© 2019, Brenna et al.

This article is distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use and redistribution provided that the original author and source are credited.

• 3,414 views
• 508 downloads
• 29 citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.

Download links

Downloads (link to download the article as pdf).

• Article PDF
• Figures PDF

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools), categories and tags.

• Cryptochrome 1
• phosphorylation
• nuclear import

Research organisms

• S. cerevisiae

Further reading

Syntaxin-6 delays prion protein fibril formation and prolongs the presence of toxic aggregation intermediates.

Prions replicate via the autocatalytic conversion of cellular prion protein (PrP C ) into fibrillar assemblies of misfolded PrP. While this process has been extensively studied in vivo and in vitro, non-physiological reaction conditions of fibril formation in vitro have precluded the identification and mechanistic analysis of cellular proteins, which may alter PrP self-assembly and prion replication. Here, we have developed a fibril formation assay for recombinant murine and human PrP (23-231) under near-native conditions (NAA) to study the effect of cellular proteins, which may be risk factors or potential therapeutic targets in prion disease. Genetic screening suggests that variants that increase syntaxin-6 expression in the brain (gene: STX6) are risk factors for sporadic Creutzfeldt–Jakob disease. Analysis of the protein in NAA revealed, counterintuitively, that syntaxin-6 is a potent inhibitor of PrP fibril formation. It significantly delayed the lag phase of fibril formation at highly sub-stoichiometric molar ratios. However, when assessing toxicity of different aggregation time points to primary neurons, syntaxin-6 prolonged the presence of neurotoxic PrP species. Electron microscopy and super-resolution fluorescence microscopy revealed that, instead of highly ordered fibrils, in the presence of syntaxin-6 PrP formed less-ordered aggregates containing syntaxin-6. These data strongly suggest that the protein can directly alter the initial phase of PrP self-assembly and, uniquely, can act as an ‘anti-chaperone’, which promotes toxic aggregation intermediates by inhibiting fibril formation.

Quantitative mapping of proteasome interactomes and substrates using ProteasomeID

Proteasomes are essential molecular machines responsible for the degradation of proteins in eukaryotic cells. Altered proteasome activity has been linked to neurodegeneration, auto-immune disorders and cancer. Despite the relevance for human disease and drug development, no method currently exists to monitor proteasome composition and interactions in vivo in animal models. To fill this gap, we developed a strategy based on tagging of proteasomes with promiscuous biotin ligases and generated a new mouse model enabling the quantification of proteasome interactions by mass spectrometry. We show that biotin ligases can be incorporated in fully assembled proteasomes without negative impact on their activity. We demonstrate the utility of our method by identifying novel proteasome-interacting proteins, charting interactomes across mouse organs, and showing that proximity-labeling enables the identification of both endogenous and small-molecule-induced proteasome substrates.

A 2-hydroxybutyrate-mediated feedback loop regulates muscular fatigue

Several metabolites have been shown to have independent and at times unexpected biological effects outside of their metabolic pathways. These include succinate, lactate, fumarate, and 2-hydroxyglutarate. 2-Hydroxybutyrate (2HB) is a byproduct of endogenous cysteine synthesis, produced during periods of cellular stress. 2HB rises acutely after exercise; it also rises during infection and is also chronically increased in a number of metabolic disorders. We show here that 2HB inhibits branched-chain aminotransferase enzymes, which in turn triggers a SIRT4-dependent shift in the compartmental abundance of protein ADP-ribosylation. The 2HB-induced decrease in nuclear protein ADP-ribosylation leads to a C/EBPβ-mediated transcriptional response in the branched-chain amino acid degradation pathway. This response to 2HB exposure leads to an improved oxidative capacity in vitro. We found that repeated injection with 2HB can replicate the improvement to oxidative capacity that occurs following exercise training. Together, we show that 2-HB regulates fundamental aspects of skeletal muscle metabolism.

Be the first to read new articles from eLife

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• v.8(10); 2017 Mar 7

Biological functions of CDK5 and potential CDK5 targeted clinical treatments

Alison shupp.

1 Departments of Cancer Biology, Medical Oncology, Sidney Kimmel Cancer Center, Thomas Jefferson University, Philadelphia, PA, USA

Mathew C. Casimiro

2 Pennsylvania Cancer and Regenerative Medicine Research Center, Baruch S Blumberg Institute, Doylestown, PA, USA

Richard G. Pestell

Cyclin dependent kinases are proline-directed serine/threonine protein kinases that are traditionally activated upon association with a regulatory subunit. For most CDKs, activation by a cyclin occurs through association and phosphorylation of the CDK’s T-loop. CDK5 is unusual because it is not typically activated upon binding with a cyclin and does not require T-loop phosphorylation for activation, even though it has high amino acid sequence homology with other CDKs. While it was previously thought that CDK5 only interacted with p35 or p39 and their cleaved counterparts, Recent evidence suggests that CDK5 can interact with certain cylins, amongst other proteins, which modulate CDK5 activity levels. This review discusses recent findings of molecular interactions that regulate CDK5 activity and CDK5 associated pathways that are implicated in various diseases. Also covered herein is the growing body of evidence for CDK5 in contributing to the onset and progression of tumorigenesis.

INTRODUCTION

Cyclin dependent kinases are proline-directed serine/threonine protein kinases that are traditionally activated upon association with a regulatory subunit. CDKs are a part of a kinase family that has been conserved throughout evolution and can be found in species from Saccharomyces cerevisia to humans. In humans there are 13 different CDKs (CDK1 - CDK13) that are highly expressed in mitotic cells [ 1 ]. For most CDKs, activation by a cyclin occurs through association and phosphorylation of the CDK's T-loop. Despite having high amino acid sequence homology with other CDKs, CDK5 is unusual because it is not typically activated upon binding with a cyclin and does not require T-loop phosphorylation for activation. Additionally, CDK5 has functions in both terminally differentiated and proliferating cells [ 2 ]. CDK5 was first identified in 1992 by multiple groups and was given a different name by each, including tau kinase II [ 3 ], neuronal Cdc2 like kinase [ 4 ], brain proline-directed kinase [ 5 ], PSSALRE [ 6 ], and CDK5 [ 7 ]. An isoform of CDK5, termed either CDK5-SV or CDK5-V1, was recently discovered [ 8 , 9 ]. One study reported that this splice variant lacks 32 amino acids encoded by exon 7 [ 8 ], while another study stated the missing 32 amino acids are encoded by exon 6 [ 9 ]. Although these two groups reported conflicting data, it has been suggested that the identified isoforms are in fact the same protein and the variances in their data are due to different methodologies [ 10 ].

CDK5 can be mapped to chromosome 7q36 and its expression is upregulated by the transcription factors Fos and CREB through the MEK/ERK pathway and by δFosB [ 11 , 12 ]. CDK5 plays a vital role in the central nervous system but has functions in other cell types. Outside of the nervous system, active CDK5 has been found in pancreatic β cells [ 13 ], corneal epithelial cells [ 14 ] and monocytes [ 15 ] amongst various other cell types [ 10 , 16 ]. In the nervous system, CDK5 is involved in neuron migration, neurite outgrowth and support, and synaptogenesis. CDK5's function in cells other than neurons includes the induction of cell motility, apoptosis, and cell cycle progression as well as functions involved with the immune system, lymphatic system, vascularization, and insulin secretion. A summary of CDK5 functions as discussed herein can be found in Table ​ TableI. I . CDK5 has recently been implicated in diseases, including the development and progression of cancer and neurodegenerative diseases. For this reason, the regulation of CDK5 activity is now emerging as a candidate therapeutic target.

CDK5 ACTIVATORS AND REGULATORS

Unlike other CDKs, CDK5 is not primarily activated by cyclins. Instead it is through specific binding with the proteins p35 or p39, or their respective cleaved counterparts p25 and p29, that CDK5 becomes active [ 1 , 17 , 18 ]. It was found that p35 knockout mice have defective cortical lamination and adults suffered from sporadic lethality and seizures [ 19 ], which is a less severe phenotype than that exhibited by Cdk5 knockout mice [ 20 ]. p39 -/- mice did not display any obvious abnormalities, however p35/p39 compound knockout mice displayed a phenotype identical to that of the Cdk5 -/- mice [ 21 ], suggesting that while p39 may not play a pivotal role in Cdk5 activation, it becomes necessary for nervous system development in the absence of p35.

p35 has a myristolation sequence that localizes it to phospholipid membranes [ 22 ]. Active CDK5 can phosphorylate p35 at Ser8 and Thr138. In the brain, phosphorylation of S8 is constant throughout development, but phosphorylation of T138 is found more abundantly in fetal brain tissue [ 23 ]. The phosphorylation at S8 leads to a more diffuse localization throughout the cytoplasm. This could be due to increased p35 mobility on membranes due to an altered interaction between the protein and phospholipids that constitute cell membranes [ 24 ]. p35 phosphorylation at T138 prevents its cleavage to p25 by calpain [ 23 ]. Because CDK5 has various regulatory functions in neuron development and migration, it is likely that the phosphorylation of p35 at T138 protects against aberrant CDK5 activation through formation of p25 in the fetal stage of brain development when CDK5 activity is also high [ 1 ]. Additionally, in vitro , under conditions of oxidative stress, p35 has been found conjugated to SUMO2 at Lys246 and Lys290, which led to increased p35/CDK5 activity [ 25 ].

As previously mentioned, the CDK5 activator p25 is formed through cleavage of p35 by calpain. This produces both the p25 product as well as a p10 product. Cleavage of p35 occurs under stress conditions such as amyloidβ presence, excitotoxicity, or oxidative stress [ 22 , 26 ]. This cleavage allows p25 to localize to nuclear and perinuclear regions by removing the p10 myristolation sequence [ 22 ]. Compared with p35, p25 has a longer half-life, and therefore prolongs the activation period of CDK5, leading to increased phosphorylation of CDK5's target proteins [ 22 , 27 ].

The functions of CDK5 activators p39 and p29 largely overlap with those of p35 and p25, respectively, however their expression throughout brain regions vary. p39 and p29 are mainly expressed in postnatal cerebral cortex and the hindbrain while p35 and p25 are largely expressed in the cerebral cortex of developing brains [ 27 ]. The localization of p39 to membranes is similar to that of p35 due to its conserved myristolation sequence [ 22 ]. Likewise, p39 also shows a more diffuse localization upon phosphorylation of Ser8 by CDK5 [ 24 ]. p39 can be phosphorylated by CDK5 at Ser173, a site equivalent to T138 in p35, and Thr84, however the effect of these phosphorylations on controlling protein stability have not yet been explored [ 1 , 24 ].

In addition to p35 and p39, cyclin I has also been shown to activate CDK5. Cyclin I-CDK5 binding targets CDK5 to the nucleus [ 28 ] and increases levels of anti-apoptotic proteins Bcl2 and Bcl2l1 via the MEK/ERK pathway [ 29 ] . This upregulation of Bcl2 and Bcl211 is observed only through cyclin I activation of CDK5, not activation via p35 [ 29 , 30 ]. CDK5 has been found to bind cyclin D1 and cyclin D3 in human fibroblasts, however this interaction had no influence on the activation and kinase activity of CDK5 [ 7 , 31 ].

While CDK5 is only activated by p35/p25, p39/p29, or cyclin I, the activity of CDK5 can be modulated by a variety of other proteins, as depicted in Figure ​ Figure1. 1 . For instance, cyclin D1 can attenuate CDK5 kinase activity by competing with p35 for binding with CDK5, thereby forming an inactive complex of cyclin D1 and CDK5 (Fig. ​ (Fig.1). 1 ). CDK5 and cyclin D1 can be found in the rat cerebellum during the first 24 days of postnatal development, albeit at varying abundances. CDK5 abundance increased while cyclin D1 decreased from day 9 on to adulthood [ 32 ]. In post-mitotic neurons, cyclin D1/CDK5 association was found to lead to cell cycle related neuronal apoptosis through sustained MEK/ERK signaling [ 33 ].

Cyclin E can directly interact with Cdk5 to reduce its activity. Cyclin E was found to sequester mouse Cdk5 away from other protein activators along with p27 KIP1 . The formation of this complex, and consequent attenuation of Cdk5 activity was found to promote synaptic plasticity, memory formation, and dendritic growth, as cyclin E -/- mice, that had increased Cdk5 activity, were deficient in these processes [ 34 ]. While this result may seem counterintuitive due to active CDK5's function with supporting neurite outgrowths, this observation could be explained by an overabundance of active CDK5 detrimentally effecting neurite outgrowth and subsequently synaptic plasticity. This theory would be consistent with findings that CDK5 expression levels are increased in certain neurodegenerative diseases, and that it is the aberrant CDK5 activity that leads to neurite collapse and death [ 35 – 37 ].

Glutathione-S-transferase (GSTP1) is another regulator of CDK5 activity that functions by competing with p35 for CDK5 binding. GSTP1 also reduces aberrant CDK5 activity by scavenging for molecules associated with oxidative stress and thereby decreasing the likelihood of p35/p39 cleavage to p25/p29 [ 38 ] (Figure ​ (Figure1 1 ).

TP53 induced glycolysis regulatory phosphatase (TIGAR) has been shown to upregulate CDK5 expression levels in the presence of induced DNA damage (Figure ​ (Figure1). 1 ). Knockdown of TIGAR led to decreased CDK5 expression, decreased phosphorylated ATM, and consequently increased levels of induced DNA damage. This suggests that DNA damage repair is mediated via TIGAR activation of the CDK5-ATM pathway [ 39 ].

CDK5 IN CELL CYCLE AND OTHER PATHWAYS

Previously, CDK5 was thought to function in a cell cycle independent manner; however, recently the retinoblastoma protein (Rb) was discovered as a downstream target of CDK5. Expression of CDK5 leads to the phosphorylation of Rb, ultimately leading to the expression of cyclins and other cdks [ 40 ]. The protein kinase CK1 is phosphorylated by CDK5, and is involved in a wide array of signaling pathways including cell cycle, DNA repair, and apoptosis [ 41 ]. When CDK5 phosphorylates CK1, its kinase activity is subsequently reduced [ 42 ]. The functional affect of CDK5-mediated phosphorylation of CK1 on cell cycle, DNA repair, or apoptosis has yet to be explored.

In pancreatic β cells, CDK5 activity reduces insulin secretion in response to glucose abundance (Figure ​ (Figure2). 2 ). This was demonstrated using CDK5 inhibitors, as well as inhibition of CDK5's activator p35. When CDK5 is active, it phosphorylates the L-type voltage-dependent Ca +2 channel (L-VDCC) at Ser783, which prevents the association of L-VDCC with syntaxin and SNAP-25, thereby preventing exocytosis of insulin from the cell [ 13 ].

Within the immune system, CDK5 has been implicated in IFNγ-induced programmed death ligand 1 (PD-L1) upregulation, which allows certain cells to evade detection by the immune system. Decreased CDK5 expression led to increased expression of the PD-L1 transcriptional repressors IRF2 and IRF2BP and consequent decreased PD-L1 expression (Figure ​ (Figure2) 2 ) [ 43 ]. PD-L1 is a ligand that binds with PD-1, which is found on various immune cells. The binding of PD-L1 and PD-1 decreases an immune response by inhibiting T-cell activation and cytokine production. In normal tissues this is vital for maintaining homeostasis [ 44 ]. However tumor cells can also express PD-L1, which allows them to avoid detection and elimination by T-cells [ 45 , 46 ].

CDK5 promotes the formation of lymphatic vessels. CDK5 phosphorylates Foxc2, a protein that regulates the expression of connexin 37, which is critical for lymphatic valve formation (Figure ​ (Figure2). 2 ). Moreover, knockout of CDK5 in the endothelium leads to lymphedema formation and embryonic lethality in mice [ 47 ].

CDK5 has previously been implicated in the migration of neurons. CDK5 knockout mice have abnormal cortical lamination, and more than 60% of CDK5 -/- mice died in utero [ 20 ]. Various studies have since implicated CDK5 in cell migration as it governs cancer metastasis. In prostate cancer cells, inhibition of CDK5 by the drug roscovitine prevented cell migration. The roscovitine treated cells did not project lamellopodia, and had reduced tubulin structures compared to untreated cells. This suggests that CDK5 inhibition prevented the establishment of cell polarity required for movement [ 48 ]. Additionally, knockdown of CDK5 in melanoma cell lines decreased cell motility and cell spreading in vitro , and decreased formation of lung and liver metastases in vivo in a mouse model of human melanoma. The decrease in CDK5 expression led to decreased phosphorylation of caldesmon, which decreased its binding affinity with actin and calmodulin (Figure ​ (Figure2) 2 ) [ 49 ]. Another mechanism by which CDK5 may promote cell migration is by enhancing pro-migratory P13K/AKT signaling. CDK5 phosphorylates the Gα –interacting vesicle associated protein (GIV), which promotes GIV interaction with Gαi, thereby enhancing P13K/AKT signaling (Figure ​ (Figure2) 2 ) [ 50 ]. Together, these studies demonstrate the importance of CDK5 in cell motility, a naturally occurring and necessary process. However, CDK5 mediated movement could also be an underlying driver of cancer metastasis and could be targeted in treatments to halt cancer metastasis.

CYTOSKELETAL ORGANIZATION

An important function of CDK5, especially in neurons, is the organization of the cytoskeleton and support of cellular outgrowths (Figure ​ (Figure2). 2 ). Expression of p35 or p39 in vitro stimulates neurite outgrowths, and a dominant negative mutant of CDK5 was found to abolish the formation of these outgrowths [ 51 ]. CDK5 supports axon and neurite outgrowth is through phosphorylation of the neurofilament heavy chain, resulting in the assembly of neurofilaments [ 52 ].

CDK5 has been shown to both prevent and promote growth cone collapse under different circumstances. CDK5 phosphorylates the protein CRMP2A at Ser27, which can be stabilized by Pin1 to support the growth of growth cones in the presence of semaphorin3A stimulation [ 53 , 54 ]. Additionally, CDK5 can promote axonal growth through indirect activation of CRMP2 by phosphorylating the protein Axin. Phosphorylated Axin inhibits GSK3β activity, leading to an increase in active, unphosphorylated CRMP2 [ 55 ] (Figure ​ (Figure2 2 ).

Conversely, CDK5 promotes the collapse of growth cones through association with CRMP2 and α2-chimerin, an adaptor protein between CRMP2 and CDK5-p35. This association of CRMP2, α2-chimerin, and CDK5-p35 promotes the phosphorylation of CRMP2 at Ser522 by CDK5. In turn, this allows for CRMP2 to associate with and be phosphorylated at T514 by GSK3β, resulting in CRMP2 inactivation, microtubule disassembly, and ultimately growth cone collapse [ 56 ]. In this manner, CDK5 activity can both prevent and promote collapse of growth cones.

CDK5 can also reduce cellular outgrowth by regulating cytoskeletal organization through phosphorylation of p35 at T138, which prevents the polymerization of microtubules. This phosphorylation at T138 is found primarily in fetal brain tissues as opposed to adult brain [ 23 ].

ROLE OF CDK5 IN NEUROLOGICAL DISEASE

Due to the many roles of CDK5 in the development of the nervous system, as well as the effects of cellular stress on CDK5 activation, CDK5 has been implicated in the progression of various neurological diseases and as a potential therapeutic target in disease treatment. For instance, while CDK5 normally phosphorylates collapsin response mediator protein 2 (CRMP2) to stimulate axon growth, it was found that hyperphosphorylation of CRMP2, as well as Tau, were implicated in the generation of neurofibrillary tangles characteristic of Alzheimer's disease [ 53 ]. Cell stress, including the presence of amyloid beta, is known to aberrantly activate CDK5 due to the formation of p25, which has been shown to cause the hyperphosphorylation of Tau, leading to atypical cell cycling, synaptotoxicity, and neuronal apoptosis [ 57 ]. Additionally, increased CDK5 activity caused by the sumoylation of p35 under oxidative stress, also contributes to neurodegeneration [ 25 ].

While CDK5 overexpression and aberrant activation are associated with neurodegenerative diseases, a loss or reduction in CDK5 activity is implicated in certain intellectual disabilities and neurodevelopmental disorders. Decreased CDK5 activity has been associated with intellectual disability in NF1 microdeletion syndrome patients [ 58 ] and schizophrenia [ 59 ]. Additionally, transgenic mice with decreased Cdk5 activity exhibited spontaneous seizures [ 60 ] as well as behaviors similar to ADHD [ 61 ].

CDK5 EXPRESSION IN CANCER

Elevated levels of CDK5 have been found in various mouse tumors and human malignant tumors [ 40 ] [ 53 , 62 – 65 ]. The mechanisms involve effects on angiogenesis, cell proliferation and the immune system. As noted above, CDK5 enhances pRb phosphorylation and thereby cell-cycle progression [ 40 ]. Furthermore, CK1 is phosphorylated by CDK5, which in turn governs cell cycle, DNA repair, and apoptosis [ 41 ]. Increased levels of CDK5 target proteins are being considered as possible biomarkers of specific cancers. For example, an increase in CRMP2 phosphorylation could be a potential biomarker for certain lung cancers, as phosphorylated CRMP2 was found in the nuclei of biopsied lung cancer cells, but not cells in the surrounding epithelium [ 53 ].

In a transgenic mouse model of sporadic medullary thyroid carcinoma (MTC), p25 overexpression led to the development of bilateral malignant thyroid tumors, and was fatal after 30 weeks. However, arresting p25 expression at 5, 11, or 16 weeks led to 100 percent survival in all mice analyzed after 30 weeks. Similar results were discovered in vitro , in which reducing p25 expression or knocking down Cdk5 expression prevented further cell proliferation. This suggests that it is the aberrant activation of Cdk5 by p25 that leads to the progression of sporadic MTC [ 40 ].

CDK5 expression in medulloblastoma allows tumor cells to evade detection by T-cells in vivo . Conversely, decreased CDK5 expression enhanced the recruitment of CD4 + T-cells to the tumor site in mice, and increased the tumor-free survival rate of the mice. CDK5 regulates the evasion of tumors from the immune system by decreasing expression of transcriptional repressors of PD-L1 expression, thus increasing the abundance of PD-L1 [ 45 ].

Inhibiting CDK5 activity in hepatocellular carcinoma (HCC) cells prevented angiogenesis in vivo by decreasing the abundance of HIF-1α. Because HCC is a highly vascularized tumor type, inhibiting CDK5 and therefore angiogenesis, could prove a promising treatment for this tumor subtype and other highly vascularized tumors [ 64 ].

CDK5 AS A TARGET FOR DISEASE TREATMENT

Due to the biological and clinically relevant importance of CDK5's function in multiple cell types, CDK5 presents an attractive therapeutic target for treating a variety of conditions such as diabetes, cancer, and neurodegeneration. Additionally, the upregulation of CDK5 associated with various cancers and neurodegenerative diseases further implicates its role in the development and progression of disease. Recently, tamoxifen (TMX), a drug currently used in breast cancer treatment, was found to decrease CDK5 activation by competitively binding with p35 and p25, and preventing their activation of CDK5. While the TMX inhibition of CDK5 activity could contribute to the anti-tumor effects of the drug, TMX treatment was also found to decrease Tau phosphorylation, suggesting a use for tamoxifen in treating Alzheimer's disease [ 66 ]. However, because of the broad functions of CDK5 in different cell and tissue types and the pan CDK inhibitory effect on other family members, the off target affects of a CDK5 inhibitory drug may create undesirable side effects. Nonetheless, CDK inhibitors are an intriguing clinical therapy for the treatment of various cancers. A list of current cyclin-dependent kinase inhibitors, including inhibitors of CDK5, and their associated clinical trials for the treatment of cancer can be seen in Table ​ TableII II .

This is representative rather than a comprehensive list of past and present clinical trials in the field.

One of the most well studied CDK inhibitors being used in cancer clinical trials is flavopiridol, a drug developed by Tolero pharmaceuticals under the name Alvocidib. Flavopiridol was found to competivively bind to the ATP-binding pocket of CDK1, CDK2, CDK4, and CDK9, consequently inducing apoptosis in both dividing and quiescent cells. Early clinical trials with flavopiridol as a monotherapy proved ineffective in that there was a narrow window between no clinical response and severe, lethal tumor lysis. Ongoing trials involve combination therapies with other novel chemotherapy agents to overcome the limitations of flavorpiridol [ 67 ].

Another relatively well studied CDK inhibitor, Dinaciclib, was found to be more efficacious than flavopiridol, with IC 50 values in the low nanomolar range (1-4 nM – in various models flavopiridol's IC50 values range from 50-350 nM) [ 67 , 68 ]. Dinaciclib selectively inhibits CDK1, CDK2, CDK5, and CDK9 [ 67 ]. Preclinical studies and early clinical trials demonstrated the cytotoxicity of Dinaciclib in solid tumors and chronic lymphocytic leukemia, while not affecting T-cell function or number [ 69 ].

Roscovitine, marketed under the name Seliciclib, is an inhibitor of CDK5 and CDK2. Many of the clinical trials for Seliciclib were intiated determine dose-limiting toxicities of the drug alone or in combination with other chemotherapeutics. While roscovitine is used widely experimentally to inhibit CDK5 activity, it is not being intensively examined as a clinical cancer therapeutic [ 67 ].

To potentially reduce broad undesirable off target effects of pan-CDK inhibitors, CDK5 inhibitory peptide (CIP) has been studied as a potential therapeutic for neurodegeneration. CIP specifically targets the hyperactivated state of CDK5 as mediated by p25/p29, while allowing normal activation of CDK5 by p35/p39. CDK5 inhibitory peptide (CIP) was found to inhibit the hyperactivation of CDK5 by p25 overexpression in vivo , which reduced neurodegeneration and improved cognitive function of transgenic mice, without affecting neurodevelopment [ 70 ]. In the future, CIP could possibly be adapted to treat certain cancers caused by aberrant CDK5 activation.

CONFLICTS OF INTEREST

There is no conflict of interest.

Loading metrics

Open Access

Peer-reviewed

Research Article

Phosphorylation of the DNA damage repair factor 53BP1 by ATM kinase controls neurodevelopmental programs in cortical brain organoids

Contributed equally to this work with: Bitna Lim, Yurika Matsui

Roles Data curation, Formal analysis, Investigation, Visualization

Current address: CHA Future Medicine Research Institute, Seongnam, Republic of Korea

Affiliation Department of Developmental Neurobiology, St Jude Children’s Research Hospital, Memphis, Tennessee, United States of America

Roles Data curation, Formal analysis, Methodology, Visualization, Writing – review & editing

Roles Formal analysis

Affiliation Center for Applied Bioinformatics, St Jude Children’s Research Hospital, Memphis, Tennessee, United States of America

Affiliation Center for Proteomics and Metabolomics, St. Jude Children’s Research Hospital, Memphis, Tennessee, United States of America

Roles Data curation, Formal analysis

Roles Supervision

Affiliation Department of Biostatistics, St Jude Children’s Research Hospital, Memphis, Tennessee, United States of America

Roles Methodology

Affiliation Department of Cell & Molecular Biology, St Jude Children’s Research Hospital, Memphis, Tennessee, United States of America

Roles Data curation

Affiliations Department of Developmental Neurobiology, St Jude Children’s Research Hospital, Memphis, Tennessee, United States of America, Department of Structural Biology, St Jude Children’s Research Hospital, Memphis, Tennessee, United States of America

Roles Formal analysis, Supervision

•  [ ... ],

Roles Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Visualization, Writing – original draft, Writing – review & editing

* E-mail: [email protected]

• [ view all ]
• [ view less ]
• Bitna Lim, 
• Yurika Matsui, 
• Seunghyun Jung, 
• Mohamed Nadhir Djekidel, 
• Wenjie Qi, 
• Zuo-Fei Yuan, 
• Xusheng Wang, 
• Xiaoyang Yang, 
• Nina Connolly, 

• Published: September 3, 2024
• https://doi.org/10.1371/journal.pbio.3002760
• Peer Review
• Reader Comments

This is an uncorrected proof.

53BP1 is a well-established DNA damage repair factor that has recently emerged to critically regulate gene expression for tumor suppression and neural development. However, its precise function and regulatory mechanisms remain unclear. Here, we showed that phosphorylation of 53BP1 at serine 25 by ATM is required for neural progenitor cell proliferation and neuronal differentiation in cortical brain organoids. Dynamic phosphorylation of 53BP1-serine 25 controls 53BP1 target genes governing neuronal differentiation and function, cellular response to stress, and apoptosis. Mechanistically, ATM and RNF168 govern 53BP1’s binding to gene loci to directly affect gene regulation, especially at genes for neuronal differentiation and maturation. 53BP1 serine 25 phosphorylation effectively impedes its binding to bivalent or H3K27me3-occupied promoters, especially at genes regulating H3K4 methylation, neuronal functions, and cell proliferation. Beyond 53BP1, ATM-dependent phosphorylation displays wide-ranging effects, regulating factors in neuronal differentiation, cytoskeleton, p53 regulation, as well as key signaling pathways such as ATM, BDNF, and WNT during cortical organoid differentiation. Together, our data suggest that the interplay between 53BP1 and ATM orchestrates essential genetic programs for cell morphogenesis, tissue organization, and developmental pathways crucial for human cortical development.

Citation: Lim B, Matsui Y, Jung S, Djekidel MN, Qi W, Yuan Z-F, et al. (2024) Phosphorylation of the DNA damage repair factor 53BP1 by ATM kinase controls neurodevelopmental programs in cortical brain organoids. PLoS Biol 22(9): e3002760. https://doi.org/10.1371/journal.pbio.3002760

Received: August 25, 2023; Accepted: July 19, 2024; Published: September 3, 2024

Copyright: © 2024 Lim et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All sequencing data are deposited in NCBI GEO database under accession number GSE231321. Codes for analyzing sequencing data are deposited in https://doi.org/10.6084/m9.figshare.7411835 . Mass spectrometry data were deposited in ProteomXchange, with project accession number PXD041699. Numerical data are in S1 Data , and uncropped Western blot images are in S1 Raw Images .

Funding: This work was supported by the American Lebanese Syrian Associated Charities ( https://www.stjude.org/ to JCP), American Cancer Society ( https://www.cancer.org/ ; 132096-RSG-18-032-01-DDC to JCP), and NIH ( https://www.nih.gov/ ; 1R01GM134358-05 to JCP). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Abbreviations: ATM, ataxia telangiectasia mutated; BSA, bovine serum albumin; FDR, false discovery rate; GO, Gene Ontology; GSEA, gene set enrichment analysis; hESC, human embryonic stem cell; KO, knockout; NPC, neural progenitor cell; PSM, peptide spectral match; TMT LC-MS/MS, liquid chromatography-tandem mass spectrometry; WB, western blot; WT, wild type; 53BP1, p53 binding protein 1; 53BP1-pS25, 53BP1 phosphorylated at serine 25

Introduction

Transcription ensures the proper expression of genetic information for the development and function of the organism, whereas DNA repair maintains the integrity of the genetic code. These 2 processes share cross-functional factors, including CSB, TFII, and XPG, which repair DNA damage caused by torsional stress from transcription-initiating RNA polymerase II [ 1 – 3 ]. Conversely, some proteins initially believed to function exclusively in DNA repair have been found to regulate gene expression. For example, 53BP1 (p53 binding protein 1) is a key regulator of DNA repair mechanisms, promoting nonhomologous end-joining over homologous recombination [ 4 ]. During the DNA damage response, 53BP1 plays a pivotal role in p53-mediated activation of tumor suppressive genetic programs [ 5 ]. Recent research has also revealed that 53BP1 collaborates with the chromatin modifier UTX in neural progenitor cells (NPCs), promoting an open chromatin to facilitate the activation of neurogenic or corticogenic programs [ 6 ]. Intriguingly, the 53BP1–UTX interaction is observed in humans but not in mice [ 6 ]; the mechanism is not well conserved and regulates primate neurodevelopment. These discoveries highlight the importance of 53BP1 in gene regulation for tumor suppression and neural development. However, the precise mechanisms underlying 53BP1’s role in gene regulation and its upstream mechanism are yet to be fully understood.

Studies of 53BP1 have primarily focused on its role in the DNA damage response. To localize to chromatin with double-stranded breaks, 53BP1 uses its BRCT domain to bind to γH2AX, the Tudor domain to bind to H4K20 dimethylation, and its UDR segment to bind to ubiquitinated H2AK15 [ 7 – 10 ]. Additionally, the phosphorylated SQ/TQ motif of 53BP1 coordinates the docking of RIF1 or SCAI, selectively promoting nonhomologous end-joining or reducing homologous recombination [ 11 , 12 ]. These interactions are likely relevant to the gene regulatory activities of 53BP1. For example, γH2AX recruits 53BP1 and is required for resolving R-loops, DNA demethylation, transcription activation, and transcription elongation [ 13 , 14 ]. These findings suggest that the activities of 53BP1 in DNA damage response are interconnected with its gene regulatory functions.

The studies mentioned above have contributed to a model of 53BP1, where posttranslational modifications of its different residues and domains coordinate various activities. Most prominently, numerous residues of 53BP1 are phosphorylated by ATM (ataxia telangiectasia mutated) kinase [ 10 , 15 – 17 ]. ATM-mediated phosphorylation of 53BP1 or 53BP1-interacting proteins controls protein interactions, cellular localization, and DNA repair mechanisms [ 11 , 12 , 18 – 20 ]. Despite these discoveries, the impact of phosphorylation on the gene regulatory activity of 53BP1 remains unknown. Here, we report that phosphorylation of 53BP1-serine 25 by ATM is crucial for the proper expression of genetic programs during the growth and development of cortical brain organoids. ATM-dependent phosphorylation controls the chromatin binding of 53BP1 to genomic targets functioning in several key pathways, including neuronal differentiation, cytoskeleton, p53, and ATM, BNDF, and WNT signaling pathways. These results highlight the essential role of 53BP1 phosphorylation in regulating genetic programs for the differentiation of cortical brain organoids.

Phosphorylated 53BP1-S25 increases during differentiation of hESCs into NPCs

Although 53BP1 is required for human embryonic stem cells (hESCs) to differentiate into NPCs [ 6 ], its levels did not change during differentiation ( Fig 1A ). Therefore, its regulation is likely posttranslational during neural differentiation. Human NPCs were analyzed by RNA-seq and immunofluorescence to validate successful NPC generation ( S1A-S1E Fig ). 53BP1 is targeted by various kinases, including ATM, and we hypothesized that 53BP1 phosphorylation regulates the differentiation of hESCs into NPCs. Intriguingly, we found that the levels of 53BP1 phosphorylated at serine 25 (53BP1-pS25) were markedly increased in NPCs compared to hESCs (Figs 1A and S1F-S1H ). The levels of the DNA damage marker γH2AX were similar in NPCs and hESCs ( S1I Fig ), suggesting that the increase in 53BP1-pS25 levels during NPC differentiation is not due to increased DNA damage.

• PPT PowerPoint slide
• PNG larger image
• TIFF original image

( A ) WB of the nuclear extract of hESCs and hNPCs showed marked increase of 53BP1-pS25 in hNPCs. WB analysis of IgG, ( B ) 53BP1, and ( C ) ATM co-immunoprecipitation in the nuclear extract of hESCs. ( D ) Quantification of the relative ATM protein levels (normalized to β-ACTIN) in 5 replicate WB analyses of hESCs and hNPCs. ( E ) Schematic diagram of the cortical organoid differentiation. Aggregates were formed in the induction media for 17 days, embedded in Matrigel droplets and cultured in cortical differentiation medium for 16 days, and then cultured in cortical maturation media thereafter. ( F ) WB analysis of WT and ATM-KO cortical organoids at day 35 of differentiation. Immunofluorescence of ( G ) PAX6 and CTIP2 and ( J ) KI67 in cryosections of cortical organoids at day 35 of differentiation. Bar, 100 μm. At day 35 of differentiation, the ( H ) area and ( I ) thickness of VZ-like regions were compared between groups. Data points represent single organoids. The mean ± SEM values were compared by one-way ANOVA with Dunnett’s multiple comparisons test to yield **** indicating p < 0.0001. n = 13 organoids/group. Underlying numerical values for figures are found in S1 Data . ATM, ataxia telangiectasia mutated; DMEM, Dulbecco’s Modified Eagle Medium; GMEM, Glasgow Modified Essential Medium; hESC, human embryonic stem cell; hNPC, human neural progenitor cell; IgG, immunoglobulin G; KO, knockout; KSR, Knockout Serum Replacement; VZ, ventricular zone; WB, western blot; WT, wild type; 53BP1, p53 binding protein 1; 53BP1-pS25, 53BP1 phosphorylated at serine 25.

https://doi.org/10.1371/journal.pbio.3002760.g001

ATM is required for 53BP1-S25 phosphorylation, cell differentiation, and tissue morphogenesis in cortical organoids

The ATM kinase phosphorylates 53BP1-S25 [ 15 ], and we thus investigated whether ATM plays a role in neural differentiation. First, we found that ATM co-immunoprecipitated with 53BP1, as did the positive control UTX, but not with the negative control SUZ12 (a core subunit of PRC2, which does not bind these proteins; Fig 1B ). Similarly, 53BP1 co-immunoprecipitated with ATM, but not with the negative control SUZ12 ( Fig 1C ). Like 53BP1-pS25, ATM levels were significantly increased in NPCs compared with hESCs ( Fig 1D ). ATM up-regulation in NPCs was shown by a previous DNA damage response study [ 21 ].

Next, we used the CRISPR-Cas9 system to generate 4 ATM -knockout (KO) hESC lines (Figs 2A , 2B , and S1J ). Data from RNA-seq and immunofluorescence showed that ATM -KO did not markedly alter hESC pluripotency ( S2C and S2D Fig ). All cell lines underwent karyotyping analysis and were characterized as karyotypically normal ( S1 Table ). ATM -KO lines had minor abnormalities, as expected due to the requirement of ATM for DNA damage repair. To analyze the role of ATM in human cortical development, we used an established protocol to differentiate wild type (WT) and ATM -KO hESCs into cortical organoids ( Fig 1E , Methods; [ 6 ]). We did not detect 53BP1-pS25 in ATM -KO D35 cortical organoids ( Fig 1F ) nor NPCs ( S2B Fig ), consistent with loss of ATM-mediated phosphorylation of 53BP1-S25 during neural differentiation of hESCs. ATM -KO modestly reduced γH2AX levels in NPCs ( S2E Fig ), suggesting that ATM promotes the phosphorylation of H2AX-S139 in NPCs.

Quantification of PAX6/CTIP2 ratios in ( A ) D28 and ( B ) D37 cortical organoids. ( C ) Quantification of NEUN/DAPI in D37 cortical organoids. ( D ) In D37 cortical organoids, 6 organoids were surveyed to count ZO-1-positive apical surfaces and proportions of PAX6-positive NPCs that are organized around the apical surfaces. ( E ) Proportions of PH3-positive cells that are adjacent to ZO-1-positive apical surfaces (“rings”). Data from 53BP1-S25A and S25D were included for comparison. **, p < 0.01; ***, p < 0.001; ****, p < 0.0001; ns, not significant by Welch’s t test in (A-C) and two-way ANOVA test in (D). From GSEA, functional terms that are highly enriched in ( F ) up-regulated and ( G ) down-regulated genes in ATM -KO vs. WT NPCs. % Match, % of genes in the enriched term that overlap the differentially expressed genes or proteins. Underlying numerical values for figures are found in S1 Data . ATM, ataxia telangiectasia mutated; GSEA, gene set enrichment analysis; KO, knockout; NES, normalized enrichment score; NPC, neural progenitor cell; WT, wild type; 53BP1, p53 binding protein 1.

https://doi.org/10.1371/journal.pbio.3002760.g002

By day 35 (D35) of differentiation, WT cortical organoids expressed the forebrain NPC marker PAX6 in ventricular zone–like regions that were radially organized ( Fig 1G ). In contrast, ATM -KO D35 cortical organoids displayed disorganized and smaller ventricular zone–like regions ( Fig 1G-1I ). We quantitatively compared NPC proliferation, neuronal differentiation, cell death, and cell organization in ATM -KO versus WT cortical organoids. Examination of endogenous DNA damage, by γH2AX immunofluorescence, did not reveal marked difference ( S3A Fig ), confirming our western blot (WB) results in S1I Fig . Although quantification of cell death marker cleaved-caspase 3 by FACS revealed a modest increase of cell death in D21 ATM-KO cortical organoids, FACS and immunofluorescence quantification showed that D28 and D35 ATM -KO and WT cortical organoids are similar in cell death frequencies ( S3B-S3E Fig ). Cell proliferation frequencies did not significantly differ between D28 and D35 ATM -KO and WT cortical organoids ( S4 Fig ). We next quantified immature neuronal marker NEUN and PAX6/CTIP2 ratios. Despite lower levels of immature neuronal differentiation, ATM-KO exhibited higher neuronal maturation (Figs 2A-2C and S5A ). These data suggest that ATM -KO fastens the phase of immature neuronal differentiation, leading to enhanced neuronal maturation. Finally, we quantified ZO-1-positive ventricular surfaces and the organization of PH3-positive and PAX6-positive cells around ventricular surfaces. The ATM -KO ventricular surfaces were similar to WT at D28 ( S5B-S5D Fig ), but the number was much reduced by D37 ( Fig 2D ). NPC organization around the ventricular surfaces were similarly organized in D37 (Figs 2D and S5E ); however, fewer ATM -KO proliferative cells were adjacent to ventricular surfaces ( Fig 2E ). These data suggest that ATM -KO enhances neuronal maturation and cellular disorganization in developing cortical organoids. By D55, ATM -KO cortical organoids had similar size distribution as the control ( S5F and S5G Fig ). Thus, ATM controls neuronal differentiation and cellular organization to form ventricular zone–like regions in cortical organoids.

ATM safeguards transcriptional and translational programs in differentiating cortical organoids

To investigate the molecular basis of the cellular defects we observed in ATM -KO, we performed RNA-seq to compare ATM -KO to WT NPCs and D35 cortical organoids derived from WT and ATM -KO hESCs. The expression of forebrain markers was similar between WT and ATM -KO cortical organoids (and low expression of midbrain and hindbrain markers; S2 Table ), suggesting that the ATM -KO cortical organoids specified to the forebrain lineage. A false discovery rate (FDR) <0.05 was used to identify differentially expressed genes. Gene set enrichment analysis (GSEA) showed that up-regulated genes in ATM -KO NPCs were enriched in forebrain development, axis specification, and metabolic pathways (Figs 2F and S6A ), whereas down-regulated genes were enriched in neuronal differentiation, epithelial mesenchymal transition, and tube morphogenesis ( Fig 2G ). Comparison of transcriptomes profiles of D35 cortical organoids from 8 ATM -KO versus 6 WT datasets yielded similar GSEA terms ( Fig 3A and 3B ). These data suggest that ATM regulates genetic programs related to forebrain development, metabolism, and neuronal differentiation in NPCs and cortical organoids. Dysregulated genetic programs likely contributed to enhanced neuronal maturation and cellular disorganization in ATM -KO cortical organoids.

From GSEA, functional terms that are highly enriched in ( A ) up-regulated and ( B ) down-regulated genes in ATM -KO D35 cortical organoids. % Match, % of genes in the enriched term that overlap the differentially expressed genes or proteins. ( C ) Schematic diagram outlining TMT LC-MS/MS profiling of total proteomics and phosphoproteomics of D35 WT and ATM -KO cortical organoids. TMT signals from total proteomics were used to normalize those of phosphopeptides. ( D ) Using FC>1.5 and FDR<0.05, 198 phosphoproteins were found to be lower in 2 ATM -KO versus WT. ( E ) Normalized levels of phosphoproteins that have ATM-dependent phosphorylation in D35 cortical organoids. 53BP1 and EIF4EBP1 were known substrates of ATM. The error bars depict the mean and standard error of the mean values, which were calculated based on the normalized levels of each phosphopeptide in the protein. ( F ) Enrichment of proteins with ATM-dependent phosphorylation in specific functional categories. ( G ) Heatmap showing altered activities of kinases between D35 ATM -KO2 and WT cortical organoids. Relative changes in kinase activity are shown as row Z‐scores. Kinase activity was inferred by IKAP [ 25 ] based on normalized substrate phosphorylation levels from phosphor-proteome. The normalization was performed by dividing phosphor-peptide abundance of each protein by corresponding protein abundance [ 57 ]. Circos plots showing kinases with inferred ( H ) higher and ( I ) lower activities in D35 ATM -KO versus WT cortical organoids and their corresponding enriched pathways. Underlying numerical values for figures are found in S1_Data.xlsx. ATM, ataxia telangiectasia mutated; FC, fold-change; FDR, false discovery rate; GSEA, gene set enrichment analysis; KO, knockout; NES, normalized enrichment score; TMT LC-MS/MS, liquid chromatography-tandem mass spectrometry; WT, wild type; 53BP1, p53 binding protein 1.

https://doi.org/10.1371/journal.pbio.3002760.g003

As ATM kinase is crucial for many cell and developmental processes, we aimed to analyze its effect on the proteome and phosphoproteome of differentiating cortical organoids. First, we used multiplexed tandem mass tag-based quantification and 2D liquid chromatography-tandem mass spectrometry (TMT LC-MS/MS) to profile the proteome of WT and ATM -KO D35 cortical organoids ( S3B Fig , Methods). We quantified 10,895 proteins between 4 WT, 4 ATM -KO2, 3 ATM -KO3, and 3 ATM -KO14 D35 cortical organoid samples by using the criteria of fold change >1.5 and FDR <0.05 ( Fig 3C ). Consistency between replicate datasets is supported by principal component analysis ( S3B Fig ). GSEA showed that compared to WT, up-regulated proteins in ATM -KO were enriched in terms related to neurotransmission, neuron spine, dendrite, synapse, and axon ( S3C Fig ), whereas down-regulated proteins were enriched in BMP/TGFβ and WNT signaling, epithelial morphogenesis, and stem cell differentiation ( S3D Fig ). These data suggest that ATM controls posttranscriptional and translational gene regulation to suppress neuronal function and promote stem cell differentiation, epithelial morphogenesis, and TGFβ and WNT signaling pathways in D35 cortical organoids.

We have observed distinct patterns in the transcriptomics and proteomics data in ATM -KO versus WT cortical organoids. Interestingly, while transcriptomic programs related to neuronal differentiation were down-regulated ( Fig 3B ), proteomic programs related to neuronal function were up-regulated ( S3C Fig ) in ATM -KO versus WT cortical organoids. These differential patterns in transcriptomics and proteomics are likely a consequence of the regulatory role of ATM in multiple cellular processes. It is possible that the higher protein expression related to neuronal functions in ATM -KO lead to down-regulation of transcriptional expression of neuronal differentiation programs. This would suggest that the dysregulated transcriptomic and proteomic programs in ATM-KO cortical organoids are interconnected and result from the complex interplay of ATM’s regulation of various cellular pathways. These findings shed light on the intricate role of ATM in coordinating gene expression and protein levels, influencing neuronal differentiation and function in cortical organoids.

ATM-dependent phosphorylation controls signaling pathways for neurogenesis, stem cell differentiation, and morphogenesis in cortical organoids

To investigate how ATM exerts its modulatory control during cortical organoid formation, we performed phosphoproteomics analysis of WT and ATM -KO cortical organoids. Using TMT LC-MS/MS, we quantified 22,646 phosphopeptides and normalized their abundance based on the protein abundance measured in the total proteomics analysis. A comparison between WT and ATM -KO lines revealed that 198 proteins had consistently lower levels of phosphorylation in at least 2 of the 3 ATM -KO lines (log2(fold change >1.5) and FDR <0.05; Fig 3D and S3 Table ). Among these proteins, 53BP1 and EIF4EBP1 were known substrates of ATM ( Fig 3E ) [ 15 , 16 , 22 , 23 ], validating the approach to identify putative ATM substrates in cortical organoids. However, it is essential to note that this approach does not distinguish between direct and indirect effects, and, therefore, some of the identified proteins could be phosphorylated by protein kinases that require ATM for their activity. Notably, many ATM-dependent phosphorylated proteins were found to be key neurodevelopmental regulators ( Fig 3E ) and enriched in functions related to neurodevelopment, neurogenesis, cell morphogenesis, and cytoskeleton ( Fig 3F ). These findings suggest that ATM plays a critical role in regulating the phosphorylation of proteins involved in essential processes for neurodevelopment and neuronal function in cortical organoids.

We further explored the effects of ATM by identifying protein kinases that had ATM-dependent phosphorylation. We used the IKAP machine learning algorithm [ 24 ] to analyze substrates (inferred from literature curation) and deduce the activities of those kinases. For example, in ATM -KO compared to WT, we found reduced phosphorylation of proteins related to MAPK9 activities, such as DCX, MAPT, and NFATC4 ( S7A Fig and S4 Table ) [ 24 ]. On the other hand, we found higher phosphorylation of proteins related to CDK5 activities, including ADD2, ADD3, DCX, DNM1L, DPYSL3, MAPT, and SRC ( S7B Fig and S4 Table ) [ 24 ]. In ATM -KO, we inferred lower activities in MAPK9, CDK2, CHEK1, ATR, CSNK1A1, MTOR, CAMK2A, and PRKACA (Figs 3G and S7C ), with enriched functions in ATM signaling, BNDF signaling, and axon guidance (Figs 3G , 3H , and S7C ). On the other hand, we inferred higher activities in GSK3B, MAPK3, PAK1, CSNK2A1, CDK5, CDK1, and PRKDC (Figs 3G , 3I , and S7C ), with enriched function in ATM signaling, WNT signaling, G2/M checkpoint, and p53 regulation in ATM -KO ( Fig 3H and 3I ). ATM KO leads to both lower and higher activities of kinases in ATM signaling. Additionally, some of the altered kinase activities could be secondary to ATM -KO, as CHEK1, ATR, and PRKDC were known substrates of ATM [ 23 , 25 ]. Overall, these data suggest that the activities of kinases related to ATM signaling, BNDF signaling, WNT signaling, G2/M checkpoint, and p53 regulation became dysregulated in ATM -KO D35 cortical organoids.

We thus conclude that ATM plays a crucial role in controlling key neurodevelopmental regulators. The dysregulated phosphorylation and activities of these regulators disrupt the normal transcriptomic program responsible for neuronal differentiation, leading to higher proteomic programs associated with neuronal function. As a consequence, the dysregulated programs in ATM -KO cortical organoids are likely responsible for the observed defects in neurogenesis and morphogenesis (formation of ventricular zone–like regions). These findings provide valuable insights into the role of ATM in neurodevelopment and shed light on potential molecular mechanisms underlying neurological disorders associated with ATM dysfunction.

Phosphorylation of 53BP1-S25 coordinates NPC proliferation and neuronal differentiation

We next examined ATM-dependent phosphorylation of 53BP1-S25. To specifically investigate the functional significance of 53BP1-pS25, we used the CRISPR-Cas9 system to mutate the endogenous 53BP1 serine 25 to alanine (S25A) or aspartic acid (S25D) (Figs 4A and S7D , Methods). The alanine substitution precludes phosphorylation, whereas aspartic acid is chemically similar to phosphoserine [ 26 ]. We generated 4 53BP1-S25A hESC lines (34–3, 34–4, 79–1, and 79–3) and 4 53BP1-S25D hESC lines (14–3, 14–15, 14–19, and 17). The total levels of 53BP1 were similar in WT, 53BP1-S25A, and 53BP1-S25D NPCs, and we did not detect pS25 in 53BP1-S25A NPCs, as expected ( S1D and S7E Figs). Control, 53BP1-S25A, and 53BP1-S25D hESC lines displayed similar transcriptomic profiles and pluripotency marker expression ( S2C , S7F , and S8A Figs), suggesting that 53BP1-S25A and 53BP1-S25D do not affect hESC self-renewal.

ATM is required for the phosphorylation of many neurodevelopmental regulators ( Fig 3F ). As the role of 53BP1-S25 beyond DNA damage repair is not known, we seek to analyze its role in human cortical development. We differentiated control WT, 53BP1-S25A, and 53BP1-S25D hESCs into cortical organoids ( Fig 1E , Methods). The 53BP1-S25A and 53BP1-S25D D35 cortical organoids displayed smaller sizes compared to WT controls ( Fig 4B and 4C ), suggesting that phosphorylation at S25 is essential for cortical organoid growth and development. Analysis of the ventricular zone–like regions showed 53BP1-S25A and 53BP1-S25D are significantly smaller than those in WT ( Fig 4D-4F ). Fewer cells were positive for KI67 (proliferation marker) or phosphorylated-serine 10 histone H3 (mitotic chromatin marker) ( S8B–S8D Fig ). Examination of endogenous DNA damage and cell death, assessed by γH2AX and cleaved-caspase 3, respectively, did not reveal significant differences between 53BP1-S25A, S25D, and WT ( S9 Fig ). To explore the developmental timing of the cellular phenotypes, we quantified KI67, NPC marker PAX6, and neuronal marker CTIP2 in D14, D21, D28, and D35 cortical organoids. At D28, 53BP1-S25A and S25D cortical organoids had significantly lower cell proliferation and higher neuronal differentiation ( Fig 5 ). Quantification of the tight junction protein ZO-1 showed significantly fewer ZO-1-positive ventricular surfaces in D28 53BP1-S25A and S25D cortical organoids compared to WT ( S10 Fig ). For the ventricles that did form in D28 53BP1-S25A and S25D, their surface areas did not significantly differ from those of WT ( S10B Fig ). These data suggest that lower ventricle formation, lower cell proliferation, and higher neuronal differentiation contributed to the depletion of progenitor pools and smaller cortical organoids in 53BP1-S25A and S25D.

( A ) In the endogenous 53BP1 locus, the codon TCT encoding serine-25 in was mutated to GCT and GAT encoding alanine and glutamate, respectively. ( B ) Bright-field images of cortical organoids formed by 4 53BP1-S25A lines, 4 53BP1-S25D lines, and 2 WT control at day 35 of differentiation. Bar, 1.5 mm. At day 35 of differentiation, the ( C ) organoid size and ( F ) area of ventricular zone–like region were compared between groups. Data points represent single organoids. The mean ± SEM values were compared by one-way ANOVA with Dunnett’s multiple comparisons test to yield ****, ***, **, *, and ns indicating p < 0.0001, 0.001, 0.01, 0.05, and not significant, respectively. n = 39–47 organoids/group for ( C ) and 15–33 organoids/group for ( F ). ( D ) Immunofluorescence of PAX6 and CTIP2 in cryosections of cortical organoids at day 35 of differentiation. Bar, 100 μm. ( E ) Illustration of ventricular zone–like areas in cortical organoids. Underlying numerical values for figures are found in S1 Data . WT, wild type; 53BP1, p53 binding protein 1; 53BP1-pS25, 53BP1 phosphorylated at serine 25.

https://doi.org/10.1371/journal.pbio.3002760.g004

FACS quantified ratios of ( A ) KI67, ( B ) PAX6, and ( C ) CTIP2 to total cells in D28 cortical organoids. ( D ) Immunofluorescence of PAX6 and CTIP2 in D28 cortical organoids. Bar, 100 μm. Quantification of immunofluorescence signals of ( E ) PAX6/DAPI, ( F ) CTIP2/DAPI, and ( G ) PAX6/CTIP2 in D28 cortical organoids. Each data point represents quantification of cells in 1 cortical organoid. Quantification of ( H ) KI67/PAX6 and ( I ) PAX6/CTIP2 ratios in immunofluorescence of D35 cortical organoids. Each data point represents quantification of cells in 1 cortical organoid. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001; ns, not significant by two-way ANOVA test. Underlying numerical values for figures are found in S1 Data .

https://doi.org/10.1371/journal.pbio.3002760.g005

At D55, the 53BP1-S25A and S25D cortical organoids remained significantly smaller than WT ( S12 Fig and S5 and S6 Tables). These data suggest that the cell biological effects of the S25A and S25D mutations were similar, despite the aspartic acid mutation (S25D) being chemically similar to phosphoserine, which is the phosphorylated form of S25. The S25D mutation may act as an inhibitory mimic of phosphorylation, akin to the S25A mutation. Consequently, the absence of S25 phosphorylation impacts NPC proliferation and overall cortical organoid growth.

Phosphorylation of 53BP1-S25 modulates the expression of genetic programs for neuronal differentiation and function

Using RNA-seq, we examined the transcriptomes of WT (6 samples), 53BP1-S25A (8 samples), and 53BP1-S25D (8 samples) D35 cortical organoids. We analyzed expressed genes with counts per million values >1 and observed few differences in gene expression between 53BP1-S25A and 53BP1-S25D D35 cortical organoids, using FDR <0.05 ( Fig 6A ). When comparing the transcriptomes of 53BP1-S25A and 53BP1-S25D organoids to WT, there were high concordant changes in gene expression, with over 87% of differentially expressed genes in 53BP1-S25A also being altered in 53BP1-S25D (Figs 6B and S12D ). However, 53BP1-S25D disrupted the expression of 2- to 3-fold more genes than 53BP1-S25A, suggesting a gain-of-function effect for the 53BP1-S25D mutation. To explore this further, we performed GSEA and found that the top terms enriched in the up-regulated genes of 53BP1-S25A and S25D organoids were highly overlapping ( Fig 6C ). In contrast, there was low overlap of the top terms in the down-regulated genes in 53BP1-S25A versus WT and 53BP1-S25D versus WT ( S12E Fig ). Both mutations led to the up-regulation of genes related to synapse, axon, and neurotransmitter functions, suggesting a shared effect on enhancing neuronal function ( Fig 6C ). The S25D mutation specifically up-regulated more genes involved in neuronal function compared to S25A, indicating a stronger impact on this aspect of gene regulation ( Fig 6C and 6D ). These findings highlight the significance of the S25 phosphorylation site in 53BP1 for the regulation of genes involved in neuronal function and support that the S25D mutation results in a gain-of-function effect, leading to more pronounced changes in gene expression related to neuronal processes.

( A ) Number of differentially expressed genes identified by pairwise comparisons at FDR <0.05. At day 35 of differentiation, 53BP1-S25A and S25D cortical organoids are molecularly similar. ( B ) Differentially expressed genes in 53BP1-S25D versus WT overlap 87% (764/875) and 91% (361/396) of those in 53BP1-S25A versus WT. ( C ) Extensive overlap of up-regulated GSEA terms between 53BP1-S25A versus WT and 53BP1-S25D versus WT. Most terms relate to axon, synapse, and neurotransmitter. ( D ) Of 53BP1 target genes up-regulated by S25A and S25D, 212 genes require WT 53BP1 for expression in cortical organoids. ( E ) The 212 genes are enriched in functions related to transcriptional regulation, neuron projection, axonogenesis, synapse, neurotransmitter synthesis and transport, and membrane depolarization. ( F ) Venn diagrams depict high overlaps between down-regulated genes in all 3 groups of mutant versus WT pairwise comparisons. ( G ) GSEA graphs showed that down-regulated genes in 53BP1-S25A or S25D vs. WT had significant enrichment in down-regulated genes of ATM-KO vs. WT cortical organoids. P values were calculated by the hypergeometric test, assuming normal data distribution. ( H ) GSEA terms of the 115 genes that were down-regulated in all 3 groups (versus WT) revealed the genetic programs copromoted by ATM and 53BP1-pS25. ATM, ataxia telangiectasia mutated; FDR, false discovery rate; GSEA, gene set enrichment analysis; KO, knockout; WT, wild type; 53BP1, p53 binding protein 1.

https://doi.org/10.1371/journal.pbio.3002760.g006

It remained unclear whether the higher expression of neuronal genetic programs in the 53BP1 mutants occurred in NPCs or neurons. Therefore, we compared the transcriptomes of 53BP1-S25A and S25D to WT NPCs, which had similar expression of NPC markers PAX6 and NES ( S12A Fig ). Up-regulated genetic programs in 53BP1-S25A and S25D NPCs shared categories such as translation control and ribosome ( S12B and S12C Fig ), whereas 53BP1-S25A NPCs also up-regulated cell cycle control and chromosome segregation ( S12C Fig ). Surprisingly, down-regulated genetic programs in 53BP1-S25A and S25D NPCs were highly enriched in neuronal differentiation ( S12D and S12E Fig ). The down-regulated genetic programs in NPCs are similar to neuronal programs that became up-regulated in 53BP1-S25A and S25D versus WT cortical organoids. These data suggest that 53BP1-S25 phosphorylation promotes the appropriate expression of neurogenic programs in NPCs and modulates the expression of the same programs in differentiating neurons in cortical organoids.

To dig deeper into analyses, we compared our data with previously published transcriptomic data that compared 53BP1 -KO and WT cortical organoids, which support a requirement of 53BP1 for activating neurogenic genes [ 6 ]. We observed that gene categories up-regulated by 53BP1-S25A and S25D were similar to those down-regulated in 53BP1 -KO cortical organoids. This was a significant overlap of 212 genes up-regulated by 53BP1-S25A and 53BP1-S25D with 53BP1-bound target genes that were down-regulated in 53BP1 -KO versus WT ( p = 0 by empirical estimation; Fig 6D ). The 212 genes were enriched in functions related to regulation of transcription, neurogenesis, neuronal projection, axonogenesis, synapse organization, and membrane depolarization ( Fig 6E ). This suggests that the expression of these genes is dependent on and up-regulated by 53BP1 phosphorylated at S25 in cortical organoids.

We next examined how transcriptomic changes in 53BP1-S25A and S25D compared to those in ATM -KO cortical organoids. We observed little overlap between the down-regulated genes in ATM -KO and the up-regulated genes in 53BP1-S25A and 53BP1-S25D. In contrast, we observed a greater overlap in concordant gene expression changes in ATM -KO, 53BP1-S25A, and 53BP1-S25D versus WT (Figs 6F , S12F , and S12G ). GSEA showed a significant enrichment of concordantly differentially expressed genes among ATM -KO, 53BP1-S25A, and 53BP1-S25D versus WT (Figs 6G and S13A-S13C ). Notably, all 3 mutant types shared down-regulated genes that were enriched functions related to TNFα signaling via NFκB, p53 pathway, IRE1-mediated unfolded protein response, FGFR signaling, TGFβ signaling, apoptosis, regulation of cell proliferation, and epithelial mesenchymal transition ( Fig 6G ). These data suggest that both ATM and 53BP1-pS25 promote the expression of these genes. From these findings, we can infer that ATM likely promotes the expression of these genes via phosphorylating 53BP1 at S25 in D35 cortical organoids. This suggests that ATM and 53BP1 may function together in a coordinated manner to regulate the expression of genes involved in critical signaling pathways and cellular processes during cortical development.

53BP1-S25A and S25D predominantly alter the expression of 53BP1 target genes

To obtain further mechanistic insights into the role of 53BP1 in controlling gene expression, we reanalyzed 53BP1 ChIP-seq data (using 2 separate anti-53BP1 antibodies) in WT NPCs [ 6 ]. Using SICER [ 27 ] and MACS2 [ 28 ] with a criterion of FDR <0.05, we identified 37,519 targets bound by 53BP1. About 41% of these 53BP1 targets localize to promoter regions, suggesting a transcriptional regulatory role of 53BP1 ( S14D Fig ). Remarkably, more than 82% of the differentially expressed genes in 53BP1-S25A and 53BP1-S25D D35 cortical organoids were found to be targets bound by 53BP1 (Figs 7A , S13E , and S13F ). 53BP1 target genes with increased transcript levels in the mutant organoids were highly enriched in neuronal development, axonogenesis, neuron projection, synapse organization, and neurotransmitter transport, transmission, and signaling ( Fig 7B ). On the other hand, 53BP1 targets with reduced transcript levels in the mutant organoids were enriched in IRE1-mediated unfolded protein response, cellular response to stress, iron import, and apoptosis regulation ( S13G Fig ). Of note, genes involved in IRE1-mediated unfolded protein response and apoptosis regulation showed reduced expression upon loss of ATM or mutation of 53BP1-S25 and were identified as direct targets of 53BP1 in NPCs. This suggests that ATM-mediated phosphorylation of 53BP1-S25 directly promotes the expression of these genes to maintain NPCs during formation of cortical organoids.

( A ) More than 82% of differentially expressed genes in 53BP1-S25A or S25D versus WT are chromatin targets bound by 53BP1 in WT NPCs. ( B ) 53BP1-S25A and S25D up-regulate 53BP1 targets that are involved in neuron development and projection, axonogenesis, synapse, and neurotransmitter synthesis and transport. ( C ) Heatmaps aligning peaks with 53BP1-pS25 CUT&RUN and 53BP1 ChIP-seq signals in WT NPCs. Input track was included as a negative control. n = numbers of peaks with differential and overlapped bindings. Criteria of FC>2 and p < 0.05 were used for comparison. ( D ) GSEA graph of 53BP1-pS25 CUT&RUN signals in genes that were lower in ESCs vs. NPCs, which were up-regulated in NPCs. P values were calculated by the hypergeometric test, assuming normal data distribution. Heatmaps aligning peaks with significantly different 53BP1 ChIP-seq signals in ( E ) 53BP1-S25A vs. WT and ( F ) 53BP1-S25D vs. WT, using the criterion of FC>2 and p < 0.05. Control peaks are those, after voom normalization, showed the least changes and served as semi-independent validation of differential ChIP-seq analysis. Bubble graphs present top enriched categories of genes that had significantly higher 53BP1 ChIP-seq in ( G ) 53BP1-S25A vs. WT and ( H ) 53BP1-S25D vs. WT. ESC, embryonic stem cell; FC, fold-change; GSEA, gene set enrichment analysis; NPC, neural progenitor cell; WT, wild type; 53BP1, p53 binding protein 1; 53BP1-pS25, 53BP1 phosphorylated at serine 25.

https://doi.org/10.1371/journal.pbio.3002760.g007

We wanted to test whether ATM alters 53BP1 binding, considering ATM is required for 53BP1-pS25 in D35 cortical organoids ( Fig 1F ) and NPCs ( S2B Fig ). A comparison of 53BP1 ChIP-seq in WT and ATM -KO NPCs showed that ATM -KO altered 53BP1 binding to chromatin ( S14A-S14C Fig ). ATM -KO reduced 53BP1 binding at specific sites, with 96.3% of these sites being promoters ( S14C Fig ). To explore the impact of 53BP1-pS25, we performed CUT&RUN in 2 separate WT NPC lines. Our analysis revealed that 67.1% of 53BP1-pS25 targets localize to promoter regions, suggesting a transcriptional regulatory role ( S14D Fig ). Under the criteria of fold-change >2 and p < 0.05, 58.6% (3,390/5,789) of 53BP1-pS25 targets overlapped with 53BP1 targets ( Fig 7C ); the nonoverlapped sites may be attributed to differences in ChIP versus CUT&RUN procedures and the accessibility of 53BP1 versus 53BP1-pS25 antibodies. 53BP1-pS25 targets were significantly enriched in 414 up-regulated genes in NPCs versus ESCs ( Fig 7D ), suggesting a role of 53BP1-pS25 in promoting their expression in NPCs. Genes having overlapped 53BP1 ChIP-seq and 53BP1-pS25 CUT&RUN signals were enriched in chromatin remodeling, DNA metabolism, RNA splicing, translation, transcription, cell cycle, and neuron development ( S14E Fig ). These data suggest that ATM can alter 53BP1 binding and that 53BP1-pS25 is enriched in the promoters of genes regulating cellular processes and neurodevelopment.

Phosphorylation of 53BP1-S25 controls the localization of 53BP1 to chromatin for gene regulation

To investigate the impact of 53BP1-S25 on the genomic distribution of 53BP1, we performed ChIP-seq in 53BP1-WT, S25A, and S25D NPCs. Two independent NPC lines were used for each group, and the ChIP-seq data were subjected to principal component analysis, which showed high consistency between the replicate dataset ( S8A Fig ). We used SICER [ 27 ] and MACS2 [ 28 ] with the criteria of fold-change >2 and p < 0.05 to perform pairwise comparisons of the merged datasets from 53BP1-WT, S25A, and S25D ChIP-seq experiments. The pairwise comparisons identified thousands of 53BP1-bound regions that were significantly different between 53BP1-WT, S25A, and S25D. Notably, the regions that significantly gained binding in 53BP1-S25A or S25D versus WT were highly enriched at promoters (within 2 kb of transcription start sites), constituting 82% and 71.1%, respectively ( S8B and S8C Fig ). In contrast, the regions that significantly lost binding in 53BP1-S25A or S25D versus WT were not as enriched at promoters, constituting 32.6% and 33%, respectively ( S8B and S8C Fig ). We generated heatmaps to visualize the genomic regions with significantly different 53BP1 binding intensity (compared against control regions). The heatmaps confirmed consistent changes in 53BP1 binding patterns between 53BP1-S25A and S25D versus WT, and between 53BP1-S25A versus S25D (Figs 5C , 5D , and S8D ). These data support that 53BP1-S25 and its phosphorylation control the genomic distribution of 53BP1 on chromatin.

We next set out to examine the correlation between changes in 53BP1 distribution on chromatin and changes in gene expression in 53BP1-WT, S25A, and S25D cortical organoids. We performed GSEA and made some notable observations. Firstly, regions that gained 53BP1 binding in 53BP1-S25A or S25D cortical organoids, as compared to WT, were enriched with up-regulated genes ( Fig 5E and 5F ). Similarly, regions that had lower 53BP1 binding were enriched with down-regulated genes in 53BP1-S25A or S25D versus WT ( S8E and S8F Fig ). These results suggest that the 53BP1-S25A or S25D mutation directly influences 53BP1 binding and gene expression and subsequently regulates gene expression, particularly at promoters where higher 53BP1 binding leads to higher gene expression.

Interestingly, genes that lost 53BP1-S25A or S25D protein binding had minimal overlap in GSEA terms, except promoters occupied with H3K4me3 and regulation of epithelial-mesenchymal transition ( S8E and S8F Fig ). In contrast, the genes that gained 53BP1-S25A or S25D protein binding were enriched with promoters marked by bivalent histone marks (H3K4me3 and H3K27me3) or occupied by H3K27me3 alone [ 29 ] ( Fig 5E and 5F ), suggesting that 53BP1-S25D or S25A proteins preferentially bind to these promoters and subsequently up-regulate gene expression. Moreover, genes that gained 53BP1-S25A or S25D binding shared common functions related to sodium ion transmembrane transporter, DNA replication, positive regulation of cell division, and regulation of histone H3K4 methylation ( Fig 5E and 5F ). This suggests that despite the 1,187 regions showing different 53BP1 bindings between 53BP1-S25A and S25D ( S8D Fig ), both mutations impact genes involved in neuronal functions and cell proliferation. Altogether, these findings show that 53BP1-S25A and S25D mutations have a direct impact on 53BP1 binding to chromatin and subsequently affecting gene regulation. We propose that 53BP1-pS25 likely inhibits 53BP1 binding to promoters associated with bivalent and H3K27me3-occupied promoters. This inhibition may lead to the reduced expression of genes involved in the regulation of H3K4me3, neuronal functions, and cell proliferation.

Molecular regulation of ATM and 53BP1-pS25 during neural differentiation

We next tried to identify a regulation of ATM, whose protein levels increased in NPCs ( Fig 1D ). This led us to test whether and how inhibitors of TGFb, WNT, and HH signaling control protein levels of ATM, 53BP1, and pS25-53BP1 by removing one inhibitor at a time from the cortical organoid differentiation media ( S16A Fig ). As we could not successfully identify physical presence of ATM at promoters, WB analysis is most apt to study ATM level and activity. By day 4 of neural differentiation, although 53BP1 protein levels were reduced by the withdrawal of SB431542 (TGFβ inhibitor) or IWR1-endo (WNT inhibitor), pS25-53BP1 was not altered ( S16B and S16C Fig ). By day 10 of neural differentiation, the withdrawal of cyclopamine (HH inhibitor) reduced pS25-53BP1 level (but not ATM or 53BP1 proteins; S16D and S16E Fig ). These signaling pathways may affect pS25-53BP1 or ATM activities during neural differentiation.

Next, we tested whether another DNA damage response factor, apart from ATM, influences 53BP1-pS25. RNF168 plays a central role in the γH2AX-MDC1-RNF8-RNF168-H2AK15ub axis, which governs the binding of 53BP1 to chromatin with DNA damage [ 30 ]. We generated RNF168 -KO hESC clone 44, which maintained pluripotency and genome integrity ( S17A-S17D Fig and S1 Table ). RNF168 -KO hESCs were differentiated to NPCs, which expressed NPC markers similar to WT NPCs ( S17E Fig ). RNA-seq analysis comparing 2 datasets each from RNF168 -KO44 and WT NPCs revealed that up-regulated genes were enriched in neuronal differentiation, translation and ribosome, and cell cycle transition ( S17F Fig ), while down-regulated genes were enriched in cilium movement, H3K27me3 targets, H3K4me3 targets, astrocyte markers, signaling pathways, and positive regulation of NPC proliferation ( S17G Fig ). We performed 53BP1-pS25 CUT&RUN and showed that RNF168 -KO disrupted 53BP1-pS25 localization on chromatin ( S17H Fig ). RNF168 -KO increased 53BP1-pS25 levels at genes enriched in neuronal differentiation, cell morphogenesis, and stem cell maintenance, whereas RNF168 -KO decreased 53BP1-pS25 levels at genes enriched in cell cycle transition, signaling receptor regulation, anterior-posterior patterning, and transcription activator ( S17I Fig ). The altered 53BP1-pS25 localization correlated with differential gene expression in RNF168-KO versus WT NPCs ( S17J Fig ). Altogether, these data suggest that DNA damage signaling regulates 53BP1 binding to chromatin, affecting genetic programs related to signaling pathways, protein translation, and NPC proliferation and differentiation.”

In our study, we made significant discoveries regarding the role of ATM and 53BP1-pS25 in controlling gene expression during the differentiation of hESCs into cortical organoids. We revealed that ATM exerts a strong influence over various aspects of gene regulation, including transcriptional, posttranscriptional, and translational control. While our in vitro model may not fully recapitulate neurodevelopment in vivo, it provides valuable insights into corticogenesis. We have shown that neural differentiation promotes ATM protein levels, and ATM-dependent phosphorylation predominantly impacts factors involved in neurogenesis, neuronal differentiation, cell morphogenesis, and microtubule cytoskeleton. Dysregulation of these processes led to the cellular defects in ATM -KO cortical organoids. We showed that key signaling pathways may affect ATM during neural induction. The activity of ATM can be regulated by DNA damage response, reactive oxygen species, hypothxia, hypothermia, and phosphatase WIP1 [ 31 – 33 ]. The exact clarification of mechanisms promoting ATM activities, especially in directing its kinase activity at specific promoters, is beyond the scope of this study. Additionally, we have identified kinases involved in ATM, BDNF, and WNT signaling, G2/M checkpoint, and p53 regulation as being influenced by ATM-dependent phosphorylation during cortical organoid differentiation. These molecular pathways may function in diseases associated with ATM, including ataxia telangiectasia [ 34 – 36 ].

We recognized the diverse effects of ATM and decided to focus our studies on 53BP1-pS25, a phosphorylation event dependent on ATM. We found that 53BP1-pS25 regulates genetic programs including signaling pathways, p53 regulation, apoptosis, and cell proliferation. To understand the mechanisms underlying 53BP1’s involvement in gene regulation, we built a model that incorporates current knowledge about 53BP1 functions in the DNA damage response. We propose that ATM phosphorylates H2AX at transcription start sites [ 13 , 14 ], facilitating the recruitment of 53BP1 and subsequent phosphorylation of 53BP1-S25. RNF168, key to DNA damage response signaling [ 30 ], also regulates 53BP1-pS25 on chromatin and genetic programs crucial to neural differentiation. Phosphorylation of 53BP1-S25 inhibits the recruitment of 53BP1 to bivalent or H3K27me3-occupied promoters for suppressing the expression of genes involved in the regulation of H3K4me3, neuronal functions, and cell proliferation. The fidelity of gene expression in cortical brain organoids requires dynamic changes in the phosphorylation of 53BP1-S25. This process is likely to involve the interactions of 53BP1 with other proteins, including RIF1, SCAI, and UTX [ 6 , 11 , 12 ]. These interactors have known roles in chromatin alterations and gene regulation. Notably, UTX is an H3K27me3 demethylase that can modify bivalent or H3K27me3-occupied promoters and has been shown to partner with 53BP1 to promote neurogenesis in humans but not in mice [ 6 ]. Given our findings, we propose that 53BP1-pS25 may influence the activities of 53BP1–UTX at bivalent or H3K27me3-occupied promoters, thus modulating gene expression and contributing to the timing of neuronal differentiation.

Our studies have uncovered the remarkable role of ATM–53BP1 in regulating neurodevelopmental programs. Its impact is multifaced. Firstly, ATM–53BP1 plays a crucial role in maintaining NPCs and controlling the size of cortical organoids. Secondly, ATM–53BP1 is involved in driving and modulating programs related to synapse formation, axon development, and neurotransmitter regulation, processes fundamental for establishing neuronal networks and communication within the brain. Thirdly, our findings reveal a temporal component in the regulation of neurodevelopmental programs by ATM–53BP1. As cortical organoids progress in differentiation, there is a temporal regulation of neuronal differentiation and function. This switch involves ATM and the 53BP1-pS25 dynamics to specifically control genes associated with synapse, axon, and neurotransmitter, which are crucial to cognition. In the future, elucidation of this mechanism will provide valuable insights into the molecular control of corticogenesis. Beyond 53BP1, ATM-dependent phosphorylation likely controls many other key neurodevelopmental regulators. Future studies of how ATM selects substrates to exert its multiple influences will significantly advance our understanding of the epigenetic programming underlying human neurodevelopment.

Materials and methods

PBS: 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4 (pH 7.4). PBST: PBS with 0.1% Triton X-100. HEPM: 25 mM HEPES (pH 6.9), 10 mM EGTA, 60 mM PIPES, 2 mM MgCl2. Immunofluorescence blocking solution: 1/3 Blocker Casein (Thermo Fisher Scientific), 2/3 HEPM with 0.05% TX-100. Buffer A: 10 mM HEPES (pH 7.9), 10 mM KCl, 1.5 mM MgCl2, 0.34 M sucrose, 10% glycerol. Buffer B: 3 mM EDTA, 0.2 mM EGTA. Buffer D: 400 mM KCl, 20 mM HEPES, 0.2 mM EDTA, 20% glycerol. ChIP lysis buffer 3: 10 mM Tris-HCl (pH 8.0), 100 mM NaCl, 1 mM EDTA, 0.5 mM EGTA, 0.1% sodium deoxycholate, 0.5% N-Lauroylsarcosine. ChIP wash buffer: 50 mM HEPES (pH 7.5), 500 mM LiCl, 1 mM EDTA, 1% NP-40, 0.7% Na-deoxycholate. ChIP elution buffer: 50 mM Tris-HCl (pH 8.0), 10 mM EDTA, 1% SDS. CUT&RUN Wash buffer: 20 mM HEPES (pH 7.5), 150 mM NaCl, 0.5 mM spermidine, protease inhibitor cocktail (Sigma-Aldrich 11873580001). CUT&RUN Binding buffer: 20 mM HEPES-KOH (pH 7.9), 10 mM KCl, 1 mM CaCl2, 1 mM MnCl2. CUT&RUN Digitonin buffer: CUT&RUN Wash buffer with 0.01% digitonin. CUT&RUN Antibody buffer: CUT&RUN Digitonin buffer with 2 mM EDTA. CUT&RUN 2X Stop buffer: 340 mM NaCl, 20 mM EDTA, 4 mM EGTA. CUT&RUN Stop buffer: Into 1 mL of 2X Stop buffer stock, add 5 μL of 10 mg/mL RNase A and 3.3 μL of 15 mg/mL. GlycoBlue Coprecipitant (Thermo Fisher AM9516)

S7 Table lists all antibodies and conditions used in this study.

ESC culture and mutagenesis

H9/WA09 (WiCell) hESCs were grown on Matrigel with reduced growth factors (Thermo Fisher Scientific, #35423) in mTeSR1 medium (STEMCELL Technologies, #85850) at 37°C and 5% CO 2 . The 53BP1 knock-in cell lines (53BP1 S25A 34–3, 34–4, 79–1, 79–3 and S25D 14–3, 14–15, 14–19, 17) and ATM KO cell lines (ATM-KO2, 3, 14, and 43) were generated using CRISPR/Cas9 gene-editing technology. Genome editing reagents were designed and validated in the Center for Advanced Genome Engineering at St. Jude Children’s Research Hospital. Briefly, a chemically modified sgRNA (Synthego) was precomplexed with SpCas9 protein (St. Jude Protein Production Core) and cotransfected with an ssODN donor template containing the desired modification into H9/WA09 cells via nucleofection (Amaxa P3 primary cell 4D nucleofector X kit L, Lonza) using the manufacturer’s recommended protocol. Transfected cells were sorted (BD FACSAria Fusion) onto Matrigel and allowed to grown single-cell clones. Clones were identified via targeted mi-seq using a 2-step PCR library setup as previously described [ 37 ]. Samples were demultiplexed using the index sequences, fastq files were generated, and NGS analysis was performed using CRIS.py [ 38 ]. S8 Table lists genome editing reagents and associated primers.

Neural progenitor cell generation and culture

ESCs were seeded onto AggreWell800 plates (STEMCELL Technologies, #34811) and fed with neural induction medium (STEMCELL Technologies, #05835) to form embryoid bodies. On day 5, embryoid bodies were replated onto Matrigel-treated 6-well plates in the same media. On day 17, cells were harvested as NPCs.

Nuclear extract preparation and western blotting

ESCs and NPCs were incubated in Buffer A + PI + DTT for 5 min on ice. After centrifugation at 1,750 g for 2 min at 4°C, the nuclei pellet was washed in Buffer A and subsequently incubated for approximately 25 min in Buffer D + PI + DTT at 4°C with rotation to obtain the nuclear fraction. Nuclear extracts were separated by SDS–PAGE and transferred onto a nitrocellulose membrane (Bio-Rad). Membranes were blocked with 3% bovine serum albumin (BSA) in HEPM, incubated in primary antibodies (HEPM containing 1% BSA and 0.1% Triton X-100) overnight at 4°C, washed in PBS-T, incubated in IRDye-conjugated secondary antibodies (LI-COR), and imaged on an Odyssey Fc imaging system (LI-COR). Signals were quantitated with the Image Studio software (version 1.0.14; LI-COR).

Immunoprecipitation

Antibody was bound to protein A and protein G Dynabeads (Thermo Fisher 10002D and 10004D) for 2 h at room temperature. Nuclear extract was incubated with the Dynabeads-antibody complex for 5 h at 4°C, washed with PBST, and eluted with 0.1 M glycine (pH 2.3). Eluates were neutralized with 1/10 volume of 1.5 M Tris buffer (pH 8.8).

Cortical organoid differentiation

Cortical organoids were generated based on previously published methods with minor modifications [ 39 , 40 ]. In brief, hESC lines were expanded and dissociated to single cells using Accutase, seeded onto low-attachment V-bottom 96-well plates (Costar, #7007) at a density of 9,000 cells per well to aggregate into embryoid bodies. The embryoid bodies formation medium (DMEM/F-12 with 20% KO serum replacement, 3% ESC-quality FBS, 2 mM GlutaMAX, 0.1 mM nonessential amino acids) was supplemented with dorsomorphin (2 μM), WNT inhibitor (IWR1, 3 μM), TGF-β inhibitor (SB431542, 5 μM), and Rho kinase inhibitor (Y-27623, 20 μM). Starting from day 4, embryoid bodies were fed with cortical differentiation medium (Glasgow-MEM, 20% KSR, 0.1 mM NEAA, 1 mM sodium pyruvate, 0.1 mM β-ME, and 1% anti-anti), supplemented with WNT inhibitor (IWR1, 3 μM), TGF-β inhibitor (SB431542, 5 μM), cyclopamine (2.5 μM) and Rho kinase inhibitor (Y-27623, 20 μM). On day 17, embryoid bodies were embedded in Matrigel droplets and transferred onto low-attachment 6-wells and cultured in suspension using DMEM/F-12 supplemented with 1% N2 supplement, 1% lipid concentrate, 2% B27 supplement without vitamin A, and 1% anti-anti under 40% O 2 /5% CO 2 conditions on shaker. Starting from day 30, medium was changed to 50% DMED/F-12, 50% neurobasal media, 0.5% N2 supplement, 1% GlutaMax, 0.05 mM NEAA, 0.025% human insulin, 0.1 mM β-ME, and 1% anti-anti, supplemented with 2% B27.

Immunofluorescence

Cells and cryosectioned organoids were blocked with IF blocking solution for 2 h at room temperature and primary antibodies (diluted in blocking buffer) added and incubated O/N at 4°C. After 3 washes in PBS-T, fluorescent dye-conjugated secondary antibodies (1:500, Alexa Fluor-CONJUGATED antibodies, Thermo Fisher Scientific) were added and incubated for 3 h at room temperature. Secondary was washed with PBS-T 3 times, and samples were washed and coverslips mounted with Prolong Glass Mounting Reagent (Thermo Fisher Scientific), which contains DAPI. Images were acquired with Zeiss LSM780.

Organoid feature characterization by image analysis

At days 35 and 55, bright-field images of organoids were captured with Axiocam 208 (Zeiss). Area of organoids, area of ventricular zone–like regions, and marker-positive cells were quantified by using the software FIJI: Signals-positive cells were identified based on signal and width thresholds. For ventricular zone–like region quantification, inner and outer edges of the regions in the image were manually traced, based on CTIP2-positive cells encircling the outer edges. FIJI was used to quantify area, perimeter, major and minor axes of the inner and outer traces. Mean perimeter and the difference between the major axes of the inner and outer traces were used to estimate the thickness of the structure. Mean Perimeter = (outer perimeter + inner perimeter) / 2. MajorAxisDiff = (outer major axis − inner major axis) / 2. MinorAxisDiff = (outer minor axis − inner minor axis) / 2. To quantify ZO-1-positive ventricular surfaces, ZO-1 signals were normalized by the Integral Image Filters plugin, and surface areas were manually traced for quantification. The VZ/SVZ structure was considered organized if PAX6-positive nuclei were densely packed with radial organization around ZO-1-positive ventricular surfaces. Ilastik [ 41 ] was used to quantify nuclear areas positive for different markers, using segmentation via a machine learning-based package and area quantification of segmented areas. Marker ratios were then calculated based on quantified areas.

Quantification of cell populations by fluorescence-activated cell sorting (FACS)

Nine to 12 organoids of each line were dissociated using the papain dissociation system (Worthington LK003153). Dissociated cells were fixed in 4% formaldehyde solution at 4°C overnight and washed once in 1X PBS. Then, cells were permeabilized in 1X PBST for 2 h at room temperature on an orbital shaker. Cells were blocked in IF blocking buffer (1/3 Blocker Casein (Thermo Fisher 37528), 2/3 HEPM with 0.05% Triton X-100) for 2 h at room temperature on a shaker. Primary antibodies in IF blocking buffer were mixed with cells at 4°C overnight followed by washing twice with 1X PBST. Secondary antibodies in IF blocking buffer were mixed with cells for 2 h at room temperature on a shaker. After washing cells once, a conjugated antibody was added and incubated for 2 h at room temperature on a shaker. Cells were washed one last time before resuspended in 1X PBS for FACS. FACSymphony A1 sorter was used for analysis. All the centrifugation steps were done at 500 × g for 4 min at room temperature. All washes were performed by incubating the cells with 1X PBS (after fixation) or PBST (after antibody staining) for 5 min at room temperature on a shaker. Primary antibodies used are Ki67 (Cell Signaling 9129), PAX6 (DSHB supernatant 1mL), CTIP2 (Abcam 18465), and cleaved Caspase3-AF405-conjugated (R&D Systems IC835V).

Total RNA was extracted with TRIzol reagent (Invitrogen, #15596026) and Direct-zol RNA Microprep (Zymo Research, # R2062) by following manufacturer’s instructions. DNA digestion with DNase I was performed during RNA extraction. Paired-end 100-cycle sequencing was performed on NovaSeq6000 sequencer by following the manufacturer’s instructions (Illumina). Raw reads were first trimmed using TrimGalore (version 0.6.3) available at: https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/ , with parameters ‘—paired—retain_unpaired’. Filtered reads were then mapped to the Homo sapiens reference genome (GRCh38 + Gencode-v31) using STAR (version 2.7.9a) [ 42 ]. Gene-level read quantification was done using RSEM (version 1.3.1) [ 43 ]. To identify the differentially expressed genes between control and experimental samples, the variation in the library size between samples was first normalized by trimmed mean of M values (TMM) and genes with CPM < 1 in all samples were eliminated. Then, the normalized data were applied to linear modeling with the voom from the limma R package [ 44 ]. GSEA was performed against using the MSigDB database (version 7.1), and differentially expressed genes were ranked based on log 2 (FC) [ 45 , 46 ].

Protein extraction, digestion, and Tandem-Mass-Tag (TMT) labeling

Organoids were harvested on day 35, and the Matrigel droplets were eliminated by multiple ice-cold PBS washes. The organoid pellet was extracted in the lysis buffer (50 mM HEPES (pH 8.5), 8 M urea, and 0.5% sodium deoxycholate, 100 μl buffer per 10 mg tissue) with 1x PhosSTOP phosphatase inhibitor cocktail (Sigma-Aldrich). Protein concentration was estimated by a Coomassie stained short gel with BSA as a standard. About 600 μg each of protein samples was digested with LysC (Wako) at an enzyme-to-substrate ratio of 1:100 (w/w) for 2 h at room temperature in the presence of 1 mM DTT. The samples were then diluted to a final 2 M urea concentration with 50 mM HEPES (pH 8.5) and digested with Trypsin (Promega) at an enzyme-to-substrate ratio of 1:50 (w/w) for 3 h. The peptides were reduced by adding 1 mM DTT for 30 min at room temperature followed by alkylation with 10 mM iodoacetamide for 30 min in the dark at room temperature. The unreacted iodoacetamide was quenched with 30 mM DTT for 30 min. Finally, the digestion was terminated and acidified by adding trifluoroacetic acid to 1%, peptides desalted using Sep-Pak C18 cartridge (Waters), and dried by speed vac. The purified peptides were resuspended in 50 mM HEPES (pH 8.5) and labeled with 16-plex Tandem Mass Tag (TMTpro) reagents (Thermo Scientific) following the manufacturer’s recommendation. The TMT labeled samples were mixed equally, desalted using Sep-Pak C18 cartridge (Waters), and dried by speed vac.

Offline fractionation and two-dimensional liquid chromatography-tandem mass spectrometry (LC/LC-MS/MS)

The dried TMT mix was resuspended and fractionated on an offline HPLC (Agilent 1220) using basic pH reverse phase liquid chromatography (pH 8.0, XBridge C18 column, 4.6 mm × 25 cm, 3.5 μm particle size, Waters). A total of 160 one-minute fractions were collected and concatenated to 80 fractions. For whole proteome analysis, 10% of these 80 fractions was used. The remaining 90% of the 80 fractions were concatenated to 20 fractions for phophopeptide enrichment. Phosphopeptide enrichment was performed according to a previously published protocol [ 47 ]. The phosphopeptide enrichment eluents and the total proteome fractions were dried and resuspended in 5% formic acid and analyzed by acidic pH reverse phase LC-MS/MS analysis. The peptide samples were loaded on a nanoscale capillary reverse phase C18 column (New objective, 75 μm ID × approximately 15 cm, 1.9 μm C18 resin from Dr. Maisch GmbH) by a HPLC system (Thermo Ultimate 3000) and eluted by either a 125-min gradient (phosphofractions) or 110-min gradient for total proteome fractions. The eluted peptides were ionized by electrospray ionization and detected by an inline Orbitrap Fusion mass spectrometer (Thermo Scientific). For total proteome fractions, the mass spectrometer is operated in data-dependent mode with a survey scan in Orbitrap (60,000 resolution, 2 × 10 5 AGC target and 50 ms maximal ion time) and MS/MS high-resolution scans (60,000 resolution, 1 × 10 5 AGC target, 150 ms maximal ion time, 36.5 HCD normalized collision energy, 1 m/z isolation window, and 15-s dynamic exclusion). For phosphoproteome fractions, the mass spectrometer is operated in data-dependent mode with a survey scan in Orbitrap (60,000 resolution, 3 × 10 5 AGC target and 50 ms maximal ion time) and MS/MS high-resolution scans (60,000 resolution, 1 × 10 5 AGC target, 150 ms maximal ion time, 36.5 HCD normalized collision energy, 1 m/z isolation window, and 10-s dynamic exclusion).

Identification of proteins and phosphopeptides

The MS/MS raw data were processed by a tag-based hybrid search engine JUMP [ 48 ]. The data were searched against the UniProt human database (168,305 protein entries; downloaded in April 2020) concatenated with a reversed decoy database for evaluating FDR. Searches were performed using a 15-ppm mass tolerance for fragment ions, fully tryptic restriction with 2 maximal missed cleavages, 3 maximal modification sites, and the assignment of b and y ions. TMT tags on Lysine residues and N-termini (+304.2071453 Da) were used for static modifications and Met oxidation (+15.99492 Da) was considered as a dynamic modification. Phosphorylation (+79.96633 Da) was considered as a dynamic modification for STY residues. Putative peptide spectral matches (PSMs) were filtered by mass accuracy and then grouped by precursor ion charge state and filtered by JUMP-based matching scores (Jscore and ΔJn) to reduce FDR below 1% for proteins during the whole proteome analysis or 1% for phosphopeptides during the phosphoproteome analysis. Phosphosites were further evaluated by JUMPl program using the concept of the phosphoRS algorithm [ 49 ] to calculate phosphosite localization scores (Lscore, 0% to 100%) for each PSM.

Quantification of proteins and phosphopeptides

TMT reporter ion intensities of each PSM were extracted and corrected based on isotopic distribution of each labeling reagent. Those PSMs with very low intensities (e.g., minimum intensity of 1,000 and median intensity of 5,000) were excluded for quantification. Sample loading bias was mitigated by normalization with the trimmed median intensity of all PSMs. Protein or phosphopeptide relative intensities were calculated by dividing the intensity of each channel by the mean intensity. Protein or phosphopeptide absolute intensities were computed by multiplying the relative intensities by the grand-mean of 3 most highly abundant PSMs.

Differential expression analysis of proteins and phosphopeptides

Differentially expressed proteins between the 2 strains and 2 different doses were identified by the limma R package [ 50 ]. The Benjamini–Hochberg method was used to control multiple-testing correction, and proteins with an adjusted p -value of <0.05 and log2 fold change of >1.5 were defined as differentially expressed.

Pathway enrichment analysis for proteomics data

Pathway enrichment analysis was carried out to infer functional groups of proteins that were enriched in a given dataset. The 4 common pathway databases were used, including Gene Ontology (GO), KEGG, Hallmark, and Reactome. The analysis was performed using Fisher’s exact test with the Benjamini–Hochberg correction for multiple testing. A cutoff of adjusted p -value < 0.2 was used to identify significantly enriched pathways.

Estimation of kinase activity

Kinase activity was inferred based on known substrates in the PhosphoSitePlus database [ 51 ] using the IKAP algorithm [ 24 ]. The phosphoproteome data were normalized against the whole proteome. We performed 100 times of calculations to overcome the potential problem of local optimization.

Chromatin immunoprecipitation

Cells were harvested in PBS. Cytoplasmic fractions were extracted using buffer A with 1× protease inhibitors and 1 mM DTT. Nuclear pellets were cross-linked by 1.1% formaldehyde in buffer B with 1× protease inhibitors and 1 mM DTT; washed; and lysed in lysis buffer 3 with 1× protease inhibitors, 1 mM DTT, and 1 mM PMSF. The fixed and lysed nuclear extract was sonicated with Bioruptor Pico (Diagenode) 10 times for 15 s each, with 45-s intervals. Chromatin was added to Dynabeads (Life Technologies) prebound with 4 μg of antibodies for overnight incubation. After incubation, beads were washed and immunoprecipitates were eluted. DNA from eluates was recovered by the GeneJET FFPE DNA purification kit (Thermo Fisher Scientific, #K0882). DNA libraries were generated using the NEBNext Ultra DNA Library Prep kit (NEB, #E7370S) and sequenced at the St. Jude Hartwell Center.

CUT&RUN

Approximately 5 × 10 5 live cells were mixed with 5 × 10 4 Drosophila S2 cells per reaction. For CUT&RUN, we followed EpiCypher CUTANA protocol. In brief, we first isolated nuclei by incubating cells on ice for 5 min in Buffer A with protease inhibitor and 0.1% Triton X-100. After centrifugation at 1,750 × g for 2 min at 4°C, nuclei were resuspended in Wash buffer. Bio-Mag Plus Concanavalin-A (Con A) coated beads (Bangs Laboratories BP531) activated in Binding buffer were then added to the nuclei and rotated for 10 min at room temperature. About 1 μg primary antibody with 0.25 μg Spike-in antibody (Active Motif 61686) diluted in Antibody buffer was added to the bead-nuclei mixture and incubated for 2 h at room temperature. Beads were washed twice with Digitonin buffer and incubated with pAG-MNase for 10 min at room temperature. Beads were then washed twice with Digitonin buffer, incubated with 2 mM CaCl 2 for 2 h at 4°C, and quenched by adding Stop buffer. DNA was released from the beads by incubating them for 10 min at 37°C and purified by CUTANA DNA purification kit (EpiCypher SKU:14–0050). Libraries were constructed using xGen ssDNA and Low-Input DNA Prep by following the manufacturer’s instructions (IDT 10009817) and sequenced at the St. Jude Hartwell Center.

Analysis of chromatin immunoprecipitation-sequencing and CUT&RUN

Approximately 50 bp single-end reads were obtained and aligned to human genome hg38 by BWA (version 0.7.170.7.12, default parameter). Duplicated reads were marked by the bamsormadup from the biobambam tool (version 2.0.87) available at https://www.sanger.ac.uk/tool/biobambam/ . Uniquely mapped reads were kept by samtools (parameter “-q 1 -F 1804,” version 1.14). Fragments <2,000 bp were kept for peak calling, and bigwig files were generated for visualization. SICER [ 27 ] and macs2 [ 28 ] were both used for peak calling to identify both the narrow and broad peak correctly. With SICER, we assigned peaks that were at the top 1 percentile as the high-confidence peaks and the top 5 percentile as the low-confidence peaks. Two sets of peaks were generated: Strong peaks called with parameter “FDR < 0.05” by at least 1 method (macs2 or SICER) and weak peaks called with parameter “FDR < 0.5” by at least 1 method (macs2 or SICER). Peaks were considered reproducible if they were supported by 1 strong peaks and at least 1 weak peak in other replicates. For downstream analyses, heatmaps were generated by deepTools [ 52 ], and gene ontology was performed with Enrichr [ 53 , 54 ] and GSEA, in addition to custom R scripts. For differential peak analysis, peaks from 2 replicates were merged and counted for number of overlapping extended reads for each sample (bedtools v2.24.0) [ 55 ]. Then, we detected the differential peaks by the empirical Bayes method (eBayes function from the limma R package) [ 44 ]. For downstream analyses, heatmaps were generated by deepTools (v3.5.0) [ 56 ]. Peaks were annotated based on Gencode following this priority: “Promoter.Up”: if they fall within TSS– 2 kb, “Promoter.Down”: if they fall within TSS– 2 kb, “Exonic” or “intronic”: if they fall within an exon or intron of any isoform, “TES peaks”: if they fall within TES ± 2 kb, “distal5” or “distal3” if they are with 50 kb upstream of TSS or 50 kb downstream of TES, respectively, and they are classified as “intergenic” if they do not fit in any of the previous categories.

Supporting information

S1 data. numerical data used to generate summary data in this study..

https://doi.org/10.1371/journal.pbio.3002760.s001

S1 Raw Images. Uncropped western blot images in this study.

https://doi.org/10.1371/journal.pbio.3002760.s002

S1 Fig. Characterization of 53BP1-pS25 and NPCs and genome editing of hESCs.

( A ) Schematic diagram of neural differentiation of hESCs: neural induction, differentiation, and maturation media to form EBs, rosettes, NPCs, and neurons. ( B ) Principal component analysis of WT ESCs, NPCs, day 10 (D10) cortical organoids, and D17 cortical organoids. GSEA terms that are highly enriched in significantly ( C ) down-regulated and ( D ) up-regulated genes in WT NPCs compared to ESCs. % Match, % of genes in the enriched term that overlap the differentially expressed genes or proteins. ( E ) Immunofluorescence of NPC markers PAX6 and NESTIN. Bar, 50 μm. ( F ) Quantification of 53BP1-pS25-positive hESCs or hNPCs. Data are presented as the mean ± SEM, with p < 0.0001. ( G ) WB analysis of control cells and 53BP1-KO clones 415, 416, and 209, which are clones KO1, KO2, and KO3 in Yang and colleagues’ study [ 6 ]. (H) WB analysis of control and 53BP1-S25A hNPCs. The S25A mutation prohibits phosphorylation. ( I ) WB analysis of hESCs and hNPCs and quantification. ( J ) Schematic diagram of genome editing in hESCs. Guide RNA 6 were complexed with Cas9 proteins and used along single-stranded nucleotide donors to transfect hESCs. Individual clones from transfection were cultured, sequenced by mi-seq across the targeted 53BP1 locus, and established as >99% pure clonal lines. Diagram was generated using open-sourced images available at biorender.com . Underlying numerical values for figures are found in S1 Data . EB, embyoid body; ESC, embryonic stem cell; GSEA, gene set enrichment analysis; hESC, human embryonic stem cell; hNPC, human neural progenitor cell; KO, knockout; NES, normalized enrichment score; NPC, neural progenitor cell; WB, western blot; WT, wild type; 53BP1-pS25, 53BP1 phosphorylated at serine 25.

https://doi.org/10.1371/journal.pbio.3002760.s003

S2 Fig. Generation and analyses of ATM-KO hESCs and cortical organoids.

( A ) Alignment of WT and ATM -KO mutation sequences on 2 alleles (al) in the ATM locus. Red indicates the gRNA sequence. ( B ) WB analysis of WT and 4 ATM -KO hNPCs. ( C ) Principal component analysis showed the intermixing and similar RNA-seq profiles from hESCs of 7 WT, 4 53BP1-S25A, 4 53BP1-S25D, 4 ATM-KO, and 4 53BP1-KO lines. ( D ) Immunofluorescence showed similar expression of OCT4 and SSEA4 proteins in control and ATM -KO hESCs. Bar, 100 μm. ( E ) WB analysis of WT and 2 ATM -KO hNPCs. Quantification suggests reduction of γH2AX in ATM -KO hNPCs. Welch’s t test was used to perform pairwise comparisons of WT and ATM -KO. Underlying numerical values for figures are found in S1 Data . ATM, ataxia telangiectasia mutated; hESC, human embryonic stem cell; hNPC, human neural progenitor cell; KO, knockout; WB, western blot; WT, wild type.

https://doi.org/10.1371/journal.pbio.3002760.s004

S3 Fig. Analysis of γH2AX and CC3 in cortical organoids.

( A ) Immunofluorescence showed D35 ATM -KO and WT cortical organoids had similar γH2AX foci. Bar, 100 μm. FACS analysis of CC3 in ( B ) D21 and ( C ) D28 cortical organoids. Two biological replicates were done, and each data point was based on 3 technical replicate analyses of 10–12 cortical organoids. ( D , E ) Immunofluorescence and quantification of CC3 in D28 cortical organoids. Bar, 100 μm. Graphs are presented in ratios (out of 1), with **, p < 0.01; ****, p < 0.0001; ns, not significant by two-way ANOVA test. Underlying numerical values for figures are found in S1 Data . ATM, ataxia telangiectasia mutated; CC3, cleaved-caspase 3; KO, knockout; WT, wild type.

https://doi.org/10.1371/journal.pbio.3002760.s005

S4 Fig. Analysis of cell proliferation in cortical organoids.

FACS analysis of PAX6 and KI67 in ( A , B ) D28 and ( C ) D35 cortical organoids. Each data point was based on the 3 technical replicate analyses of 10 cortical organoids. ( D ) Quantification of KI67/PAX6 ratios in immunofluorescence of D35 cortical organoids. Each data point represents quantification of cells in 1 cortical organoid. ( E , F ) Immunofluorescence and quantification of H3-pS10 (PH3) in D28 cortical organoids. Bar, 100 μm. ( F - H ) Immunofluorescence and quantification of PH3 and KI67 in D35 cortical organoids. Bar, 100 μm. ***, p < 0.001; ns, not significant by two-way ANOVA test. Underlying numerical values for figures are found in S1 Data .

https://doi.org/10.1371/journal.pbio.3002760.s006

S5 Fig. Immunofluorescence analyses of cortical organoids and NPCs.

Immunofluorescence of ( A ) NEUN and ( E ) ZO-1 and PAX6 in D37 cortical organoids. Bar, 100 μm. ( B ) Immunofluorescence of ZO-1 in D28 cortical organoids. Bar, 100 μm. Quantification of the ( C ) number and ( D ) surface area of ZO-1-positive ventricles in D28 cortical organoids. *, p < 0.05; ***, p <0 .001; ns, not significant by two-way ANOVA test. ( F ) Bright-field images of cortical organoids formed by ATM -KO2, 3, 14, 43, and WT control at day 55 of differentiation. Bar, 1.5 mm. ( G ) The size of cortical organoids was compared between groups by one-way ANOVA with Dunnett’s multiple comparisons test, with ns, not significant and ***, p < 0.001. n = 13 organoids/group. Underlying numerical values for figures are found in S1 Data . ATM, ataxia telangiectasia mutated; NPC, neural progenitor cell; WT, wild type.

https://doi.org/10.1371/journal.pbio.3002760.s007

S6 Fig. Characterization of NPCs and D35 cortical organoids.

( A ) Immunofluorescence of PAX6 and NES in NPCs. Bar, 50 μm. ( B ) Principal component analysis of proteomics data of D35 WT and ATM -KO cortical organoids. GSEA terms that are highly enriched in significantly ( C ) higher and ( D ) lower total proteins in D35 ATM -KO versus WT cortical organoids. ( E ) GSEA terms that are highly enriched in significantly higher phosphoproteins, which were normalized to total proteomics, in D35 ATM -KO versus WT cortical organoids. Underlying numerical values for figures are found in S1 Data . ATM, ataxia telangiectasia mutated; GSEA, gene set enrichment analysis; KO, knockout; NES, normalized enrichment score; NPC, neural progenitor cell; WT, wild type.

https://doi.org/10.1371/journal.pbio.3002760.s008

S7 Fig. Kinase activities in cortical organoids and characterization of the 53BP1-S25A and 53BP1-S25D hESCs.

Heatmaps showing relative phosphorylation levels of ( A ) 7 MAPK9 substrates that are significantly lower and ( B ) 7 CDK5 substrates that are significantly higher in D35 ATM -KO versus WT cortical organoids. ( C ) Heatmaps showing activity of selected protein kinases between ATM-KO3, ATM-KO4, and WT cell lines. ( D ) Alignment of WT and 53BP1-S25A and S25D mutation sequences on 2 alleles (al). Red indicates the gRNA sequence. Underline indicates codon encoding the WT serine 25, mutant alanine, or mutant aspartic acid. ( E ) WB analysis of control and 53BP1-S25D hNPCs, which have comparable levels of 53BP1 protein. ( F ) Transcripts per million values of 10 pluripotency genes were used for comparison to show that control, 53BP1-S25A, and 53BP1-S25D hESCs did not differ in pluripotency. Underlying numerical values for figures are found in S1 Data . ATM, ataxia telangiectasia mutated; hESC, human embryonic stem cell; hNPC, human neural progenitor cell; KO, knockout; WB, western blot; WT, wild type; 53BP1, p53 binding protein 1.

https://doi.org/10.1371/journal.pbio.3002760.s009

S8 Fig. Characterization of the 53BP1-S25A and 53BP1-S25D hESCs and cortical organoids.

( A ) Immunofluorescence showed similar expression of OCT4 and SSEA4 proteins in WT, 53BP1-S25A, and 53BP1-S25D hESCs. Bar, 100 μm. Immunofluorescence of ( B ) KI67 and ( D ) PH3 in cryosections of cortical organoids at day 35 of differentiation. Bar, 100 μm. ( C ) Quantification of KI67-positive cells in D35 cortical organoids. Data points represent single organoids. The mean ± SEM values were compared by one-way ANOVA with Dunnett’s multiple comparisons test to yield ****, ***, and ** indicating p < 0.0001, 0.001, and 0.01, respectively. n = 3 organoids/group. Underlying numerical values for figures are found in S1 Data .

https://doi.org/10.1371/journal.pbio.3002760.s010

S9 Fig. Analysis of γH2AX and CC3 in cortical organoids.

Immunofluorescence of ( A ) γH2AX in D35 cortical organoids and ( D ) CC3 in D28 cortical organoids. Bar, 100 μm. CC3 quantification by FACS of ( B ) D21 and ( C ) D28 cortical organoids. For each datapoint, 10–12 organoids from each line were analyzed via 3 technical replicates, and data from 4 mutant lines were consolidated to achieve rigorous comparisons. **, p < 0.01 and ns, not significant by two-way ANOVA test. ( E ) CC3 quantification of immunofluorescence images of D28 cortical organoids. For each line, 4–6 images and >10,000 cells were analyzed. *, p < 0.05; **, p < 0.01; ns, not significant by two-way ANOVA test. Graphs in ( B , C , E ) are presented in ratios (out of 1). Underlying numerical values for figures are found in S1 Data .

https://doi.org/10.1371/journal.pbio.3002760.s011

S10 Fig. 53BP1-pS25 promotes ventricle formation in cortical organoids.

Quantification of the ( A ) number and ( B ) surface area of ZO-1-positive ventricles. ( C ) Immunofluorescence of ZO-1 in D28 cortical organoids. Bar, 100 μm. *, p < 0.05; **, p < 0.01; ***, p < 0.001; ns, not significant by two-way ANOVA test. Underlying numerical values for figures are found in S1 Data .

https://doi.org/10.1371/journal.pbio.3002760.s012

S11 Fig. Characterization of 53BP1-S25A and 53BP1-S25D cortical organoids.

( A ) Bright-field images of cortical organoids formed by cell lines 53BP1-S25A 34–3, 34–4, 79–1, 79–3 and S25D 14–3, 14–15, 14–19, 17, and 2 WT control at day 55 of differentiation. Bar, 1.5 mm. Blue transparent structures around organoids are Matrigel embedment. ( B ) At day 55 of differentiation, the size of cortical organoids was compared between groups. ( C ) The growth (comparing organoids at days 35 and 55) of cortical organoids were compared between groups. Data points represent single organoids. The mean ± SEM values were compared by one-way ANOVA with Dunnett’s multiple comparisons test to yield **** and ** indicating p < 0.0001 and 0.01, respectively. n = 15–36 organoids/group. ( D ) Two genes overlapped between up-regulated genes in 53BP1-S25A versus WT and down-regulated genes in 53BP1-S25D versus WT cortical organoids. No gene overlapped between down-regulated genes in 53BP1-S25A versus WT and up-regulated genes in 53BP1-S25D versus WT cortical organoids. ( E ) Down-regulated GSEA terms between 53BP1-S25A versus WT and 53BP1-S25D versus WT were not highly overlapped. Ten GSEA terms were specific to 53BP1-S25D versus WT. Underlying numerical values for figures are found in S1 Data .

https://doi.org/10.1371/journal.pbio.3002760.s013

S12 Fig. Characterization of NPCs and comparative analyses of RNA-seq data.

( A ) Immunofluorescence of NPC markers PAX6 and NESTIN. Bar, 50 μm. GSEA identified top enrichment of differentially expressed genes in ( B , D ) 53BP1-S25A or ( C , E ) S25D versus WT NPCs. % Match, % of genes in the enriched term that overlap the differentially expressed genes or proteins. Venn diagrams depict overlaps between down-regulated genes in ATM -KO with 53BP1- ( F ) S25A or ( G ) S25D cortical organoids. Underlying numerical values for figures are found in S1 Data . ATM, ataxia telangiectasia mutated; GSEA, gene set enrichment analysis; KO, knockout; NES, normalized enrichment score; NPC, neural progenitor cell; WT, wild type.

https://doi.org/10.1371/journal.pbio.3002760.s014

S13 Fig. Comparisons of RNA-seq data and 53BP1 ChIP-seq analyses.

( A ) GSEA graphs showed that up-regulated genes in 53BP1-S25A or S25D vs. WT had significant enrichment in down-regulated genes of ATM-KO vs. WT cortical organoids. P values were calculated by the hypergeometric test, assuming normal data distribution. ( B ) Concordantly differential expression of genes in 53BP1-S25D vs. WT were enriched in those in 53BP1-S25A vs. WT. ( C ) Concordantly differential expression of genes in 53BP1-S25A vs. WT were enriched in those in 53BP1-S25D vs. WT. For ( A - C ), P values were calculated by the hypergeometric test, assuming normal data distribution. ( D ) Proportions of 53BP1 binding to genomic features. 53BP1 ChIP-seq tracks at loci of representative ( E ) up-regulated and ( F ) down-regulated genes in 53BP1-S25A and S25D versus WT D35 cortical organoids. ( G ) S25A and S25D down-regulate 53BP1 targets that are enriched in IRE1-mediated unfolded protein response, regulation of cellular response to stress, iron import into cells, and regulation of apoptosis. Underlying numerical values for figures are found in S1 Data . ATM, ataxia telangiectasia mutated; GSEA, gene set enrichment analysis; KO, knockout; WT, wild type.

https://doi.org/10.1371/journal.pbio.3002760.s015

S14 Fig. 53BP1 ChIP-seq and 53BP1-pS25 CUT&RUN.

( A ) MA plot displays 53BP1 ChIP-seq signals at genomic sites that are significantly different in ATM -KO vs. WT NPCs. Proportions of genomic features and gene ontology of genes with ( B ) higher or ( C ) lower 53BP1 binding in ATM -KO vs. WT NPCs. ( D ) Proportions of 53BP1-pS25 binding to genomic features. ( E ) GSEA identified top enrichment of genes occupied by 53BP1-pS25 in WT NPCs. % Match, % of genes in the enriched term that overlap the differentially expressed genes or proteins. Underlying numerical values for figures are found in S1 Data . ATM, ataxia telangiectasia mutated; GSEA, gene set enrichment analysis; KO, knockout; NES, normalized enrichment score; NPC, neural progenitor cell; WT, wild type.

https://doi.org/10.1371/journal.pbio.3002760.s016

S15 Fig. Differential 53BP1 ChIP-seq in 53BP1-WT, S25A, and S25D NPCs.

( A ) Principal component analysis of top 3,000 most variable peaks in 53BP1 ChIP-seq of 53BP1-WT, S25A, and S25D NPCs. Two independent cell lines for each group were used for ChIP-seq. Proportions of genomic features in regions with significantly different 53BP1 ChIP-seq in ( B ) 53BP1-S25A vs. WT and ( C ) 53BP1-S25D vs. WT, using the criterion of FC>2 and p < 0.05. ( D ) Heatmaps aligning peaks with significantly different 53BP1 ChIP-seq in 53BP1-S25A vs. S25D. Control regions are those, after voom normalization, showed the least changes and served as semi-independent validation of differential ChIP-seq analysis. Bubble graphs present top enriched categories of genes that had significantly lower 53BP1 ChIP-seq in ( E ) 53BP1-S25A vs. WT and ( F ) 53BP1-S25D vs. WT. Underlying numerical values for figures are found in S1 Data . FC, fold-change; NPC, neural progenitor cell; WT, wild type.

https://doi.org/10.1371/journal.pbio.3002760.s017

S16 Fig. Analysis of ATM activities during the inhibition of TGFβ, WNT, and HH signaling.

( A ) Schematic diagram of neural specification of hESCs with HH (SB421542), TGFβ (dorsomorphin), and WNT (IWR1e and cyclopamine) signaling inhibitors. Nuclear extract was harvested on day 4 and day 10. ( B ) WB analysis of day 4 samples. ( C ) Quantification of day 4 WB. Data are presented as the mean ± SEM, and Student t test was performed for pairwise comparisons. n.s., *, and ** indicate not significant, p < 0.05, and p < 0.01, respectively. ( D ) WB analysis of day 10 samples. ( E ) Quantification of day 10 WB. Data are presented as the mean ± SEM, and Student t test was performed for pairwise comparisons. n.s., *, and ** indicate not significant, p < 0.05, and p < 0.01, respectively. Underlying numerical values for figures are found in S1 Data . ATM, ataxia telangiectasia mutated; hESC, human embryonic stem cell; WB, western blot.

https://doi.org/10.1371/journal.pbio.3002760.s018

S17 Fig. RNF168 -KO alters key genetic programs and 53BP1-pS25 binding to chromatin.

( A ) Alignment of WT and RNF168 -KO mutation sequences in the RNF168 locus. Red indicates the gRNA sequences. ( B ) WB analysis of WT and RNF168 -KO hESCs. ( C ) RT-qPCR analysis showing that pluripotent genes in RNF168 -KO were expressed higher or the same as those in WT. RNF168 -KO did not reduce pluripotent gene expression. *, p < 0.05; ns, not significant by two-way ANOVA text. ( D ) Immunofluorescence showed similar expression of OCT4 and SSEA4 proteins in WT and RNF168 -KO hESCs. Bar, 100 μm. ( E ) Immunofluorescence of showed similar expression of PAX6 and NES in NPCs. WT and RNF168 -KO NPCs. Bar, 50 μm. Functional terms that are highly enriched in ( F ) up-regulated and ( G ) down-regulated genes in RNF168 -KO D35 cortical organoids. % Match, % of genes in the enriched term that overlap the differentially expressed genes or proteins. ( H ) Heatmaps aligning peaks with 53BP1-pS25 CUT&RUN signals that were gained, the same, or lost in RNF168 -KO vs. WT NPCs, using the criterion of FC>2 and p < 0.05. n = numbers of peaks. Regions with the same signals, are n = 899, which showed the least changes after voom normalization and served as semi-independent validation of differential ChIP-seq analysis. ( I ) Functional terms of 53BP1-pS25-bound genes in WT NPCs. % Match, % of genes in the enriched term that overlap the differentially bound genes. ( J ) Number of differentially expressed genes identified by comparison of RNF168 -KO vs. WT NPCs at p < 0.05. Of these genes, we list the numbers of 53BP1-pS25-bound targets and targets with higher or lower 53BP1-pS25 CUT&RUN signals in RNF168 -KO NPCs. Underlying numerical values for figures are found in S1 Data . FC, fold-change; hESC, human embryonic stem cell; KO, knockout; NES, normalized enrichment score; NPC, neural progenitor cell; RT-qPCR, quantitative reverse transcription PCR; WB, western blot; WT, wild type; 53BP1-pS25, 53BP1 phosphorylated at serine 25.

https://doi.org/10.1371/journal.pbio.3002760.s019

S1 Table. All cell lines generated for this study were treated with trypsin and Wright’s stain and then analyzed by the Cytogenetic Shared Resource at St. Jude.

Typically normal karyotypes and 3 abnormalities are shown.

https://doi.org/10.1371/journal.pbio.3002760.s020

S2 Table. The expression of forebrain, midbrain, and hindbrain markers in D35 WT and ATM -KO cortical organoids.

Data suggest that D35 ATM -KO cortical organoids specified to the forebrain lineage.

https://doi.org/10.1371/journal.pbio.3002760.s021

S3 Table. List of phosphoproteins, normalized to total protein levels, that were significantly lower in D35 ATM -KO versus WT cortical organoids.

https://doi.org/10.1371/journal.pbio.3002760.s022

S4 Table. Normalized (to total proteome) levels of phosphopeptide substrates of MAPK9 and CDK5 in D35 WT and ATM -KO cortical organoids.

https://doi.org/10.1371/journal.pbio.3002760.s023

S5 Table. Two-sample t test examines the sizes of cortical organoids that change between day 35 and day 55 of differentiation.

Data from WT and 53BP1 mutants are compared pairwise by using the two-sample t test. The sizes of organoids are significantly different between each comparison pair (all p < 0.05).

https://doi.org/10.1371/journal.pbio.3002760.s024

S6 Table. The changes in organoid size at days 35 and 55 of differentiation were compared to yield S3C Fig .

This table lists the calculation for different combinations of data and the descriptive statistics.

https://doi.org/10.1371/journal.pbio.3002760.s025

S7 Table. Primary antibodies used in this study.

https://doi.org/10.1371/journal.pbio.3002760.s026

S8 Table. gRNA sequences used in this study.

https://doi.org/10.1371/journal.pbio.3002760.s027

Acknowledgments

The authors thank A. Andersen and I. Chen for discussions and editing the manuscript; A. N. Kettenbach for advice; J. Houston and K. Lowe for FACS; P. Sinojia and E. Rivera-Peraza for preliminary experiments and data analyses. Sequencing was performed at the Harwell Center for Biotechnology, images were acquired at the Cell & Tissue Imaging Center, and karyotyping was analyzed by J. Wilbourne and V. Valentine at the Cytogenetics Core; all are supported by SJCRH and NCI P30 (CA021765).

• View Article
• PubMed/NCBI
• Google Scholar

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Published: 04 September 2024

CDK5–cyclin B1 regulates mitotic fidelity

• Xiao-Feng Zheng   ORCID: orcid.org/0000-0001-8769-4604 1   na1 ,
• Aniruddha Sarkar   ORCID: orcid.org/0000-0002-9393-1335 1   na1 ,
• Humphrey Lotana 2 ,
• Aleem Syed   ORCID: orcid.org/0000-0001-7942-3900 1 ,
• Huy Nguyen   ORCID: orcid.org/0000-0002-4424-1047 1 ,
• Richard G. Ivey 3 ,
• Jacob J. Kennedy 3 ,
• Jeffrey R. Whiteaker 3 ,
• Bartłomiej Tomasik   ORCID: orcid.org/0000-0001-5648-345X 1 , 4   nAff7 ,
• Kaimeng Huang   ORCID: orcid.org/0000-0002-0552-209X 1 , 5 ,
• Feng Li 1 ,
• Alan D. D’Andrea   ORCID: orcid.org/0000-0001-6168-6294 1 , 5 ,
• Amanda G. Paulovich   ORCID: orcid.org/0000-0001-6532-6499 3 ,
• Kavita Shah 2 ,
• Alexander Spektor   ORCID: orcid.org/0000-0002-1085-3205 1 , 5 &
• Dipanjan Chowdhury   ORCID: orcid.org/0000-0001-5645-3752 1 , 5 , 6  

Nature ( 2024 ) Cite this article

48 Altmetric

Metrics details

CDK1 has been known to be the sole cyclin-dependent kinase (CDK) partner of cyclin B1 to drive mitotic progression 1 . Here we demonstrate that CDK5 is active during mitosis and is necessary for maintaining mitotic fidelity. CDK5 is an atypical CDK owing to its high expression in post-mitotic neurons and activation by non-cyclin proteins p35 and p39 2 . Here, using independent chemical genetic approaches, we specifically abrogated CDK5 activity during mitosis, and observed mitotic defects, nuclear atypia and substantial alterations in the mitotic phosphoproteome. Notably, cyclin B1 is a mitotic co-factor of CDK5. Computational modelling, comparison with experimentally derived structures of CDK–cyclin complexes and validation with mutational analysis indicate that CDK5–cyclin B1 can form a functional complex. Disruption of the CDK5–cyclin B1 complex phenocopies CDK5 abrogation in mitosis. Together, our results demonstrate that cyclin B1 partners with both CDK5 and CDK1, and CDK5–cyclin B1 functions as a canonical CDK–cyclin complex to ensure mitotic fidelity.

This is a preview of subscription content, access via your institution

Access options

Access Nature and 54 other Nature Portfolio journals

Get Nature+, our best-value online-access subscription

24,99 € / 30 days

cancel any time

Subscribe to this journal

Receive 51 print issues and online access

185,98 € per year

only 3,65 € per issue

Buy this article

• Purchase on SpringerLink
• Instant access to full article PDF

Prices may be subject to local taxes which are calculated during checkout

Similar content being viewed by others

Core control principles of the eukaryotic cell cycle

CDC7-independent G1/S transition revealed by targeted protein degradation

Evolution of opposing regulatory interactions underlies the emergence of eukaryotic cell cycle checkpoints

Data availability.

All data supporting the findings of this study are available in the Article and its Supplementary Information . The LC–MS/MS proteomics data have been deposited to the ProteomeXchange Consortium 60 via the PRIDE 61 partner repository under dataset identifier PXD038386 . Correspondence regarding experiments and requests for materials should be addressed to the corresponding authors.

Wieser, S. & Pines, J. The biochemistry of mitosis. Cold Spring Harb. Perspect. Biol. 7 , a015776 (2015).

Article   PubMed   PubMed Central   Google Scholar  

Dhavan, R. & Tsai, L. H. A decade of CDK5. Nat. Rev. Mol. Cell Biol. 2 , 749–759 (2001).

Article   CAS   PubMed   Google Scholar  

Malumbres, M. Cyclin-dependent kinases. Genome Biol. 15 , 122 (2014).

Coverley, D., Laman, H. & Laskey, R. A. Distinct roles for cyclins E and A during DNA replication complex assembly and activation. Nat. Cell Biol. 4 , 523–528 (2002).

Desai, D., Wessling, H. C., Fisher, R. P. & Morgan, D. O. Effects of phosphorylation by CAK on cyclin binding by CDC2 and CDK2. Mol. Cell. Biol. 15 , 345–350 (1995).

Article   CAS   PubMed   PubMed Central   Google Scholar  

Brown, N. R. et al. CDK1 structures reveal conserved and unique features of the essential cell cycle CDK. Nat. Commun. 6 , 6769 (2015).

Article   ADS   CAS   PubMed   PubMed Central   Google Scholar  

Strauss, B. et al. Cyclin B1 is essential for mitosis in mouse embryos, and its nuclear export sets the time for mitosis. J. Cell Biol. 217 , 179–193 (2018).

Gavet, O. & Pines, J. Activation of cyclin B1-Cdk1 synchronizes events in the nucleus and the cytoplasm at mitosis. J. Cell Biol. 189 , 247–259 (2010).

Barbiero, M. et al. Cell cycle-dependent binding between cyclin B1 and Cdk1 revealed by time-resolved fluorescence correlation spectroscopy. Open Biol. 12 , 220057 (2022).

Pines, J. & Hunter, T. Isolation of a human cyclin cDNA: evidence for cyclin mRNA and protein regulation in the cell cycle and for interaction with p34cdc2. Cell 58 , 833–846 (1989).

Clute, P. & Pines, J. Temporal and spatial control of cyclin B1 destruction in metaphase. Nat. Cell Biol. 1 , 82–87 (1999).

Potapova, T. A. et al. The reversibility of mitotic exit in vertebrate cells. Nature 440 , 954–958 (2006).

Basu, S., Greenwood, J., Jones, A. W. & Nurse, P. Core control principles of the eukaryotic cell cycle. Nature 607 , 381–386 (2022).

Santamaria, D. et al. Cdk1 is sufficient to drive the mammalian cell cycle. Nature 448 , 811–815 (2007).

Article   ADS   CAS   PubMed   Google Scholar  

Zheng, X. F. et al. A mitotic CDK5-PP4 phospho-signaling cascade primes 53BP1 for DNA repair in G1. Nat. Commun. 10 , 4252 (2019).

Article   ADS   PubMed   PubMed Central   Google Scholar  

Fagerberg, L. et al. Analysis of the human tissue-specific expression by genome-wide integration of transcriptomics and antibody-based proteomics. Mol. Cell. Proteom. 13 , 397–406 (2014).

Article   CAS   Google Scholar  

Pozo, K. & Bibb, J. A. The emerging role of Cdk5 in cancer. Trends Cancer 2 , 606–618 (2016).

Sharma, S. & Sicinski, P. A kinase of many talents: non-neuronal functions of CDK5 in development and disease. Open Biol. 10 , 190287 (2020).

Sun, K. H. et al. Novel genetic tools reveal Cdk5’s major role in Golgi fragmentation in Alzheimer’s disease. Mol. Biol. Cell 19 , 3052–3069 (2008).

Sharma, S. et al. Targeting the cyclin-dependent kinase 5 in metastatic melanoma. Proc. Natl Acad. Sci. USA 117 , 8001–8012 (2020).

Nabet, B. et al. The dTAG system for immediate and target-specific protein degradation. Nat. Chem. Biol. 14 , 431–441 (2018).

Simpson, L. M. et al. Target protein localization and its impact on PROTAC-mediated degradation. Cell Chem. Biol. 29 , 1482–1504 e1487 (2022).

Vassilev, L. T. et al. Selective small-molecule inhibitor reveals critical mitotic functions of human CDK1. Proc. Natl Acad. Sci. USA 103 , 10660–10665 (2006).

Janssen, A. F. J., Breusegem, S. Y. & Larrieu, D. Current methods and pipelines for image-based quantitation of nuclear shape and nuclear envelope abnormalities. Cells 11 , 347 (2022).

Thompson, S. L. & Compton, D. A. Chromosome missegregation in human cells arises through specific types of kinetochore-microtubule attachment errors. Proc. Natl Acad. Sci. USA 108 , 17974–17978 (2011).

Kline-Smith, S. L. & Walczak, C. E. Mitotic spindle assembly and chromosome segregation: refocusing on microtubule dynamics. Mol. Cell 15 , 317–327 (2004).

Prosser, S. L. & Pelletier, L. Mitotic spindle assembly in animal cells: a fine balancing act. Nat. Rev. Mol. Cell Biol. 18 , 187–201 (2017).

Zeng, X. et al. Pharmacologic inhibition of the anaphase-promoting complex induces a spindle checkpoint-dependent mitotic arrest in the absence of spindle damage. Cancer Cell 18 , 382–395 (2010).

Warren, J. D., Orr, B. & Compton, D. A. A comparative analysis of methods to measure kinetochore-microtubule attachment stability. Methods Cell. Biol. 158 , 91–116 (2020).

Gregan, J., Polakova, S., Zhang, L., Tolic-Norrelykke, I. M. & Cimini, D. Merotelic kinetochore attachment: causes and effects. Trends Cell Biol 21 , 374–381 (2011).

Etemad, B., Kuijt, T. E. & Kops, G. J. Kinetochore-microtubule attachment is sufficient to satisfy the human spindle assembly checkpoint. Nat. Commun. 6 , 8987 (2015).

Tauchman, E. C., Boehm, F. J. & DeLuca, J. G. Stable kinetochore-microtubule attachment is sufficient to silence the spindle assembly checkpoint in human cells. Nat. Commun. 6 , 10036 (2015).

Mitchison, T. & Kirschner, M. Microtubule assembly nucleated by isolated centrosomes. Nature 312 , 232–237 (1984).

Fourest-Lieuvin, A. et al. Microtubule regulation in mitosis: tubulin phosphorylation by the cyclin-dependent kinase Cdk1. Mol. Biol. Cell 17 , 1041–1050 (2006).

Ubersax, J. A. et al. Targets of the cyclin-dependent kinase Cdk1. Nature 425 , 859–864 (2003).

Yang, C. H., Lambie, E. J. & Snyder, M. NuMA: an unusually long coiled-coil related protein in the mammalian nucleus. J. Cell Biol. 116 , 1303–1317 (1992).

Yang, C. H. & Snyder, M. The nuclear-mitotic apparatus protein is important in the establishment and maintenance of the bipolar mitotic spindle apparatus. Mol. Biol. Cell 3 , 1259–1267 (1992).

Kotak, S., Busso, C. & Gonczy, P. NuMA phosphorylation by CDK1 couples mitotic progression with cortical dynein function. EMBO J. 32 , 2517–2529 (2013).

Kitagawa, M. et al. Cdk1 coordinates timely activation of MKlp2 kinesin with relocation of the chromosome passenger complex for cytokinesis. Cell Rep. 7 , 166–179 (2014).

Schrock, M. S. et al. MKLP2 functions in early mitosis to ensure proper chromosome congression. J. Cell Sci. 135 , jcs259560 (2022).

Sun, M. et al. NuMA regulates mitotic spindle assembly, structural dynamics and function via phase separation. Nat. Commun. 12 , 7157 (2021).

Chen, Q., Zhang, X., Jiang, Q., Clarke, P. R. & Zhang, C. Cyclin B1 is localized to unattached kinetochores and contributes to efficient microtubule attachment and proper chromosome alignment during mitosis. Cell Res. 18 , 268–280 (2008).

Kabeche, L. & Compton, D. A. Cyclin A regulates kinetochore microtubules to promote faithful chromosome segregation. Nature 502 , 110–113 (2013).

Hegarat, N. et al. Cyclin A triggers mitosis either via the Greatwall kinase pathway or cyclin B. EMBO J. 39 , e104419 (2020).

Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596 , 583–589 (2021).

Wood, D. J. & Endicott, J. A. Structural insights into the functional diversity of the CDK-cyclin family. Open Biol. 8 , 180112 (2018).

Brown, N. R., Noble, M. E., Endicott, J. A. & Johnson, L. N. The structural basis for specificity of substrate and recruitment peptides for cyclin-dependent kinases. Nat. Cell Biol. 1 , 438–443 (1999).

Tarricone, C. et al. Structure and regulation of the CDK5-p25 nck5a complex. Mol. Cell 8 , 657–669 (2001).

Poon, R. Y., Lew, J. & Hunter, T. Identification of functional domains in the neuronal Cdk5 activator protein. J. Biol. Chem. 272 , 5703–5708 (1997).

Oppermann, F. S. et al. Large-scale proteomics analysis of the human kinome. Mol. Cell. Proteom. 8 , 1751–1764 (2009).

van den Heuvel, S. & Harlow, E. Distinct roles for cyclin-dependent kinases in cell cycle control. Science 262 , 2050–2054 (1993).

Article   ADS   PubMed   Google Scholar  

Nakatani, Y. & Ogryzko, V. Immunoaffinity purification of mammalian protein complexes. Methods Enzymol. 370 , 430–444 (2003).

Tyanova, S., Temu, T. & Cox, J. The MaxQuant computational platform for mass spectrometry-based shotgun proteomics. Nat. Protoc. 11 , 2301–2319 (2016).

Tyanova, S. et al. The Perseus computational platform for comprehensive analysis of (prote)omics data. Nat. Methods 13 , 731–740 (2016).

Ritchie, M. E. et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 43 , e47 (2015).

R Core Team. R: a language and environment for statistical computing (2021).

Wickham, H. ggplot2: elegant graphics for data analysis (2016).

Slowikowski, K. ggrepel: automatically position non-overlapping text labels with “ggplot2” (2018).

Wu, T. et al. clusterProfiler 4.0: a universal enrichment tool for interpreting omics data. Innovation 2 , 100141 (2021).

CAS   PubMed   PubMed Central   Google Scholar  

Deutsch, E. W. et al. The ProteomeXchange consortium in 2020: enabling ‘big data’ approaches in proteomics. Nucleic Acids Res. 48 , D1145–D1152 (2020).

CAS   PubMed   Google Scholar  

Perez-Riverol, Y. et al. The PRIDE database and related tools and resources in 2019: improving support for quantification data. Nucleic Acids Res. 47 , D442–D450 (2019).

Robinson, M. D., McCarthy, D. J. & Smyth, G. K. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26 , 139–140 (2010).

Nagahara, H. et al. Transduction of full-length TAT fusion proteins into mammalian cells: TAT-p27Kip1 induces cell migration. Nat. Med. 4 , 1449–1452 (1998).

Mirdita, M. et al. ColabFold: making protein folding accessible to all. Nat. Methods 19 , 679–682 (2022).

Lu, C. et al. OPLS4: improving force field accuracy on challenging regimes of chemical space. J. Chem. Theory Comput. 17 , 4291–4300 (2021).

Obenauer, J. C., Cantley, L. C. & Yaffe, M. B. Scansite 2.0: proteome-wide prediction of cell signaling interactions using short sequence motifs. Nucleic Acids Res. 31 , 3635–3641 (2003).

Download references

Acknowledgements

We thank D. Pellman for comments on the manuscript; W. Michowski, S. Sharma, P. Sicinski, B. Nabet and N. Gray for the reagents; J. A. Tainer for providing access to software used for structural analysis; and S. Gerber for sharing unpublished results. D.C. is supported by grants R01 CA208244 and R01 CA264900, DOD Ovarian Cancer Award W81XWH-15-0564/OC140632, Tina’s Wish Foundation, Detect Me If You Can, a V Foundation Award, a Gray Foundation grant and the Claudia Adams Barr Program in Innovative Basic Cancer Research. A. Spektor would like to acknowledge support from K08 CA208008, the Burroughs Wellcome Fund Career Award for Medical Scientists, Saverin Breast Cancer Research Fund and the Claudia Adams Barr Program in Innovative Basic Cancer Research. X.-F.Z. was an American Cancer Society Fellow and is supported by the Breast and Gynecologic Cancer Innovation Award from Susan F. Smith Center for Women’s Cancers at Dana-Farber Cancer Institute. A. Syed is supported by the Claudia Adams Barr Program in Innovative Basic Cancer Research. B.T. was supported by the Polish National Agency for Academic Exchange (grant PPN/WAL/2019/1/00018) and by the Foundation for Polish Science (START Program). A.D.D is supported by NIH grant R01 HL52725. A.G.P. by National Cancer Institute grants U01CA214114 and U01CA271407, as well as a donation from the Aven Foundation; J.R.W. by National Cancer Institute grant R50CA211499; and K.S. by NIH awards 1R01-CA237660 and 1RF1NS124779.

Author information

Bartłomiej Tomasik

Present address: Department of Oncology and Radiotherapy, Medical University of Gdańsk, Faculty of Medicine, Gdańsk, Poland

These authors contributed equally: Xiao-Feng Zheng, Aniruddha Sarkar

Authors and Affiliations

Division of Radiation and Genome Stability, Department of Radiation Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA

Xiao-Feng Zheng, Aniruddha Sarkar, Aleem Syed, Huy Nguyen, Bartłomiej Tomasik, Kaimeng Huang, Feng Li, Alan D. D’Andrea, Alexander Spektor & Dipanjan Chowdhury

Department of Chemistry and Purdue University Center for Cancer Research, Purdue University, West Lafayette, IN, USA

Humphrey Lotana & Kavita Shah

Translational Science and Therapeutics Division, Fred Hutchinson Cancer Research Center, Seattle, WA, USA

Richard G. Ivey, Jacob J. Kennedy, Jeffrey R. Whiteaker & Amanda G. Paulovich

Department of Biostatistics and Translational Medicine, Medical University of Łódź, Łódź, Poland

Broad Institute of Harvard and MIT, Cambridge, MA, USA

Kaimeng Huang, Alan D. D’Andrea, Alexander Spektor & Dipanjan Chowdhury

Department of Biological Chemistry & Molecular Pharmacology, Harvard Medical School, Boston, MA, USA

Dipanjan Chowdhury

You can also search for this author in PubMed   Google Scholar

Contributions

X.-F.Z., A. Sarkar., A. Spektor. and D.C. conceived the project and designed the experiments. X.-F.Z. and A. Sarkar performed the majority of experiments and associated analyses except as listed below. H.L. expressed relevant proteins and conducted the kinase activity assays for CDK5–cyclin B1, CDK5–p35 and CDK5(S46) variant complexes under the guidance of K.S.; A. Syed performed structural modelling and analysis. R.G.I., J.J.K. and J.R.W. performed MS and analysis. B.T. and H.N. performed MS data analyses. K.H. provided guidance to screen CDK5(as) knocked-in clones and performed sequence analysis to confirm CDK5(as) knock-in. F.L. and A.D.D. provided reagents and discussion on CDK5 substrates analyses. X.-F.Z., A. Sarkar, A. Spektor and D.C. wrote the manuscript with inputs and edits from all of the other authors.

Corresponding authors

Correspondence to Alexander Spektor or Dipanjan Chowdhury .

Ethics declarations

Competing interests.

A.D.D. reports consulting for AstraZeneca, Bayer AG, Blacksmith/Lightstone Ventures, Bristol Myers Squibb, Cyteir Therapeutics, EMD Serono, Impact Therapeutics, PrimeFour Therapeutics, Pfizer, Tango Therapeutics and Zentalis Pharmaceuticals/Zeno Management; is an advisory board member for Cyteir and Impact Therapeutics; a stockholder in Cedilla Therapeutics, Cyteir, Impact Therapeutics and PrimeFour Therapeutics; and reports receiving commercial research grants from Bristol Myers Squibb, EMD Serono, Moderna and Tango Therapeutics. The other authors declare no competing interests.

Peer review

Peer review information.

Nature thanks Yibing Shan and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended data fig. 1 inhibition of cdk5 in analogue-sensitive (cdk5- as ) system..

a , Schematics depicting specific inhibition of the CDK5 analogue-sensitive ( as ) variant. Canonical ATP-analogue inhibitor (In, yellow) targets endogenous CDK5 (dark green) at its ATP-binding catalytic site nonspecifically since multiple kinases share structurally similar catalytic sites (left panel). The analogue-sensitive ( as , light green) phenylalanine-to-glycine (F80G) mutation confers a structural change adjacent to the catalytic site of CDK5 that does not impact its catalysis but accommodates the specific binding of a non-hydrolysable bulky orthogonal inhibitor 1NM-PP1(In*, orange). Introduction of 1NM-PP1 thus selectively inhibits CDK5- as variant (right panel). b , Immunoblots showing two clones (Cl 23 and Cl 50) of RPE-1 cells expressing FLAG-HA-CDK5- as in place of endogenous CDK5. Representative results are shown from three independent repeats. c , Proliferation curve of parental RPE-1 and RPE-1 CDK5- as cells. Data represent mean ± s.d. from three independent repeats. p -value was determined by Mann Whitney U test. d , Immunoblots showing immunoprecipitated CDK1-cyclin B1 complex or CDK5- as -cyclin B1 complex by the indicated antibody-coupled agarose, from nocodazole arrested RPE-1 CDK5- as cells with treated with or without 1NM-PP1 for inhibition of CDK5- as , from three independent replicate experiments. e , In-vitro kinase activity quantification of immunoprecipitated complex shown in d . Data represent mean ± s.d. from three independent experiments. p -values were determined by unpaired, two-tailed student’s t-test. f , Immunoblots of RPE-1 CDK5- as cells treated with either DMSO or 1NM-PP1 for 2 h prior to and upon release from RO-3306 and collected at 60 min following release. Cells were lysed and blotted with anti-bodies against indicated proteins (upper panel). Quantification of the relative intensity of PP4R3β phosphorylation at S840 in 1NM-PP1-treated CDK5- as cells compared to DMSO-treatment (lower panel). g , Experimental scheme for specific and temporal abrogation of CDK5 in RPE-1 CDK5- as cells. Data represent mean ± S.D from quadruplicate repeats. p -value was determined by one sample t and Wilcoxon test. h , Hoechst staining showing primary nuclei and micronuclei of RPE-1 CDK5- as with indicated treatment; scale bar is as indicated (left panel). Right, quantification of the percentage of cells with micronuclei after treatment. Data represent mean ± s.d. of three independent experiments from n = 2174 DMSO, n = 1788 1NM-PP1 where n is the number of cells. p- values were determined by unpaired, two-tailed student’s t-test. Scale bar is as indicated. Uncropped gel images are provided in Supplementary Fig. 1 .

Extended Data Fig. 2 Degradation of CDK5 in degradation tag (CDK5- dTAG ) system.

a , Schematic depicting the dTAG-13-inducible protein degradation system. Compound dTAG-13 links protein fused with FKBP12 F36V domain (dTAG) to CRBN-DDB1-CUL4A E3 ligase complex, leading to CRBN-mediated degradation. b , Immunoblots showing two clones of RPE-1 cells that express dTAG -HA-CDK5 in place of endogenous CDK5 (Cl N1 and Cl N4). Representative results are shown from three independent repeats. c , Proliferation curve of parental RPE-1 and RPE-1 CDK5-dTAG. Data represent mean ± s.d. of three independent repeats. p -value was determined by Mann Whitney U test. d and e , Representative images of RPE-1 CDK5- dTAG clone 1 (N1) ( d ) and RPE-1 CDK5- dTAG clone 4 (N4) ( e ) treated with DMSO or dTAG-13 for 2 h prior to and upon release from G2/M arrest and fixed at 120 min after release (top panel); quantification of CDK5 total intensity per cell (lower panels). Data represent mean ± s.d. of at least two independent experiments from n = 100 cells each condition. p- values were determined by unpaired, two-tailed student’s t-test. f , Immunoblots showing level of indicated proteins in RPE-1 CDK5- dTAG cells. Cells were treated with either DMSO or dTAG-13 for 2 h prior to and upon release from RO-3306 and lysed at 60 min following release (upper panel). Quantification of the relative intensity of PP4R3β phosphorylation at S840 in dTAG13-treated CDK5- dTAG cells compared to DMSO-treatment (lower panel). Data represent mean ± s.d. of four independent experiments. p -value was determined by one sample t and Wilcoxon test. g , Experimental scheme for specific and temporal abrogation of CDK5 in RPE-1 CDK5- dTAG cells. h , Hoechst staining showing primary nuclei and micronuclei of RPE-1 CDK5- dTAG with indicated treatment; scale bar is as indicated (left panel). Right, quantification of the percentage of cells with micronuclei after treatment. Data represent mean ± s.d. of three independent experiments from n = 2094 DMSO and n = 2095 dTAG-13, where n is the number of cells. p- values were determined by unpaired, two-tailed student’s t-test. Scale bar is as indicated. Uncropped gel images are provided in Supplementary Fig. 1 .

Extended Data Fig. 3 CDK5 abrogation render chromosome alignment and segregation defect despite intact spindle assembly checkpoint and timely mitotic duration.

a and b , Live-cell imaging snapshots of RPE-1 CDK5- as cells ( a ) and RPE-1 CDK5- dTAG cells ( b ) expressing mCherry-H2B and GFP-α-tubulin, abrogated of CDK5 by treatment with 1NM-PP1 or dTAG-13, respectively. Imaging commenced in prophase following release from RO-3306 into fresh media containing indicated chemicals (left); quantification of the percentage of cells with abnormal nuclear morphology (right). c and d , Representative snapshots of the final frame prior to metaphase-to-anaphase transition from a live-cell imaging experiment detailing chromosome alignment at the metaphase plate of RPE- CDK5- as (c) and RPE-1 CDK5- dTAG ( d ) expressing mCherry-H2B, and GFP-α-tubulin (left); quantification of the percentage of cells displaying abnormal chromosome alignment following indicated treatments (top right). e , Representative images showing the range of depolymerization outcomes (low polymers, high polymers and spindle-like) in DMSO- and 1NM-PP1-treated cells, as shown in Fig. 2e , from n = 50 for each condition, where n is number of metaphase cells . f , Quantifications of mitotic duration from nuclear envelope breakdown (NEBD) to anaphase onset of RPE-1 CDK5- as (left ) and RPE-1 CDK5- dTAG (right) cells, following the indicated treatments. Live-cell imaging of RPE-1 CDK5- as and RPE-1 CDK5- dTAG cells expressing mCherry-H2B and GFP-BAF commenced following release from RO-3306 arrest into fresh media containing DMSO or 1NM-PP1 or dTAG-13. g , Quantifications of the percentage of RPE-1 CDK5- as (left) and RPE-1 CDK5- dTAG (right) cells that were arrested in mitosis following the indicated treatments. Imaging commenced in prophase cells as described in a , following release from RO-3306 into fresh media in the presence or absence nocodazole as indicated. The data in a, c , and g represent mean ± s.d. of at least two independent experiments from n = 85 DMSO and n = 78 1NM-PP1 in a and c ; from n = 40 cells for each treatment condition in g . The data in b , d , and f represent mean ± s.d. of three independent experiments from n = 57 DMSO and n = 64 dTAG-13 in b and d ; from n = 78 DMSO and n = 64 1NM-PP1; n = 59 DMSO and n = 60 dTAG-13, in f , where n is the number of cells. p- values were determined by unpaired, two-tailed student’s t-test. Scale bar is as indicated.

Extended Data Fig. 4 CDK5 and CDK1 regulate tubulin dynamics.

a, b , Immunostaining of RPE-1 cells with antibodies against CDK1 and α-tubulin ( a ); and CDK5 and α-tubulin ( b ) at indicated stages of mitosis. c, d , Manders’ overlap coefficient M1 (CDK1 versus CDK5 on α-tubulin) ( c ); and M2 (α-tubulin on CDK1 versus CDK5) ( d ) at indicated phases of mitosis in cells shown in a and b . The data represent mean ± s.d. of at least two independent experiments from n = 25 cells in each mitotic stage. p- values were determined by unpaired, two-tailed student’s t-test.

Extended Data Fig. 5 Phosphoprotoemics analysis to identify mitotic CDK5 substrates.

a , Scheme of cell synchronization for phosphoproteomics: RPE-1 CDK5- as cells were arrested at G2/M by treatment with RO-3306 for 16 h. The cells were treated with 1NM-PP1 to initiate CDK5 inhibition. 2 h post-treatment, cells were released from G2/M arrest into fresh media with or without 1NM-PP1 to proceed through mitosis with or without continuing inhibition of CDK5. Cells were collected at 60 min post-release from RO-3306 for lysis. b , Schematic for phosphoproteomics-based identification of putative CDK5 substrates. c , Gene ontology analysis of proteins harbouring CDK5 inhibition-induced up-regulated phosphosites. d , Table indicating phospho-site of proteins that are down-regulated as result of CDK5 inhibition. e , Table indicating the likely kinases to phosphorylate the indicated phosphosites of the protein, as predicted by Scansite 4 66 . Divergent score denotes the extent by which phosphosite diverge from known kinase substrate recognition motif, hence higher divergent score indicating the corresponding kinase is less likely the kinase to phosphorylate the phosphosite.

Extended Data Fig. 6 Cyclin B1 is a mitotic co-factor of CDK5 and of CDK1.

a , Endogenous CDK5 was immunoprecipitated from RPE-1 cells collected at time points corresponding to the indicated cell cycle stage. Cell lysate input and elution of immunoprecipitation were immunoblotted by antibodies against the indicated proteins. RPE-1 cells were synchronized to G2 by RO-3306 treatment for 16 h and to prometaphase (M) by nocodazole treatment for 6 h. Asynch: Asynchronous. Uncropped gel images are provided in Supplementary Fig. 1 . b , Immunostaining of RPE-1 cells with antibodies against the indicated proteins at indicated mitotic stages (upper panels). Manders’ overlap coefficient M1 (Cyclin B1 on CDK1) and M2 (CDK1 on Cyclin B1) at indicated mitotic stages for in cells shown in b (lower panels). The data represent mean ± s.d. of at least two independent experiments from n = 25 mitotic cells in each mitotic stage. p- values were determined by unpaired, two-tailed student’s t-test. c , Table listing common proteins as putative targets of CDK5, uncovered from the phosphoproteomics anlaysis of down-regulated phosphoproteins upon CDK5 inhibition (Fig. 3 and Supplementary Table 1 ), and those of cyclin B1, uncovered from phosphoproteomics analysis of down-regulated phospho-proteins upon cyclin B1 degradation (Fig. 6 and Table EV2 in Hegarat et al. EMBO J. 2020). Proteins relevant to mitotic functions are highlighted in red.

Extended Data Fig. 7 Structural prediction and analyses of the CDK5-cyclin B1 complex.

a , Predicted alignment error (PAE) plots of the top five AlphaFold2 (AF2)-predicted models of CDK5-cyclin B1 (top row) and CDK1-cyclin B1 (bottom row) complexes, ranked by interface-predicted template (iPTM) scores. b , AlphaFold2-Multimer-predicted structure of the CDK5-cyclin B1 complex. c , Structural comparison of CDK-cyclin complexes. Left most panel: Structural-overlay of AF2 model of CDK5-cyclin B1 and crystal structure of phospho-CDK2-cyclin A3-substrate complex (PDB ID: 1QMZ ). The zoomed-in view of the activation loops of CDK5 and CDK2 is shown in the inset. V163 (in CDK5), V164 (in CDK2) and Proline at +1 position in the substrates are indicated with arrows. Middle panel: Structural-overlay of AF2 model of CDK5-cyclin B1 and crystal structure of CDK1-cyclin B1-Cks2 complex (PDB ID: 4YC3 ). The zoomed-in view of the activation loops of CDK5 and CDK1 is shown in the inset. Cks2 has been removed from the structure for clarity. Right most panel: structural-overlay of AF2 models of CDK5-cyclin B1 and CDK1-cyclin B1 complex. The zoomed view of the activation loops of CDK5 and CDK1 is shown in the inset. d , Secondary structure elements of CDK5, cyclin B1 and p25. The protein sequences, labelled based on the structural models, are generated by PSPript for CDK5 (AF2 model) ( i ), cyclin B1 (AF2 model) ( ii ) and p25 (PDB ID: 3O0G ) ( iii ). Structural elements ( α , β , η ) are defined by default settings in the program. Key loops highlighted in Fig. 4d are mapped onto the corresponding sequence.

Extended Data Fig. 8 Phosphorylation of CDK5 S159 is required for kinase activity and mitotic fidelity.

a , Structure of the CDK5-p25 complex (PDB ID: 1h41 ). CDK5 (blue) interacts with p25 (yellow). Serine 159 (S159, magenta) is in the T-loop. b , Sequence alignment of CDK5 and CDK1 shows that S159 in CDK5 is the analogous phosphosite as that of T161 in CDK1 for T-loop activation. Sequence alignment was performed by CLC Sequence Viewer ( https://www.qiagenbioinformatics.com/products/clc-sequence-viewer/ ). c , Immunoblots of indicated proteins in nocodazole-arrested mitotic (M) and asynchronous (Asy) HeLa cell lysate. d , Myc-His-tagged CDK5 S159 variants expressed in RPE-1 CDK5- as cells were immunoprecipitated from nocodazole-arrested mitotic lysate by Myc-agarose. Input from cell lysate and elution from immunoprecipitation were immunoblotted with antibodies against indicated protein. EV= empty vector. In vitro kinase activity assay of the indicated immunoprecipitated complex shown on the right panel. Data represent mean ± s.d. of four independent experiments. p -values were determined by unpaired two-tailed student’s t-test. e , Immunoblots showing RPE-1 FLAG-CDK5- as cells stably expressing Myc-His-tagged CDK5 WT and S159A, which were used in live-cell imaging and immunofluorescence experiments to characterize chromosome alignment and spindle architecture during mitosis, following inhibition of CDK5- as by 1NM-PP1, such that only the Myc-His-tagged CDK5 WT and S159A are not inhibited. Representative results are shown from three independent repeats. f , Hoechst staining showing nuclear morphology of RPE-1 CDK5- as cells expressing indicated CDK5 S159 variants following treatment with either DMSO or 1NMP-PP1 and fixation at 120 min post-release from RO-3306-induced arrest (upper panel); quantification of nuclear circularity and solidity (lower panels) g , Snapshots of live-cell imaging RPE-1 CDK5- as cells expressing indicated CDK5 S159 variant, mCherry-H2B, and GFP-α-tubulin, after release from RO-3306-induced arrest at G2/M, treated with 1NM-PP1 2 h prior to and upon after release from G2/M arrest (upper panel); quantification of cells displaying abnormal chromosome alignment in (lower panel). Representative images are shown from two independent experiments, n = 30 cells each cell line. h , Representative images of RPE-1 CDK5- as cells expressing indicated CDK5 S159 variants in metaphase, treated with DMSO or 1NM-PP1 for 2 h prior to and upon release from RO-3306-induced arrest, and then released into media containing 20 µM proTAME for 2 h, fixed and stained with tubulin and DAPI (upper panel); metaphase plate width and spindle length measurements for these representative cells were shown in the table on right; quantification of metaphase plate width and spindle length following the indicated treatments (lower panel). Data in f and h represent mean ± s.d. of at least two independent experiments from n = 486 WT, n = 561 S159A, and n = 401 EV, where n is the number of cells in f ; from n = 65 WT, n = 64 S159A, and n = 67 EV, where n is the number of cells in h . Scale bar is as indicated. Uncropped gel images are provided in Supplementary Fig. 1 .

Extended Data Fig. 9 The CDK5 co-factor-binding helix regulates CDK5 kinase activity.

a , Structure of the CDK5-p25 complex (PDB ID: 1h41 ). CDK5 (blue) interacts with p25 (yellow) at the PSSALRE helix (green). Serine 46 (S46, red) is in the PSSALRE helix. Serine 159 (S159, magenta) is in the T-loop. b , Sequence alignment of CDK5 and CDK1 shows that S46 is conserved in CDK1 and CDK5. Sequence alignment was performed by CLC Sequence Viewer ( https://www.qiagenbioinformatics.com/products/clc-sequence-viewer/ ). c , Immunoblots of CDK5 immunoprecipitation from lysate of E. coli BL21 (DE3) expressing His-tagged human CDK5 WT or CDK5 S46D, mixed with lysate of E. coli BL21 (DE3) expressing His-tagged human cyclin B1. Immunoprecipitated CDK5 alone or in the indicated complex were used in kinase activity assay, shown in Fig. 5b . Representative results are shown from three independent repeats. d , Immunoblots showing RPE-1 FLAG-CDK5- as cells stably expressing Myc-His-tagged CDK5 S46 phospho-variants, which were used in live-cell imaging and immunofluorescence experiments to characterize chromosome alignment and spindle architecture during mitosis, following inhibition of CDK5- as by 1NM-PP1, such that only the Myc-His-tagged CDK5 S46 phospho-variants are not inhibited. Representative results are shown from three independent repeats. e , Immunostaining of RPE-1 CDK5- as cells expressing Myc-His-tagged CDK5 WT or S46D with anti-PP4R3β S840 (pS840) antibody following indicated treatment (DMSO vs 1NM-PP1). Scale bar is as indicated (left). Normalized intensity level of PP4R3β S840 phosphorylation (right). Data represent mean ± s.d. of at least two independent experiments from n = 40 WT and n = 55 S46D, where n is the number of metaphase cells. p- values were determined by unpaired two-tailed student’s t-test. f , Immunoblots showing level of indicated proteins in RPE-1 CDK5- as cells expressing Myc-His-tagged CDK5 WT or S46D. Cells were treated with either DMSO or 1NM-PP1 for 2 h prior to and upon release from RO-3306 and collected and lysed at 60 min following release (left). Quantification of the intensity of PP4R3β phosphorylation at S840 (right). Data represent mean ± s.d. of four independent experiments. p -values were determined by two-tailed one sample t and Wilcoxon test. g , Representative snapshots of live-cell imaging of RPE-1 CDK5- as cells harbouring indicated CDK5 S46 variants expressing mCherry-H2B and GFP-α-tubulin, treated with 1NM-PP1, as shown in Fig. 5d , from n = 35 cells. Imaging commenced in prophase following release from RO-3306 into fresh media containing indicated chemicals. Uncropped gel images are provided in Supplementary Fig. 1 .

Extended Data Fig. 10 Localization of CDK5 S46 phospho-variants.

Immunostaining of RPE-1 CDK5- as cells stably expressing Myc-His CDK5-WT ( a ), S46A ( b ), and S46D ( c ) with antibodies against indicated protein in prophase, prometaphase, and metaphase. Data represent at least two independent experiments from n = 25 cells of each condition in each mitotic stage.

Extended Data Fig. 11 RPE-1 harbouring CDK5- as introduced by CRISPR-mediated knock-in recapitulates chromosome mis-segregation defects observed in RPE-1 overexpressing CDK5- as upon inhibition of CDK5- as by 1NM-PP1 treatment.

a , Chromatogram showing RPE-1 that harbours the homozygous CDK5- as mutation F80G introduced by CRISPR-mediated knock-in (lower panel), replacing endogenous WT CDK5 (upper panel). b , Immunoblots showing level of CDK5 expressed in parental RPE-1 and RPE-1 that harbours CDK5- as F80G mutation in place of endogenous CDK5. c , Representative images of CDK5- as knocked-in RPE-1 cells exhibiting lagging chromosomes following indicated treatments. d , Quantification of percentage of cells exhibiting lagging chromosomes following indicated treatments shown in (c). Data represent mean ± s.d. of three independent experiments from n = 252 DMSO, n = 220 1NM-PP1, where n is the number of cells. p -value was determined by two-tailed Mann Whitney U test.

Extended Data Fig. 12 CDK5 is highly expressed in post-mitotic neurons and overexpressed in cancers.

a , CDK5 RNAseq expression in tumours (left) with matched normal tissues (right). The data are analysed using 22 TCGA projects. Note that CDK5 expression is higher in many cancers compared to the matched normal tissues. BLCA, urothelial bladder carcinoma; BRCA, breast invasive carcinoma; CESC cervical squamous cell carcinoma and endocervical adenocarcinoma; CHOL, cholangiocarcinoma; COAD, colon adenocarcinoma; ESCA, esophageal carcinoma; HNSC, head and neck squamous cell carcinoma; KICH, kidney chromophobe; KIRC, kidney renal clear cell carcinoma; KIRP, kidney renal papillary cell carcinoma; LIHC, liver hepatocellular carcinoma; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; PAAD, pancreatic adenocarcinoma; PCPG, pheochromocytoma and paraganglioma; PRAD, prostate adenocarcinoma; READ, rectum adenocarcinoma; SARC, sarcoma; STAD, stomach adenocarcinoma; THCA, thyroid carcinoma; THYM, thymoma; and UCEC, uterine corpus endometrial carcinoma. p -value was determined by two-sided Student’s t-test. ****: p <= 0.0001; ***: p <= 0.001; **: p <= 0.01; *: p <= 0.05; ns: not significant, p  > 0.05. b , Scatter plots showing cells of indicated cancer types that are more dependent on CDK5 and less dependent on CDK1. Each dot represents a cancer cell line. The RNAi dependency data (in DEMETER2) for CDK5 and CDK1 were obtained from the Dependency Map ( depmap.org ). The slope line represents a simple linear regression analysis for the indicated cancer type. The four indicated cancer types (Head/Neck, Ovary, CNS/Brain, and Bowel) showed a trend of more negative CDK5 RNAi effect scores (indicative of more dependency) with increasing CDK1 RNAi effect scores (indicative of less dependency). The p value represents the significance of the correlation computed from a simple linear regression analysis of the data. Red circle highlights the subset of the cells that are relatively less dependent on CDK1 but more dependent on CDK5. c , Scatter plots showing bowel cancer cells that expresses CDK5 while being less dependent on CDK1. Each dot represents a cancer cell line. The data on gene effect of CDK1 CRISPR and CDK5 mRNA level were obtained from the Dependency Map ( depmap.org ). The slope line represents a simple linear regression analysis. Red circle highlights the subset of cells that are relatively less dependent on CDK1 but expresses higher level of CDK5. For b and c , solid line represents the best-fit line from simple linear regression using GraphPad Prism. Dashed lines represent 95% confidence bands of the best-fit line. p -value is determined by the F test testing the null hypothesis that the slope is zero. d , Scatter plots showing rapidly dividing cells of indicated cancer types that are more dependent on CDK5 and less dependent on CDK1. Each dots represents a cancer cell line. The doubling time data on the x-axis were obtained from the Cell Model Passports ( cellmodelpassports.sanger.ac.uk ). The RNAi dependency data (in DEMETER2) for CDK5, or CDK1, on the y-axis were obtained from the Dependency Map ( depmap.org ). Only cell lines with doubling time of less than 72 h are displayed and included for analysis. Each slope line represents a simple linear regression analysis for each cancer type. The indicated three cancer types were analysed and displayed because they showed a trend of faster proliferation rate (lower doubling time) with more negative CDK5 RNAi effect (more dependency) but increasing CDK1 RNAi effect (less dependency) scores. The p value represents the significance of the association of the three cancer types combined, computed from a multiple linear regression analysis of the combined data, using cancer type as a covariate. Red circle depicts subset of fast dividing cells that are relatively more dependent on CDK5 (left) and less dependent on CDK1 (right). Solid lines represent the best-fit lines from individual simple linear regressions using GraphPad Prism. p -value is for the test with the null hypothesis that the effect of the doubling time is zero from the multiple linear regression RNAi ~ Intercept + Doubling Time (hours) + Lineage.

Supplementary information

Supplementary figure 1.

Full scanned images of all western blots.

Reporting Summary

Peer review file, supplementary table 1.

Phosphosite changes in 1NM-PP1-treated cells versus DMSO-treated controls as measured by LC–MS/MS.

Supplementary Table 2

Global protein changes in 1NM-PP1-treated cells versus DMSO-treated controls as measured by LC–MS/MS.

Supplementary Video 1

RPE-1 CDK5(as) cell after DMSO treatment, ×100 imaging.

Supplementary Video 2

RPE-1 CDK5(as) cell after 1NM-PP1 treatment (example 1), ×100 imaging.

Supplementary Video 3

RPE-1 CDK5(as) cell after 1NM-PP1 treatment (example 2), ×100 imaging.

Supplementary Video 4

RPE-1 CDK5(dTAG) cell after DMSO treatment, ×100 imaging.

Supplementary Video 5

RPE-1 CDK5(dTAG) cell after dTAG-13 treatment (example 1), ×100 imaging.

Supplementary Video 6

RPE-1 CDK5(dTAG) cell after dTAG-13 treatment (example 2) ×100 imaging.

Supplementary Video 7

RPE-1 CDK5(as) cells expressing MYC-CDK5(WT) after 1NM-PP1 treatment, ×20 imaging.

Supplementary Video 8

RPE-1 CDK5(as) cells expressing MYC-EV after 1NM-PP1 treatment, ×20 imaging.

Supplementary Video 9

RPE-1 CDK5(as) cells expressing MYC-CDK5(S159A) after 1NM-PP1 treatment (example 1), ×20 imaging.

Supplementary Video 10

RPE-1 CDK5(as) cells expressing MYC-CDK5(S159A) after 1NM-PP1 treatment (example 2), ×20 imaging.

Supplementary Video 11

RPE-1 CDK5(as) cells expressing MYC-CDK5(WT) after 1NM-PP1 treatment, ×100 imaging.

Supplementary Video 12

RPE-1 CDK5(as) cells expressing MYC-CDK5(S46A) after 1NM-PP1 treatment (example 1), ×100 imaging.

Supplementary Video 13

RPE-1 CDK5(as) cells expressing MYC-CDK5(S46A) after 1NM-PP1 treatment (example 2), ×100 imaging.

Supplementary Video 14

RPE-1 CDK5(as) cells expressing MYC-CDK5(S46D) after 1NM-PP1 treatment (example 1), ×100 imaging.

Supplementary Video 15

RPE-1 CDK5(as) cells expressing MYC-CDK5(S46D) after 1NM-PP1 treatment (example 2), ×100 imaging.

Supplementary Video 16

RPE-1 CDK5(as) cells expressing MYC-EV after 1NM-PP1 treatment,×100 imaging.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Cite this article.

Zheng, XF., Sarkar, A., Lotana, H. et al. CDK5–cyclin B1 regulates mitotic fidelity. Nature (2024). https://doi.org/10.1038/s41586-024-07888-x

Download citation

Received : 24 March 2023

Accepted : 30 July 2024

Published : 04 September 2024

DOI : https://doi.org/10.1038/s41586-024-07888-x

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.