UniProt: P46527 · HGNC: CDKN1B · 198 aa · CDI family (Pfam PF02234, InterPro IPR003175)
p27Kip1 is a Cip/Kip-family cyclin-dependent kinase regulator, related to p21/CDKN1A and
p57/CDKN1C. It is not an enzyme and has no chemical substrate; it acts stoichiometrically by
binding cyclin–CDK complexes. It is largely intrinsically disordered in isolation, and the
disorder is functional rather than an artefact: the kinase-inhibitory domain folds on binding
PMID:11790096,
and preforming its helix slows complex formation, so p27 "derives a kinetic advantage from
intrinsic structural disorder" in the authors' phrasing.
Mechanism of inhibition, from the crystal structure of the KID bound to phospho-cyclin A–CDK2
PMID:8684460 — it engages both subunits:
on the cyclin, PMID:8684460 formed by conserved cyclin-box
residues (the MRAIL groove); on the kinase it rearranges the N-terminal lobe and
PMID:8684460
Binding is sequential and this is what sets specificity
PMID:15024385 data
support cyclin-first engagement, and
PMID:15890360
cell-cycle CDKs. The negative control is informative:
PMID:15890360 CDK5–p25 complex, because p25 lacks
the MRAIL motif — hence p27 does not regulate CDK5 in neurons despite binding CDK5 itself.
The same logic explains the Spy1/RINGO escape:
PMID:28666995 — Spy1 lacks the cyclin-binding
site that mediates p27 affinity. This is the basis of the UniProt statement that p27 has little
inhibitory activity on SPDYA-bound CDK2.
The cloning paper is still the cleanest statement of the core activity
PMID:8033212 complexes in vivo and in vitro,
PMID:8033212 previously activated complexes, and
PMID:8033212 into S phase. Substrate-level:
PMID:8033212 cyclin A-Cdk2,
and cyclin D2-Cdk4. Physiological trigger:
PMID:8033212 arrest by TGF-beta
and cell–cell contact.
Independent in-cell readout at a defined CDK2 substrate: p27 blocks CDK2-dependent
phosphorylation of p220/NPAT and histone H4 promoter activation, and is more effective than p21
PMID:19170105.
Adenoviral delivery is pRb-dependent, placing p27 upstream of Rb
PMID:10208428
PMID:10208428
functional pRb — in pRb-deficient cells the CKIs did not arrest growth.
This is why p27 must not be described as a pure inhibitor:
PMID:9106657
D-type cyclins, mostly by decreasing k_off; and
PMID:9106657 the
nucleus. LaBaer et al. drew the adaptor conclusion themselves — the p21 family,
PMID:9106657 adaptor proteins
that assemble and program kinase complexes. Tyr74/Tyr88 phosphorylation is required for CDK4
binding (UniProt SUBUNIT) and can leave CDK4 active. Consequence for curation: the parent term
GO:0019914 cyclin-dependent protein kinase regulator activity is the right shape for this gene
and should not be collapsed into the inhibitor child; both signs are real.
Thr187 / SCF(SKP2)–CKS1. The switch that destroys p27 at G1/S:
PMID:16209941
(Skp1-Cul1-Rbx1-Skp2) complex, with CKS1 obligatory —
PMID:16209941
binding site, and Glu185 inserts into the SKP2–CKS1 interface. Note the direction of the
relationship: p27 is the substrate; the structure gives no evidence of p27 activating the
ligase (see §6).
Ser10 / UHMK1(KIS) and CRM1 export.
PMID:12093740 p27 and
phosphorylates Ser10, promoting PMID:12093740;
PMID:12093740 KIS-siRNA
enhances arrest, and p27-null cells are refractory,
PMID:12093740 critical target.
Export itself is CRM1/Ran-driven: PMID:12529437
transiently cytoplasmic at G1/S;
PMID:12529437 but not by
T187A; the model is that PMID:12529437
cytoplasmic degradation in early G1. p27 is the cargo here, not part of the export machinery.
Thr157/Thr198, AKT/RSK, 14-3-3. AKT
PMID:15057270
p27, and PMID:15057270 localisation
by competing with importin alpha5 for the phospho-NLS.
PMID:17053782 ubiquitin-dependent degradation of free
p27, synergising with the T187 route
PMID:17053782
T187. T198 also matters for motility: PMID:21423803
p27/stathmin interaction.
Tyrosine phosphorylation. A
PMID:17254966
phosphorylated by LYN and BCR-ABL; this does not prevent binding but
PMID:17254966
ejected from the CDK2 active site, restoring partial activity and enabling Thr187
phosphorylation. Src acts similarly:
PMID:17254967
Src transfection reduces p27–cyclin E–CDK2 complexes.
SKP2-independent turnover. In Wnt10b mammary tumours,
PMID:19056892
SCF(SKP2)-mediated degradation; CUL4A
PMID:19056892 and
PMID:19056892.
Again: substrate, not ligase subunit (see §6). Lithium in that work is a GSK3 inhibitor used to
turn Wnt on — a reagent, not a biological stimulus for p27.
RhoA: PMID:19470470 RSK1 transfectants show
PMID:19470470 and loss of actin
stress fibres, all reversed by p27 shRNA. Stathmin: p27 acts through
PMID:15652749
stathmin, and PMID:15652749 inhibits
mesenchymal motility. Note the directional conflict: the RhoA axis makes cytoplasmic p27
pro-migratory, the stathmin axis anti-migratory. I therefore annotated the molecular activity
(small GTPase binding, GO:0031267 — the old "Rho GTPase binding" GO:0017048 is merged into it)
and the cytosolic location, and deliberately proposed no directional migration process term;
raised as a question for experts instead.
MCM7 is a further CDK-independent route:
PMID:16289477, where the p27 C-terminus
PMID:16289477 CDK inhibitor.
Substrate ≠ participant. Three annotations invert the direction of a relationship in which
p27 is the thing being acted on. Applying the participation test from CLAUDE.md (which entity
performs the step?):
GO:1990757 ubiquitin ligase activator activity (EXP, PMID:16209941) → REMOVE. TheGO:0031625 ubiquitin protein
ligase binding, is separately annotated, so nothing is lost.GO:0051168 nuclear export (IMP, PMID:12529437) → REMOVE. CRM1/RanGTP does the work; p27GO:0045732 positive regulation of protein catabolic process (IDA, PMID:19056892) → REMOVE.GO:0031464 Cul4A-RING E3 ubiquitin ligase complex (IDA, part_of) → MODIFY →GO:0031625. part_of asserts subunit composition; CRL4 is CUL4A/B–RBX1–DDB1–DCAF and p27 isSign errors. GO:0045740 positive regulation of DNA replication (Reactome TAS) → MODIFY
→ GO:0008156 negative regulation of DNA replication: p27 appears in "Cyclin A:Cdk2-associated
events at S phase entry" as the inhibitor that must be degraded. Direct evidence for the negative
sign: the MCM7 paper above and S-phase-entry block. GO:2000045 regulation of G1/S transition
(IDA) → MODIFY → GO:2000134 (negative), matching the pRb-dependent arrest data.
Growth vs proliferation. GO:0030308 negative regulation of cell growth and
GO:0045926 negative regulation of growth → MODIFY → GO:0008285. In GO, "growth" is
increase in size/mass; every underlying assay measured division (arrest, DNA synthesis, Rb
phosphorylation, tumour latency). No evidence p27 controls cell size.
Too general in the inhibitor hierarchy. GO:0004860 protein kinase inhibitor activity (IMP)
and GO:0140678 molecular function inhibitor activity (IMP) → MODIFY → GO:0004861. p27 does
not inhibit kinases generally; the cited papers are cyclin–CDK2 experiments.
p53-class DDR. Both GO:0030330 TAS rows → MARK_AS_OVER_ANNOTATED. The p53-responsive CKI
is p21: PMID:19170105
reducing histone H4 mRNA. The annotation comes from pathway membership (p27/p21 CDK2 inactivation
is a sub-event of the p53-dependent G1 checkpoint), not from evidence that p27 transduces damage
signals. Not removed outright because p27 can contribute to arrest downstream of damage.
Overexpression phenotypes. GO:0048102 autophagic cell death (IDA, PMID:12698196) →
MARK_AS_OVER_ANNOTATED: adenoviral p27
PMID:12698196 not apoptosis,
and PMID:12698196
— yet the same construct neither killed nor induced autophagy in normal astrocytes. Dose- and
context-dependent, not a physiological function. GO:0071285 cellular response to lithium ion →
MARK_AS_OVER_ANNOTATED (reagent, see §4). GO:0007165 signal transduction (ARBA) →
MARK_AS_OVER_ANNOTATED: p27 is the endpoint of antimitogenic signalling, not a transducer, and
a root-level term invites bad propagation.
Pleiotropic / tissue-specific process terms → KEEP_AS_NON_CORE, graded consistently: heart
development (×2) and negative regulation of cardiac muscle tissue regeneration (×2), both from the
mouse cardiomyocyte study — PMID:24380855 and
PMID:24380855, where the
organ-level consequence follows because
PMID:24380855
cardiomyocytes; negative regulation of vSMC proliferation
(PMID:19088079);
cellular senescence (Reactome TAS; p16/p21 are the principal senescence CKIs); endosome (UniProt
SL-0101 is a degradative colocalisation with SNX6, and "By similarity" at that).
Protein binding (126 rows). Per the repo policy, GO:0005515 is never ACCEPT/
KEEP_AS_NON_CORE. Graded by partner class: cyclin partners (CCNA1/A2, CCNB1, CCND1/2/3, CCNE1/E2)
→ MODIFY → GO:0030332 cyclin binding; CDK partners (CDK1/2/4/5) → MODIFY →
GO:0019901 protein kinase binding; RHOA → MODIFY → GO:0031267 small GTPase binding
(the paper supports the specific activity); all other partners → REMOVE as uninformative,
explicitly noting that removal concerns the term, not the reality of the interaction.
Deferred to curators (rule: do not overrule an experimental annotation from an abstract-only
cache): GO:0005634 nucleus IDA from PMID:11800646 (cached abstract reports p18INK4C/p14ARF
staining — ACCEPTed anyway, since p27's nuclear localisation is not in doubt and is corroborated by
PMID:12529437 and PMID:15057270); GO:0004861 IDA from PMID:10918569 (PTEN/p27 thyroid paper,
abstract shows only the p27-dependence of growth suppression); GO:0019903 protein phosphatase
binding with PTPN6/SHP-1 (abstract states
PMID:19838216
gene transcription and an SHP-1–PI3K interaction, but no direct p27–SHP-1 complex — kept as
non-core on the curator's reading; note also PPM1H genuinely dephosphorylates Thr187 per UniProt).
IBA rows (GO:0004861, GO:0005634, GO:0005737, GO:2000134; PANTHER:PTN004142838) all ACCEPTed.
Human P46527 appears in its own WITH/FROM for three of them, which is expected: the gene's own
experimental annotation is one of the descendant evidences used to place the IBD node, so this
marks experimental grounding on the target and adds the claim that the function is inherited.
Descendant evidence spans Drosophila dacapo and C. elegans cki-1/cki-2, consistent with a
Cip/Kip-wide function.
GO:0060090 molecular adaptor
activity is the closest and was ACCEPTed, but a cyclin-CDK-complex-assembly term would besuggested_questions rather than as proposed_new_terms.proposed_new_terms: [] — no NEW annotations added. Everything supportable is either already