A cyclin-dependent kinase (CDK) is a member of a family of enzymes that regulate cellular processes by transferring phosphate groups to other proteins. CDKs are best known for controlling the cell cycle, the ordered sequence through which a cell duplicates its genetic material and divides. Other family members regulate the synthesis and processing of RNA, neuronal functions, and responses to cellular stress. Their name reflects their characteristic dependence on regulatory proteins called cyclins, although some CDKs also associate with non-cyclin activators.
CDKs convert regulatory signals into changes in protein activity, localization, interactions, or stability. Their catalytic activity is controlled by partner binding, phosphorylation, inhibitory proteins, and protein degradation. Together, these mechanisms allow cells to initiate major events—such as DNA replication or chromosome segregation—at appropriate times and in the correct order. Disruption of CDK regulation contributes to cancer and other diseases, while selective CDK inhibitors have become established treatments for particular cancers.
Discovery and classification
The identification of CDKs brought together genetic studies in yeast and biochemical studies of dividing animal cells. In the 1970s, Leland Hartwell and colleagues used temperature-sensitive mutants of the budding yeast Saccharomyces cerevisiae to identify genes required at particular stages of cell division. One of these, CDC28, proved essential for passing the major commitment point known as Start. In the fission yeast Schizosaccharomyces pombe, Paul Nurse and colleagues identified cdc2 as a central regulator of both entry into DNA synthesis and entry into mitosis. These experiments showed that individual genes could control transitions between large-scale cellular states.
A complementary line of research began with maturation-promoting factor, or MPF. In 1971, Yoshio Masui and Clement Markert showed that cytoplasm from a maturing frog oocyte could induce maturation in another oocyte. Subsequent work established that MPF activity also controlled entry into mitosis. In 1983, Tim Hunt and colleagues described proteins in sea urchin embryos whose abundance rose during the cell cycle and fell abruptly at division, naming them cyclins. Biochemical work later established that a central component of MPF was a complex of a Cdc2-related kinase and a B-type cyclin.
A particularly important finding was the conservation of this machinery. In 1987, a human gene encoding the counterpart of fission-yeast Cdc2 was shown to complement a yeast cdc2 defect. The protein is now called CDK1. Such results demonstrated that a regulatory system identified in microorganisms had been retained across widely separated eukaryotic lineages. Hartwell, Hunt, and Nurse received the 2001 Nobel Prize in Physiology or Medicine for discoveries concerning key regulators of the cell cycle.
Human CDKs are conventionally designated CDK1 through CDK20, although gene duplication and historical nomenclature complicate a simple one-number–one-gene interpretation. For example, CDK11 proteins are encoded by two closely related human genes. CDK1, CDK2, CDK4, and CDK6 are prominent cell-cycle regulators; CDK7, CDK8, CDK9, CDK12, and CDK13 have major roles in transcriptional regulation. CDK5 is especially associated with neuronal biology. These categories overlap: CDK7 activates other CDKs as well as participating in transcription, and several transcription-associated CDKs influence proliferation indirectly.
The CDK family belongs to the broader group of CMGC protein kinases, which also includes mitogen-activated protein kinases and glycogen synthase kinases. Family membership is defined by sequence and structural relationships, not solely by a demonstrated role in cell division. Consequently, “cyclin-dependent kinase” denotes an evolutionary and biochemical family rather than an exclusively cell-cycle-specific activity.
Catalytic structure and substrate recognition
CDKs are serine/threonine protein kinases. They usually transfer the terminal phosphate group of adenosine triphosphate (ATP) to the hydroxyl group of a serine or threonine residue in a substrate protein. The reaction can be represented schematically as
Magnesium ions participate in nucleotide binding and catalysis. Phosphorylation can alter a substrate’s shape, create a binding site for another protein, obstruct an interaction, or mark the substrate for subsequent modification. The biological effect therefore depends on the substrate and the position modified, rather than on phosphorylation having an intrinsically activating or inhibitory meaning.
The catalytic domain has the two-lobed architecture characteristic of protein kinases. A smaller amino-terminal lobe contributes to ATP positioning, while a larger carboxy-terminal lobe provides much of the substrate-binding surface. ATP binds in the cleft between them. Important structural elements include the glycine-rich loop, the αC helix, the catalytic loop, and the activation segment, often called the T-loop in CDKs. Conserved residues organize ATP and the substrate hydroxyl group for phosphate transfer.
In an inactive monomeric CDK, these elements are not optimally aligned. Cyclin binding reorganizes the kinase, moving the αC helix and changing the position of the activation segment. Structural studies of CDK2 provided especially clear examples. Its αC helix contains the characteristic PSTAIRE sequence, although related CDKs contain variants of this motif. Cyclin binding helps align catalytic residues, and phosphorylation of the activation segment can stabilize a more fully active conformation. Structures determined by X-ray crystallography established how regulatory subunits control an enzyme without contributing a second catalytic site.
Many cell-cycle CDKs favor a minimal substrate sequence containing serine or threonine followed immediately by proline. A frequently used extended consensus is , where X represents any amino acid. This preference is not an absolute rule, and not every matching sequence is phosphorylated in a cell. Accessibility, competing enzymes, subcellular location, and the availability of docking interactions all affect substrate selection.
Cyclins contribute specificity by binding substrates as well as activating the kinase. Several cyclins recognize docking motifs containing an arginine followed by a short sequence that includes leucine, often described as an RXL or Cy motif. Additional partners, including Cks proteins, help organize interactions with phosphorylated substrates. A protein bearing many phosphorylation sites may therefore respond differently to two CDK–cyclin complexes even when their catalytic sites have similar sequence preferences. CDK specificity emerges from the entire molecular complex and its cellular setting.
Layers of activation and inhibition
Cyclin binding is a major regulatory step, but CDK activity is not determined by cyclin abundance alone. Cyclins differ in their expression patterns, degradation rates, binding preferences, and intracellular locations. D-type cyclins connect extracellular growth signals to CDK4 and CDK6; E- and A-type cyclins associate prominently with CDK2; A- and B-type cyclins associate with CDK1. These are principal physiological partnerships rather than an exhaustive list of possible interactions.
For many CDKs, activation also requires phosphorylation of a threonine in the activation segment. Well-characterized examples are Thr160 in CDK2 and Thr161 in CDK1. In animal cells, the CDK-activating kinase, or CAK, is a complex containing CDK7, cyclin H, and MAT1. The organization differs in budding yeast, where Cak1 performs the principal CDK-activating function. This distinction illustrates that the core catalytic machinery is conserved more strongly than every detail of its regulation.
Phosphorylation at other positions has the opposite effect. In CDK1, phosphorylation of Thr14 and Tyr15 inhibits activity despite cyclin binding. Wee1 primarily supplies inhibitory Tyr15 phosphorylation, while Myt1 can phosphorylate Thr14 and Tyr15. Members of the Cdc25 phosphatase family remove these inhibitory phosphates. Entry into mitosis therefore depends partly on the balance between inhibitory kinases and activating phosphatases, not simply on the accumulation of cyclin B.
Dedicated CDK inhibitor proteins provide another layer of control. The INK4 family, comprising p15, p16, p18, and p19, preferentially inhibits CDK4 and CDK6. These proteins alter kinase conformation and interfere with productive cyclin association. The Cip/Kip family includes p21, p27, and p57, which regulate several cyclin–CDK complexes. Their effects are context-dependent: they can inhibit CDK2-containing complexes, while p21 and p27 can also assist the assembly, localization, or regulation of cyclin D–CDK4/6 complexes. The name “inhibitor” thus does not capture every molecular function of these proteins.
Protein degradation makes regulation directional. The ubiquitin–proteasome system removes cyclins and CDK inhibitors through regulated ubiquitination. SCF ubiquitin ligases participate in the destruction of phosphorylated inhibitors such as p27, while the anaphase-promoting complex/cyclosome, APC/C, promotes the destruction of mitotic cyclins and other cell-cycle regulators. These events are not interchangeable: eliminating an inhibitor can increase CDK activity, whereas destroying an activating cyclin can terminate it.
The combined system contains both positive and negative feedback. CDK1 can promote its own activation by stimulating Cdc25 and opposing Wee1-dependent inhibition. Later, CDK-dependent events help enable APC/C activation and cyclin destruction. Such circuitry produces rapid transitions and oscillations rather than a simple proportional relationship between cyclin concentration and cellular behavior.
Coordination of DNA replication and mitosis
During a typical proliferative mammalian cell cycle, different CDK activities predominate in G1, S, G2, and M phases. This succession coordinates growth, DNA replication, and mitosis. The familiar sequence—cyclin D–CDK4/6, cyclin E–CDK2, cyclin A-associated kinases, and cyclin B–CDK1—is useful, but it should not be interpreted as a rigid relay in which each enzyme has only one function and no substitute.
In G1, growth-factor signaling promotes D-type cyclin production and CDK4/6 activity. An important substrate is the retinoblastoma protein, RB, which restrains E2F transcription factors. Changes in RB phosphorylation alter this restraint and support expression of genes needed for DNA synthesis, including genes encoding replication proteins and additional cell-cycle regulators. Cyclin E–CDK2 participates in the transition toward sustained S-phase activity. The precise order and functional meaning of individual RB phosphorylation events depend on cellular context, but the central relationship between CDK activity, RB, and E2F is well established.
Replication requires an additional distinction between licensing an origin and firing it. During low-CDK states, prereplication complexes can load the MCM helicase machinery onto DNA. S-phase CDKs subsequently cooperate with Dbf4-dependent kinase to activate licensed origins. At the same time, rising CDK activity prevents new licensing through effects on replication-initiation proteins and associated regulators. This separation helps ensure that a chromosome region is normally replicated once, rather than repeatedly, within one division cycle.
CDK1 drives the major reorganizations of mitosis. Through a broad substrate network, it promotes chromosome condensation, nuclear-envelope breakdown, reorganization of the cytoskeleton, and assembly of the mitotic spindle. These outputs require more than one kinase: CDK1 works with Polo-like kinases, Aurora kinases, and other regulators. Phosphorylation of nuclear lamins, for example, contributes to disassembly of the nuclear lamina, while changes in microtubule-associated proteins help transform the interphase microtubule network into a division apparatus.
Mitotic exit requires reversal of much of this phosphorylation. Once the conditions for chromosome segregation are satisfied, APC/C-dependent degradation of securin permits separase activation, while destruction of cyclin B lowers CDK1 activity. Protein phosphatases, especially PP1 and PP2A-containing complexes, then promote chromosome decondensation, nuclear-envelope reassembly, and other aspects of exit. CDKs therefore establish a mitotic state, but its termination requires coordinated proteolysis and dephosphorylation.
Genetic experiments reveal considerable flexibility in this system. Mouse embryonic cells can proliferate without several interphase CDKs because CDK1 can associate with cyclins normally assigned to those enzymes. Nevertheless, individual CDKs remain essential for particular tissues or developmental processes; CDK2, for example, has a critical role in meiosis. Such findings distinguish the minimum machinery needed for a division cycle from the specialized circuitry used by an intact organism.
Checkpoints, growth arrest, and genome protection
CDK regulation provides a major connection between cell-cycle progression and surveillance of cellular conditions. A cell-cycle checkpoint is a signaling mechanism that delays progression when a necessary event is incomplete or damage has occurred. Checkpoints do not generally inspect a single CDK molecule directly. Instead, they alter the regulatory network that determines whether CDK-dependent transitions can proceed.
DNA damage activates signaling pathways centered on ATM or ATR and downstream checkpoint kinases. ATM is strongly associated with responses to DNA double-strand breaks, whereas ATR responds prominently to replication-associated structures containing single-stranded DNA. CHK1 and CHK2 contribute to control of Cdc25 phosphatases and other regulators. Suppression of Cdc25 activity can preserve inhibitory phosphorylation on CDK1 or CDK2, delaying mitotic entry or restraining replication-associated activity while DNA repair and recovery mechanisms operate.
The p53 pathway provides a transcriptional route to CDK inhibition. Following appropriate stress signals, p53 can increase expression of CDKN1A, which encodes p21. By inhibiting selected cyclin–CDK complexes, p21 helps impose cell-cycle arrest. The outcome is not invariably permanent: transient arrest, prolonged arrest, cellular senescence, and apoptosis are distinct responses whose occurrence depends on cell type, damage, and signaling context. The p53–p21 axis is therefore one component of a broader decision-making network.
Replication stress reveals a particularly important balance. Some CDK activity is necessary to initiate DNA synthesis, but excessive initiation can increase the number of active replication forks and strain available resources. Checkpoint signaling limits inappropriate origin firing while helping stabilize stressed forks. If cells enter mitosis with incompletely replicated or damaged DNA, chromosome breakage and segregation errors can result. CDK restraint protects the coordination between genome duplication and chromosome partitioning.
The spindle assembly checkpoint acts later and regulates CDK activity less directly. Unattached kinetochores generate a signal that inhibits APC/C activation by Cdc20. This delays securin and cyclin B destruction, helping preserve the mitotic state until chromosome attachment requirements are satisfied. Thus, one checkpoint can inhibit a transition by reducing CDK activity, while another can delay a transition by preventing the fall in CDK activity.
Longer-term withdrawal from proliferation involves overlapping mechanisms. Quiescent cells generally maintain low activities of cell-cycle CDKs but can re-enter the cycle under suitable conditions. Senescent cells exhibit a more stable proliferative arrest, often involving p16–RB and p53–p21 pathways. Neither state can be defined by a single inhibitor measurement. CDK activity is a functional output of growth signals, stress responses, differentiation programs, and the state of the genome.
Transcription and RNA processing
Several CDKs regulate transcription rather than primarily controlling transitions between cell-cycle phases. Their substrates include components of the transcription machinery, regulatory proteins, and factors involved in RNA maturation. A central target is the carboxy-terminal domain, or CTD, of the largest subunit of RNA polymerase II. The human CTD contains 52 repeats related to the seven-residue sequence Tyr–Ser–Pro–Thr–Ser–Pro–Ser. Different phosphorylation patterns help organize successive stages of transcription and RNA processing.
CDK7, cyclin H, and MAT1 form a kinase module within the general transcription factor TFIIH. In this setting, CDK7 phosphorylates the polymerase II CTD, especially at Ser5 and Ser7. Ser5 phosphorylation is associated with early transcription and helps recruit the machinery that adds the protective cap to a newly synthesized RNA. CDK7 therefore links promoter-associated events to the production and maturation of messenger RNA, while its separate CAK activity connects it to cell-cycle control.
CDK9 associates principally with T-type cyclins to form positive transcription elongation factor b, or P-TEFb. After initiation, RNA polymerase II frequently pauses near a gene’s promoter under the influence of DSIF and NELF. CDK9 phosphorylates components of these factors, including SPT5 within DSIF, and contributes to polymerase CTD phosphorylation. These changes promote release into productive elongation. A portion of P-TEFb is held in an inactive complex containing 7SK RNA and HEXIM proteins, providing a regulated reservoir of elongation activity.
CDK12 and CDK13, usually associated with cyclin K, also contribute to transcription elongation and RNA processing. They phosphorylate the polymerase II CTD and influence the production of full-length transcripts. CDK12 is particularly important for proper expression of many long genes, including genes involved in genome maintenance. Loss of its activity can increase premature cleavage and polyadenylation within genes. Although CDK12 and CDK13 are closely related, their functions overlap only partly.
CDK8 and CDK19 operate in a kinase module associated with Mediator, together with cyclin C, MED12, and MED13 or their related partners. This module influences interactions among transcriptional regulators, Mediator, and polymerase II. Its effects can be activating or repressive depending on the gene and signaling environment. It is therefore misleading to classify these kinases simply as general stimulators of transcription.
Transcription-associated CDKs demonstrate why CDK inhibition cannot automatically be equated with cell-cycle arrest. Inhibiting a kinase such as CDK9 can rapidly alter transcript production and reduce proteins whose messenger RNAs or protein products turn over quickly. Some consequences may precede any detectable change in cell-cycle distribution. Conversely, altered transcription can secondarily change cyclin production, DNA repair capacity, or survival signaling, coupling the transcriptional and cell-cycle branches of the family.
Specialized functions and evolutionary diversity
CDK5 is an important exception to the association between CDKs, conventional cyclins, and proliferation. Its best-established activators are p35 and p39, proteins that are not typical cell-cycle cyclins but adopt a cyclin-like activating structure. CDK5 is especially active in the nervous system, including in differentiated neurons that do not normally divide. It participates in neuronal migration, cytoskeletal organization, axonal transport, and synaptic function.
The location and lifetime of a CDK5 activator can change the biological consequences of kinase activity. The amino-terminal region of p35 helps target the complex to membranes. Proteolytic cleavage of p35 by calpain can generate p25, which lacks that targeting region and is more stable. Consequently, p25-associated CDK5 can differ from p35-associated CDK5 in duration and localization of activity. This mechanism has been studied in models of neuronal injury and neurodegeneration, but findings from such models do not by themselves establish a single causal pathway for human neurological disease.
Other family members have specialized or incompletely resolved functions. CDK11 proteins have been implicated in transcription, RNA splicing, and mitotic regulation. CDK14, in association with cyclin Y, can participate in regulation of Wnt signaling through phosphorylation of the LRP6 coreceptor. Such examples show how a conserved catalytic framework can be recruited into signaling networks that are not adequately described by a simple G1-to-M sequence.
Comparisons among organisms reveal both conservation and diversification. Budding yeast relies heavily on a single major cell-cycle CDK, Cdc28, partnered with different cyclins. Fission yeast similarly uses Cdc2 for central cell-cycle transitions. Mammals distribute related functions among several CDKs, enabling tissue-specific control and more elaborate responses to extracellular signals. This expansion creates redundancy, but also provides opportunities for specialization.
Plants possess recognizable CDK machinery alongside lineage-specific features. CDKA proteins have important roles around both major cell-cycle transitions, whereas plant-specific CDKB proteins contribute prominently to later cell-cycle events. Plant growth also frequently involves endoreduplication, in which DNA is replicated without a corresponding mitosis. Regulation of CDKs and their inhibitors helps determine whether a cell continues mitotic division or enters such an alternative cycle.
The distribution of CDKs across eukaryotes supports an ancient origin for this regulatory strategy. Bacteria do not use the canonical cyclin–CDK system to organize their division cycle, despite possessing other protein kinases and elaborate mechanisms coordinating replication and division. Evolution has therefore preserved CDK-centered control within eukaryotes while modifying cyclin repertoires, activating enzymes, inhibitors, and substrate networks to suit different cellular and developmental demands.
Cancer biology and therapeutic inhibition
CDK pathways are frequently altered in cancer, although the alteration is often in a regulator rather than in the kinase itself. Examples include amplification of the cyclin D1 gene, CCND1; loss or silencing of CDKN2A, which encodes p16; and disruption of the RB pathway. These changes can weaken the normal connection between growth conditions and entry into the cell cycle. They do not produce identical biological effects, and dependence on a particular CDK varies among tumors.
The cyclin D–CDK4/6–RB axis became a major therapeutic target because it provides a relatively defined link between mitogenic signaling and proliferation. Palbociclib, ribociclib, and abemaciclib inhibit CDK4 and CDK6 and have established uses in hormone receptor-positive, HER2-negative breast cancer. Their first United States approvals occurred in 2015 for palbociclib and in 2017 for ribociclib and abemaciclib. Exact indications differ among agents and have expanded through subsequent regulatory decisions.
These drugs are commonly used with endocrine therapy in specified clinical settings. Their direct biochemical action is inhibition of kinase activity; reduced RB phosphorylation and suppression of cell-cycle entry are important downstream consequences. Functional RB is generally necessary for the canonical antiproliferative response. However, measuring a single component of the pathway does not fully predict response, because tumors differ in signaling inputs, compensatory pathways, and evolutionary history.
Resistance can involve loss of RB function, increased cyclin E–CDK2 activity, or changes in growth-factor and other signaling pathways. Some resistant tumors acquire identifiable genetic changes, while others alter regulatory states without a single explanatory mutation. Laboratory models have also revealed effects of CDK4/6 inhibition on metabolism, senescence-associated programs, and interactions with the immune system. The contribution of each effect to clinical outcomes depends on context.
Transcription-associated CDKs represent a different therapeutic strategy. Inhibiting CDK7, CDK9, CDK12, or CDK13 can disrupt transcriptional programs or the production of survival and DNA-repair proteins. Such approaches have been extensively investigated, but biological rationale, activity in cultured cells, and clinical benefit are separate levels of evidence. Earlier broad-spectrum CDK inhibitors illustrated the difficulty of obtaining useful tumor selectivity when several essential kinases are inhibited simultaneously.
Selectivity also shapes adverse effects. Normal bone marrow, intestinal tissues, and other proliferative compartments depend on regulated cell-cycle activity, while transcriptional CDKs support functions in dividing and nondividing cells alike. CDK4/6 inhibitors have clinically important differences in their adverse-effect profiles, despite sharing principal targets. Thus, “CDK inhibitor” describes a mechanistic class, not a uniform clinical effect, and findings about one compound cannot automatically be transferred to another.
Experimental approaches and unresolved mechanisms
CDK research combines genetics, biochemistry, structural biology, and live-cell measurements. Purified kinase assays test whether a CDK complex can phosphorylate a substrate under controlled conditions. ATP consumption, phosphate incorporation, or formation of a phosphorylated product can be measured. Such experiments establish biochemical capability, but high enzyme concentrations and simplified conditions can permit interactions that are uncommon inside cells. Demonstrating a physiological substrate requires additional evidence.
Genetic methods include gene deletion, conditional inactivation, RNA interference, and CRISPR-based perturbation. These approaches reveal requirements over different timescales, but cells may compensate for chronic loss of a kinase. Acute inhibitors or inducible protein degradation can expose immediate functions that are masked after prolonged adaptation. Chemical-genetic strategies provide another route: an engineered “analog-sensitive” kinase can accept a bulky inhibitor or nucleotide analog that interacts poorly with most unmodified kinases.
Large-scale proteomics and mass spectrometry identify phosphorylation sites whose abundance changes when CDK activity is perturbed. Interpretation requires care because a disappearing phosphosite may be an indirect consequence of altered phosphatase activity, protein abundance, or cell-cycle position. Strong substrate assignments often combine rapid perturbation, direct phosphorylation in vitro, appropriate docking interactions, and functional testing of the modified residue. Replacing serine or threonine with alanine can prevent phosphorylation, but substitutions with acidic residues do not necessarily reproduce a true phosphate group.
Live-cell reporters have changed understanding of cell-cycle commitment. Fluorescent sensors can report CDK activity through phosphorylation-dependent changes in localization or molecular conformation. Other reporters mark the degradation of cell-cycle-regulated proteins. Combined with time-lapse microscopy, these tools show that genetically similar cells can differ in whether and when they re-enter the cycle after division. A population-average increase in CDK activity may therefore conceal distinct single-cell trajectories.
A central mechanistic question concerns how partly overlapping kinases generate an ordered series of events. One explanation emphasizes qualitative specificity: different cyclins recruit different substrates. Another emphasizes quantitative thresholds: some substrates respond at low CDK activity, whereas others require higher activity. These mechanisms can coexist. Multisite phosphorylation, phosphatase opposition, substrate localization, and feedback can create distinct transition points even when the same catalytic subunit acts throughout much of the cycle.
Other unresolved issues concern the organization of transcription-associated CDKs, the division of labor between closely related family members, and the extent to which experimental results transfer across tissues. High-resolution structures reveal possible molecular interactions, while cellular experiments establish when those interactions occur and whether they matter. Distinguishing direct catalytic effects from changes caused by altered cell state remains essential, particularly when interpreting drug responses, genome-wide phosphorylation data, or long-term genetic perturbations.