p53 signaling — stress-kinase stabilisation of p53 (ATM/ATR, CHK2/CHK1), MDM2-MDM4 ubiquitin-ligase negative feedback, ARF inhibition of MDM2, acetylation control, and p53 tetramer transactivation of target genes; TP53/MDM2/MDM4/CDKN2A(ARF)/ATM/ATR/CHEK2/CHEK1/EP300/CREBBP/KAT5/SIRT1

The tumour suppressor p53 (TP53) is a sequence-specific transcription factor that is kept at low levels in unstressed cells by the RING E3 ubiquitin ligase MDM2. MDM2 binds the N-terminal transactivation domain of p53, which both conceals the domain from the transcription machinery and positions p53 for polyubiquitylation and proteasomal degradation. The MDM2 paralog MDM4 (MDMX) has no ligase activity of its own toward p53 but heterodimerises with MDM2 through the RING domain, stimulating MDM2 E3 activity, and also binds and inhibits the p53 transactivation domain. Because the MDM2 gene is itself a p53 target, the circuit is a negative feedback loop. Stress releases p53 from this loop in two main ways. DNA damage activates the PI3K-like kinases ATM (double-strand breaks) and ATR (replication stress and UV damage), which phosphorylate p53 on Ser15 and MDM2, and activate the effector kinases CHK2 and CHK1, which phosphorylate p53 on Ser20 in the MDM2-binding region; these modifications weaken the p53-MDM2 interaction and stabilise p53. Oncogenic signalling induces the ARF product of the CDKN2A locus, which binds MDM2, inhibits its E3 activity and sequesters it in the nucleolus. Stabilised p53 binds its response elements as a tetramer and recruits the coactivator acetyltransferases p300 and CBP, which acetylate the p53 C-terminal domain as well as promoter histones; KAT5 (TIP60) acetylates p53 Lys120 in the DNA-binding domain, a mark needed for induction of pro-apoptotic targets, and the NAD-dependent deacetylase SIRT1 removes p53 acetylation and dampens its activity. The transcriptional outputs, such as CDKN1A (p21), BBC3 (PUMA), PMAIP1 (NOXA) and BAX, then execute cell-cycle arrest, apoptosis or senescence through modules downstream of this circuit.

Derived from the RGD Pathway Ontology diagram PW:0000718 (p53 signaling pathway) through the RGD_PATHWAY_ONTOLOGY project's diagram->module route (seed in projects/RGD_PATHWAY_ONTOLOGY/data/module_seeds.yaml, tier related_module_only). The seed's only existing related module was c_elegans_core_apoptosis, matched because it uses a descendant of the mapped GO term; it models CEP-1-dependent germline apoptosis in C. elegans and is not a human p53 circuit, so it is neither referenced nor duplicated here. The seed's best-overlap GO-CAM (gomodel:65a1f4f800001246, lactylation of MRE11 by CREBBP and of NBN by KAT5 in homologous recombination) shares only DNA-repair genes with the pathway and is not linked. The diagram draws only three genes (Mdm2, Mdm4, Tp53); RGD curates 40 human genes to PW:0000718: ABL1 ATM BRCA1 CHEK1 CHEK2 COP1 CREBBP ETS1 HDAC1 KAT2A KAT2B KAT5 KAT8 MDM2 MDM4 MRE11 MUC1 NFYA NFYB NFYC PIN1 POU4F1 POU4F2 PPP1R13B PPP1R13L RCHY1 SETD7 SIRT1 SMYD2 SREBF2 TP53 TP53BP2 TP63 TP73 USP7 VDR YBX1 ZNF385A ZNF420 ZNHIT1. Nine are modelled as participants (TP53, MDM2, MDM4, ATM, CHEK1, CHEK2, CREBBP, KAT5, SIRT1). Three gene products not on the RGD list are added because the circuit cannot be drawn without them: ATR (the replication-stress kinase upstream of CHK1), ARF (the CDKN2A alternative-reading-frame product, the oncogenic-stress inhibitor of MDM2), and EP300 as the paralog shown by Gu and Roeder to acetylate p53 (grouped with CREBBP in one family annoton). Molecular functions come from each protein's experimental GO annotations in QuickGO (November 2026): TP53 GO:0001228 (IDA), MDM2 GO:0061630 (IDA/IMP/EXP), MDM4 GO:0008047 (IDA, with extension has_input MDM2 and directly positively regulates ubiquitin protein ligase activity), ARF GO:1990948 (IDA), ATM, ATR and CHEK2 GO:0004674 (IDA), CHEK1 GO:0106310 (EXP/IDA), KAT5 and p300/CBP GO:0061733 (IDA), SIRT1 GO:0034979 (IDA). Gene reviews exist for TP53, CDKN2A (p16 entry P42771) and CDKN2A__Q8N726 (ARF), ATM, CHEK1, CHEK2, CREBBP, EP300, SIRT1, COP1 and ABL1. Substrate versus participant (CLAUDE.md). p53 target genes are transcriptional outputs of the circuit, not participants in it. CDKN1A, BAX, BBC3 (PUMA), PMAIP1 (NOXA) and GADD45A are therefore not annotons; their products act in downstream modules referenced as optional output parts (bcl2_family_momp_control#bh3_only_activation for the BH3-only proteins and BAX; g1_s_transition#g1s_cdk_activation for p21 inhibition of G1/S cyclin-CDK). The comparator check agrees with this convention: in human GOA, BBC3, BAX and GADD45A carry no annotation to GO:0072331 or its descendants, and CDKN1A carries only IEA/TAS rows to GO:0030330, whereas the circuit components MDM2 (IMP), KAT5 (IDA) and SIRT1 (IMP, GO:0042771) carry experimental ones. PMAIP1 carries GO:0072332 (IMP), which is the apoptotic output process rather than the p53 signal itself. MDM2 is the exception the rule allows: it is a p53 target, but its product does work in the circuit (E3 ligase activity on p53), so it is a participant, and the p53 -> MDM2 transcriptional edge closes the feedback loop. p53 itself is a substrate of every enzyme in parts 2-4 and 6-7, but it is also the pathway's transcription factor, so it is the central participant. Two human GO-CAMs do model target-gene products as downstream activities (BAX channel activity in gomodel:69b3372b00001963; CDKN1A CDK inhibitor activity in gomodel:69d8496c00000318, p21 induction via HIPK2); that is consistent with treating them as outputs causally downstream of the p53 activity. Boundary decisions. (1) ATM activation at double-strand breaks is not re-modelled: part 1 is a module_ref to atm_dsb_signaling#atm_activation and is optional, because p53 also responds to ATR-dependent and ARF-dependent stress. Only ATM's own action on p53 and MDM2 is an annoton here. MRE11 and BRCA1, on the RGD list, belong to that module. ATR activation (ATRIP, TOPBP1, 9-1-1) is out of scope; no human ATR module exists (dmel_atr_atrip_complex and dmel_9_1_1_checkpoint_clamp are fly modules). (2) Ribosomal-protein inhibition of MDM2 (RPL5, RPL11, nucleolar stress) is named in the ARF part description but not modelled. (3) Further p53 ubiquitin-system components on the RGD list are not modelled: COP1 and RCHY1 (Pirh2) are additional p53 E3 ligases (RCHY1 is itself p53-induced, a second feedback loop), and USP7 (HAUSP) deubiquitinates both p53 (PMID:11923872) and MDM2, so its net effect is context dependent. They are candidates for a later extension of part 3. (4) Other p53 modifiers on the RGD list are not modelled to keep the core circuit legible: KAT2A and KAT2B (PCAF; Lys320 acetylation, in the p21 induction GO-CAM gomodel:69c59f8a00000219), KAT8 (MOF; Lys120, redundant with KAT5 and named in the KAT5 annoton), SETD7 (Lys372 methylation), SMYD2 (repressive Lys370 methylation), HDAC1 (deacetylation with NuRD/MTA2), PIN1 (phosphorylation-dependent prolyl isomerisation) and ABL1 (damage-induced phosphorylation of MDM2 and MDM4). (5) The ASPP proteins PPP1R13B (ASPP1), TP53BP2 (ASPP2) and PPP1R13L (iASPP) modulate p53 promoter selection toward or away from apoptotic targets; they are context regulators of target choice, not part of the stabilisation circuit. (6) Gene products that bind p53 or its target promoters as transcriptional partners or regulators of TP53 expression (NFYA, NFYB, NFYC, ETS1, YBX1, VDR, SREBF2, POU4F1, POU4F2, MUC1, ZNF385A, ZNF420/Apak, ZNHIT1) are peripheral and not modelled. (7) The p53 paralogs TP63 and TP73 share response elements but have their own developmental and DNA-damage roles; they are not variants of this human p53 module. (8) Repression of RNA polymerase I and III transcription, non-coding RNA targets and regulators, and p53 transcription-independent mitochondrial apoptosis are drawn or described by RGD but left out. (9) The RGD diagram links PW:0000104 (intrinsic apoptotic pathway), PW:0000127 and PW:0001274 (RNA polymerase II transcription and its initiation) and PW:0001361 (ATM signaling pathway). These are mentioned here, reached through module_ref or as context, and are not listed as evidence sources; PW:0001361 is the evidence source of atm_dsb_signaling. (10) The module concept is GO:0072331 (the curated closeMatch), not its DNA-damage child GO:0030330, because ARF-driven activation is not a DNA damage response. GO-CAM and GOA observations for upstream review. In human GOA, CHEK1's experimental MF rows use GO:0106310 protein serine kinase activity while CHEK2, ATM and ATR use GO:0004674; both are correct for serine kinases, but the split is a curation inconsistency rather than a biological difference. The MDM4 enzyme activator annotation (GO:0008047) rests on PMID:25384516, a paper about RAD54B as a scaffold for the MDM2-MDMX ligase, rather than on the primary MDM2-MDMX heterodimer studies. The ARF GO:1990948 IDA comes from PMID:18560357 (a ribosomal protein L5/L11 study) rather than from the original ARF-MDM2 papers. The p53 GO-CAM activities are typed inconsistently across models (GO:0003700, GO:0000981, GO:0001228 and GO:1990841 for the same p53 role). TP53 and CHEK1 gene reviews do not list GO:0001228 or GO:0106310 in core_functions (module validator conformance warnings). PANTHER family ids and PTNs are deliberately not asserted.