Regulation of CDK Activity

A cyclin-dependent kinase is a serine/threonine kinase that is catalytically inert on its own. Binding to a cyclin subunit — whose concentration oscillates through the cell cycle — repositions a structural element called the T-loop (or activation loop) away from the active site, partially activating the enzyme and creating a docking surface for substrates. But partial activation from cyclin binding alone is not the whole story: a second, independent layer of control determines whether the assembled cyclin-CDK dimer is fully active, held in reserve, or destroyed outright. That second layer is the subject of this article. Three regulatory mechanisms act on cyclin-CDK complexes, and it is essential to keep them conceptually distinct even though they cooperate in the same decisions. First, phosphorylation of the CDK subunit itself can be activating (a single phosphate on the T-loop threonine, added by CDK-activating kinase, CAK) or inhibitory (phosphates on residues near the ATP-binding pocket, added by Wee1/Myt1 kinases and removed by Cdc25 phosphatases). Second, CDK inhibitor proteins (CKIs) are separate polypeptides that bind cyclin-CDK complexes directly and block kinase activity or substrate access; mammalian cells use two structurally distinct CKI families (INK4 and Cip/Kip), while budding yeast uses a single major inhibitor, Sic1. Third, ubiquitin-mediated proteolysis targets cyclins (and CKIs, and other cell-cycle regulators) for destruction by the 26S proteasome, using two dedicated E3 ubiquitin ligase complexes — the SCF (Skp1-Cullin-F-box) complex and the anaphase-promoting complex/cyclosome (APC/C) — that act at different points in the cycle. These mechanisms are not redundant; they solve different problems. Phosphorylation is fast and fully reversible, ideal for holding a complex "poised" and then flipping it on abruptly. CKIs provide a stoichiometric brake that can be titrated by controlling CKI abundance. Ubiquitin-mediated destruction is deliberately irreversible — once a cyclin is degraded, that particular activity cannot simply be switched back on; a new cyclin must be synthesized from scratch. This asymmetry between reversible and irreversible control is what allows the cell cycle to have both a "hold" state (accumulate cyclin-CDK in an inactive, phosphorylated form, waiting for a signal) and a genuine point of no return (destroy the cyclin, and mitosis cannot simply reverse).

Step 1 — Cyclin binding and partial activation. A newly synthesized cyclin binds its CDK partner. This binding event repositions the T-loop and the PSTAIRE helix (in CDK1/CDK2) so that ATP can be properly oriented in the catalytic cleft, but the complex is still only weakly active.

Step 2 — CAK phosphorylation locks in activity. CDK-activating kinase (in vertebrates, itself a cyclin-CDK complex — cyclin H–CDK7, part of the TFIIH transcription factor complex) phosphorylates a conserved threonine on the T-loop (Thr161 in human CDK1; Thr160 in CDK2). This phosphate stabilizes the active conformation and is required for full kinase activity. Unlike the other regulatory phosphorylations described below, CAK phosphorylation is generally constitutive — CAK activity does not oscillate much across the cycle — so this step is necessary but not, by itself, the timing switch.

Step 3 — Inhibitory phosphorylation holds the complex in reserve. Even with CAK phosphorylation in place, Wee1 kinase (and, in vertebrates, the related Myt1) adds inhibitory phosphate groups to a tyrosine (Tyr15 in human CDK1) and, for Myt1, also a neighboring threonine (Thr14). These phosphates distort the ATP-binding site so the enzyme cannot properly bind or hydrolyze ATP, even though cyclin is bound and the activating T-loop phosphate is present. This is the crucial "safety catch": a cell can have abundant, CAK-phosphorylated cyclin B–CDK1 sitting in the cytoplasm, fully assembled, and still be catalytically silent because Tyr15 is phosphorylated.

Step 4 — Cdc25 phosphatase removes the brake. Cdc25 phosphatases (Cdc25A, B, C in humans) remove the inhibitory phosphates from Tyr15 (and Thr14), instantly converting the held complex into an active kinase. Cdc25 itself is activated by phosphorylation (partly by the very CDK1 it activates, and by other kinases such as Polo-like kinase 1), creating a double-negative feedback loop: active CDK1 phosphorylates and activates more Cdc25, and active CDK1 also phosphorylates and inhibits Wee1. Both feedback loops push in the same direction — toward full CDK1 activation — once a threshold is crossed. This positive-feedback architecture converts what would otherwise be a graded, noisy signal into an abrupt, switch-like, nearly all-or-none activation of CDK1 at the G2/M transition. Once enough cyclin B–CDK1 becomes active, it rapidly inactivates all the Wee1 in the cell and activates all the Cdc25, flipping the entire cellular pool of CDK1 to the active state within minutes.

Step 5 — CDK inhibitor proteins impose an independent brake. In parallel with phosphorylation control, small inhibitor proteins bind cyclin-CDK complexes directly.

• In mammals, the INK4 family (p16^INK4a, p15^INK4b, p18^INK4c, p19^INK4d) binds CDK4 and CDK6 alone (before cyclin D binds), distorting the cyclin-binding interface and preventing cyclin D association altogether.
• The Cip/Kip family (p21^Cip1, p27^Kip1, p57^Kip2) binds already-assembled cyclin-CDK complexes (cyclin E–CDK2, cyclin A–CDK2, and at high concentration cyclin D–CDK4/6) and inserts a segment into the catalytic cleft, blocking ATP orientation.
• In budding yeast, a single stoichiometric inhibitor, Sic1, binds and inhibits cyclin B (Clb)–CDK1 (Cdc28) complexes, preventing premature entry into S phase until Sic1 itself is destroyed.

CKI levels are themselves controlled transcriptionally (p21 is a direct transcriptional target of p53, linking DNA damage signaling to CDK inhibition) and post-translationally by ubiquitin-mediated degradation.

Step 6 — Ubiquitin-mediated destruction makes inactivation irreversible. Two E3 ubiquitin ligase complexes provide the "point of no return" for CDK regulation.

• The SCF complex (Skp1–Cullin1–F-box protein, with Rbx1) is active throughout the cycle but selects substrates via F-box proteins that recognize specific phosphodegrons. SCF^Skp2 targets phosphorylated p27 for destruction at the G1/S transition, removing the brake on cyclin E–CDK2 and allowing S-phase entry. SCF also targets Sic1 in yeast (once Sic1 is phosphorylated by cyclin E-analog Cln–CDK complexes), and targets G1/S cyclins for degradation once their job is done.

• The anaphase-promoting complex/cyclosome (APC/C) is activated at the metaphase-to-anaphase transition. Bound to its activator subunit Cdc20 (APC/C^Cdc20), it ubiquitinates securin, whose destruction releases the protease separase to cleave cohesin and trigger sister chromatid separation, and it ubiquitinates cyclin B, whose destruction inactivates CDK1 and allows mitotic exit. Later in the cycle, APC/C switches to a second activator, Cdc20's paralog Cdh1 (APC/C^Cdh1), which keeps cyclin levels low through G1, preventing premature S-phase entry until cyclins re-accumulate at the next G1/S transition.

Because ubiquitination targets the cyclin subunit itself for irreversible proteolysis by the 26S proteasome, this mechanism differs fundamentally from phosphorylation-based control: dephosphorylating Tyr15 can turn CDK1 back on in seconds, but once cyclin B is degraded, CDK1 activity can only be restored by synthesizing an entirely new cyclin B molecule. This irreversibility is what makes mitotic exit a true commitment rather than a reversible pause.

Rao and Johnson (1970s) — cell fusion reveals dominant cytoplasmic activities. Potu Rao and Robert Johnson fused mammalian cells arrested in different cell-cycle phases and examined the fate of the resulting heterokaryons. When an M-phase cell was fused with a G1, S, or G2 cell, the interphase nucleus was forced prematurely into a mitotic-like state — chromatin condensed into visible chromosomes, even a G1 nucleus with unreplicated DNA. This showed that M-phase cytoplasm contains a diffusible, dominant activity capable of driving another nucleus into mitosis regardless of its own cycle stage — direct evidence for a soluble "maturation/mitosis-promoting factor" rather than a purely nucleus-autonomous clock.

Masui and Markert (1971) — MPF in frog oocytes. Yoshio Masui and Clement Markert showed that cytoplasm from a Xenopus oocyte induced to mature (by progesterone) could be injected into an immature oocyte and trigger it to mature as well — condensing chromatin and resuming meiosis — even without new protein synthesis in the recipient at the moment of injection. They named the activity maturation-promoting factor (MPF). This was later shown to be the same activity responsible for mitotic entry in somatic cells (hence its dual name, M-phase-promoting factor), and it was ultimately identified biochemically as cyclin B bound to CDK1.

Hartwell (1970s) and Nurse (1970s–80s) — genetic identification of CDC genes. Leland Hartwell isolated temperature-sensitive cell division cycle (cdc) mutants in budding yeast that arrested at specific points of the cycle at the restrictive temperature, defining genes required for "start," DNA replication, and mitosis. Paul Nurse, working in fission yeast, identified cdc2 as essential for entry into both S phase and mitosis, and showed that fission yeast cdc2 and budding yeast CDC28 were functionally interchangeable — and that the human homolog (later named CDK1) could rescue a yeast cdc2 mutant, demonstrating deep evolutionary conservation of the core cell-cycle engine across roughly a billion years of divergence. Nurse also identified wee1, whose loss-of-function mutants divide at abnormally small size because cells enter mitosis prematurely — the original genetic evidence for an inhibitory brake on CDK1 activity, later shown biochemically to be the Tyr15 kinase.

Murray and Kirschner (1989) — cyclin destruction is necessary for mitotic exit. Andrew Murray and Marc Kirschner used cell-free Xenopus egg extracts that cycle in and out of mitosis, driven by endogenous cyclin B synthesis and destruction, even in the absence of nuclei or chromosomes. They showed that adding a non-degradable form of cyclin B (lacking the N-terminal "destruction box" motif required for ubiquitination) caused extracts to arrest permanently in a mitotic state — chromosomes condensed and never decondensed, spindle never disassembled. This was direct, causal proof that cyclin B destruction (not just its dephosphorylation or dilution) is required to exit mitosis, and it established the destruction box as the molecular signal recognized by the ubiquitination machinery (later identified as the APC/C).

Evans, Rosenthal, Youngblom, Distel, and Hunt (1983) — the discovery of cyclin. Tim Hunt and colleagues, studying protein synthesis in fertilized sea urchin and clam embryos, noticed a protein band on gels whose abundance rose steadily through interphase and then abruptly disappeared at each cell division, in near-perfect synchrony with the cleavage cycles. They named it "cyclin" for this oscillating behavior — a discovery made essentially by careful, repeated observation of one gel lane over time, and one of the cleanest examples in cell biology of a hypothesis (there must be a diffusible factor whose periodic destruction times cell division) generated directly from the data. Together, these independent lines of evidence — cell fusion physiology, oocyte injection, yeast genetics, and cell-free biochemistry — converged on the same molecular picture: a cyclin-CDK complex whose activity is switched on by dephosphorylation/CAK action and switched off irreversibly by ubiquitin-dependent cyclin destruction.

Example 1 — Predicting phenotypes of Wee1 and Cdc25 mutants in fission yeast. Fission yeast (Schizosaccharomyces pombe) divide by fission at a fairly fixed cell length, so cell size at division is a convenient readout of when CDK1 (Cdc2) becomes active.

• wee1⁻ (loss-of-function) mutants: without Wee1 to phosphorylate Tyr15, CDK1 activates as soon as cyclin B accumulates to a threshold, without waiting for the normal growth period. Prediction: cells enter mitosis too early, at abnormally small size. This matches Nurse's original wee1 phenotype (hence the gene's name — mutant cells are "wee," i.e., small).
• cdc25⁻ (loss-of-function, temperature-sensitive) mutants: without Cdc25 to remove the inhibitory phosphate, Tyr15 stays phosphorylated and CDK1 cannot activate even when cyclin B is abundant. Prediction: cells fail to enter mitosis and keep growing, becoming abnormally elongated, eventually arresting.
• Double mutant wee1⁻ cdc25⁻: since Wee1 activity is absent, Tyr15 is never phosphorylated in the first place, so Cdc25 has nothing to remove — the wee1⁻ phenotype (small cells) is epistatic. This genetic epistasis test is exactly how the antagonistic, opposite-direction relationship between Wee1 and Cdc25 acting on the same residue was established, independent of any biochemistry.

Example 2 — Reasoning through a p53 pathway clinical scenario. A cell line is exposed to ionizing radiation, causing double-strand DNA breaks. In wild-type cells, ATM/ATR kinases phosphorylate and stabilize p53, which transcriptionally activates the CDK inhibitor gene CDKN1A, encoding p21. Suppose you compare three cell lines after identical radiation doses:

Reasoning through line C is the instructive case: it shows that p53 stabilization alone is not sufficient for arrest — the CKI itself (p21) is the actual effector that inhibits cyclin E–CDK2 and enforces the G1 checkpoint. This distinguishes "sensing damage" (p53/ATM/ATR) from "executing arrest" (CKI inhibition of CDK), a distinction with direct clinical relevance: tumors that inactivate TP53 (roughly half of all human cancers carry TP53 mutations) lose this checkpoint and tend to accumulate further mutations, and this same logic foreshadows the more detailed checkpoint circuitry covered in Cell Cycle Checkpoints and Their Logic.

Sir Paul Nurse's career is a study in how yeast genetics, thought to be almost comically remote from human biology, turned out to reveal the literal molecular engine shared by essentially all eukaryotic cells. Working first with fission yeast in the 1970s and 80s at the University of Edinburgh and later the Imperial Cancer Research Fund in London, Nurse identified the cdc2 gene as central to both the G1/S and G2/M transitions — a single gene product controlling two supposedly separate decisions, which was itself a conceptual surprise. His most celebrated single experiment was almost audacious in its simplicity: he introduced a human gene library into fission yeast cells carrying a temperature-sensitive cdc2 mutation, and asked whether any human gene could rescue growth at the restrictive temperature. One did — the human CDK1 gene — demonstrating that the core cell-cycle machinery had been conserved essentially unchanged since the last common ancestor of yeast and humans, over a billion years of evolutionary divergence. This result did more than identify a gene; it collapsed the artificial boundary between "model organism biology" and "human biology" for an entire generation of cell biologists, and it directly enabled the mapping of the human CDK/cyclin/Wee1/Cdc25/CKI network described in this article onto the yeast genetic pathway that first revealed its logic.

CDK regulation is now central to oncology. Many cancers inactivate CKIs directly — CDKN2A, encoding p16^INK4a, is one of the most frequently silenced tumor suppressor loci in human cancer (deleted or methylated in a substantial fraction of melanomas, pancreatic cancers, and gliomas), removing the brake on cyclin D–CDK4/6 and driving unrestrained progression past the restriction point. This biology directly motivated the development of CDK4/6 inhibitors (palbociclib, ribociclib, abemaciclib), now standard-of-care drugs in hormone-receptor-positive, HER2-negative metastatic breast cancer, used precisely because they restore a pharmacological brake on CDK4/6 in tumors that have lost their natural one. Because APC/C-mediated cyclin B destruction is required for mitotic exit, drugs that interfere with mitotic progression (such as the microtubule poisons paclitaxel and vincristine, used in many chemotherapy regimens) exploit the spindle assembly checkpoint's ability to hold APC/C^Cdc20 inactive, keeping cyclin B undegraded and cells arrested in mitosis until they die by apoptosis — a direct clinical application of the ubiquitin-destruction logic described above, and a mechanism explored further under Cell Cycle Checkpoints and Their Logic. In reproductive biology, the original MPF/Wee1/Cdc25 axis discovered in frog oocytes underlies assisted reproductive technologies: understanding what holds vertebrate oocytes arrested in meiotic metaphase II (cytostatic factor acting through APC/C inhibition) has informed in vitro maturation protocols used in human fertility treatment and livestock breeding. And in basic research, Xenopus egg extracts — the same cell-free system Murray and Kirschner used to prove cyclin destruction was required for mitotic exit — remain a standard tool for reconstituting and dissecting cell-cycle transitions biochemically, from chromatin licensing to spindle assembly, decades after the original experiments.

Ask the tutor: How does this phosphorylation/CKI/destruction machinery get wired into the actual checkpoint decisions the cell makes at G1/S, G2/M, and the spindle assembly checkpoint?