Mammalian nucleotide excision repair — GG-NER (XPC-RAD23B-CETN2, UV-DDB/CRL4-DDB2) and TC-NER (stalled RNA Pol II, CSB, CRL4-CSA, UVSSA, ELOF1) damage recognition, TFIIH opening, XPA-RPA verification, XPF-ERCC1 and XPG dual incision, PCNA/RFC-dependent repair synthesis by Pol delta, epsilon or kappa, and ligation by LIG1 or LIG3-XRCC1

Nucleotide excision repair (NER) removes a wide range of structurally unrelated, helix-distorting DNA lesions, most prominently the UV photoproducts (cyclobutane pyrimidine dimers and 6-4 photoproducts) and bulky chemical adducts, by excising a 24-32 nucleotide single-stranded oligonucleotide containing the lesion and resynthesising the gap. Mammalian NER has two damage-recognition sub-pathways that converge on a common core. In global genome NER (GG-NER), the XPC-RAD23B-centrin 2 complex detects the unpaired bases opposite a lesion anywhere in the genome; for CPDs, which distort the helix only slightly, the UV-DDB complex (DDB2 within the DDB1-CUL4-RBX1 ubiquitin ligase) binds the lesion first, ubiquitylates XPC, histones and itself, and hands the lesion over to XPC. In transcription-coupled NER (TC-NER), RNA polymerase II stalled at a lesion in the transcribed strand is the sensor: CSB (ERCC6), a SWI2/SNF2-family DNA translocase, binds the arrested polymerase and recruits the CSA (ERCC8) substrate receptor of a second DDB1-CUL4-RBX1 ligase (CRL4-CSA); CSA brings UVSSA, and ELOF1 positions UVSSA and CRL4-CSA so that RPB1 is ubiquitylated on Lys1268. Both sub-pathways then recruit the ten-subunit transcription factor TFIIH, in GG-NER through XPC and in TC-NER through UVSSA. The XPB (ERCC3) ATPase anchors TFIIH and opens the DNA, and the XPD (ERCC2) 5'-3' helicase extends the bubble to about 30 nucleotides and scans the strand to verify the lesion; the CDK-activating kinase subcomplex leaves TFIIH during this step. XPA and the single-stranded DNA-binding protein RPA bind the opened intermediate, verify the damaged strand and position the two structure-specific endonucleases: XPF-ERCC1 incises 5' to the lesion and XPG incises 3' to it. The 5' incision lets RFC load PCNA, and DNA polymerase delta, epsilon or kappa fills the gap; the 3' incision by XPG then releases the excised oligonucleotide. DNA ligase I or the DNA ligase III-XRCC1 complex seals the remaining nick. Inherited defects in NER proteins cause xeroderma pigmentosum, Cockayne syndrome, trichothiodystrophy and UV-sensitive syndrome.

Derived from the RGD Pathway Ontology diagram PW:0000130 (nucleotide excision repair 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 because the GO term is shared with the prokaryotic module). RGD curates 39 human genes to PW:0000130: CETN2 DDB1 DDB2 ERCC1 ERCC2 ERCC3 ERCC4 ERCC5 ERCC6 ERCC8 GTF2H1 GTF2H2 HMGN1 LIG1 LIG3 PCNA POLD1 POLD2 POLD3 POLD4 POLE POLE2 POLE3 POLE4 POLK RAD23B RFC1 RFC2 RFC3 RFC4 RFC5 RPA1 RPA2 RPA3 RPA4 XAB2 XPA XPC XRCC1. Thirty-six are modelled as participants; HMGN1, XAB2 and RPA4 are not (decision 6). Nine gene products not on the RGD list are added because the steps RGD draws cannot be stated without them: the remaining TFIIH core subunits GTF2H3, GTF2H4 and GTF2H5 (RGD draws "TFIIH complex" but lists only Gtf2h1 and Gtf2h2), the cullin and RING subunits CUL4A, CUL4B and RBX1 of the two CRL4 ligases (RGD draws Ddb1-Ddb2 and Ercc8 without the ligase core), and, for TC-NER, the stalled RNA polymerase II (represented by POLR2A), UVSSA and ELOF1, which the human TC-NER GO-CAM models. Molecular functions were taken from each protein's experimental GO annotations in QuickGO (November 2026) and, where a gene review exists (DDB1, DDB2, ERCC2, ERCC3, ERCC4, ERCC5, ERCC6, ERCC8, CUL4B, RBX1, POLD2, POLE4), from its core functions: XPB is typed DNA translocase activity and CSB DNA translocase activity as in their reviews, and XPF-ERCC1 ds/ssDNA junction-specific dsDNA endonuclease activity as in the ERCC4 review. Deliberate departures: PCNA carries DNA polymerase processivity factor activity, which has no experimental human GOA row; LIG1 is typed DNA ligase (ATP) activity, the child of its experimental DNA ligase activity; RNA polymerase II carries its transcription activity, not the GO:0140612 DNA damage sensor activity used for XPC. That term describes a molecule that recognises a DNA structure and initiates a response, as XPC does by binding the helix distortion. The polymerase recognises nothing: it detects the lesion only by stalling while it transcribes, and no human GOA row puts GO:0140612 on RNA polymerase II. Subunits with no experimental molecular function relevant to NER (RAD23B, CETN2, GTF2H1-5, RPA3, RFC1-5, POLD2-4, POLE2-4, XRCC1) are units without a function; their roles are given in prose. RAD23B has an IDA row for DNA damage sensor activity (PMID:19941824), which is the activity of the XPC complex and is recorded on XPC and on the complex annoton. UniProt labels follow the current recommended names (UniProt release 2026_04, 4 November 2026), including the recent "Lysine-specific demethylase RAD23B" for P54727, the UV excision repair protein RAD23 homolog B. Boundary decisions. (1) Sub-pathways as a ROUTE variant set. GG-NER and TC-NER differ only in lesion recognition and TFIIH recruitment, so they are two variants of part 1 (selection ONE_OR_MORE, since both operate in the same cell); everything from TFIIH onward is shared core. In GG-NER, UV-DDB is modelled as an annoton within the GG-NER variant rather than as a separate required step because it is dispensable in the reconstituted reaction (PMID:7697716) though needed in cells for CPD repair. (2) TFIIH is one complex annoton whose active units are the seven core subunits. Only XPB and XPD carry functions (DNA translocase and 5'-3' helicase); p62, p44, p52, p34 and p8 are units without a GO function, with their NER roles (XPC/UVSSA binding, XPD and XPB stimulation, stabilisation) in prose. The CDK-activating kinase subcomplex (CDK7, CCNH, MNAT1), which completes the ten-subunit holo-TFIIH that RGD describes, is excluded: it is released from TFIIH during NER and its kinase is a transcription activity; RGD does not list it either. (3) Repair synthesis is modelled here rather than referenced. modules/ was searched for human DNA replication, PCNA, RFC, clamp loader, polymerase and ligase modules: the only related modules are other-taxon ones (dmel_dna_polymerase_complexes, dmel_elg1_rfc_like_complex, dmel_9_1_1_checkpoint_clamp, bacterial_chromosomal_dna_replication), so there is nothing human to module_ref. The repair-synthesis and ligation parts are kept to the proteins RGD draws (PCNA, RFC, Pol delta, Pol epsilon, Pol kappa, LIG1, LIG3-XRCC1); when a human replisome or Okazaki-fragment module exists, the Pol delta, Pol epsilon, RFC and PCNA annotons should become module_refs. CTF18-RFC (CHTF18-DSCC1-CTF8 with RFC2-5), which loads PCNA for Pol epsilon in NER (PMID:20227374), and RAD18-dependent PCNA monoubiquitination for Pol kappa are named here but not modelled. (4) RPA is the canonical RPA1-RPA2-RPA3 trimer. RPA4, the RPA2 paralogue of the alternative RPA, is on the RGD list but there is no cached evidence for its participation in NER, so it is excluded. (5) GO's NER incision terms (5'- and 3'-to-lesion, and the parent DNA incision term) are obsolete, so the dual-incision node has no process concept and its annotons carry nucleotide-excision repair. (6) HMGN1 and XAB2 are excluded. RGD draws both associated with Ercc8. HMGN1 is a nucleosome-binding chromatin-unfolding protein recruited in TC-NER through CSA, and XAB2 is a core spliceosomal (Prp19-associated) protein that binds XPA, CSA and CSB; neither has an experimental molecular function in human GOA that would place it as a performer of an NER step, and neither is in the human TC-NER GO-CAM, so both are left as candidate accessory factors. The diagram's link to PW:0001337 (chromatin modification/remodeling pathway) likewise covers the "access-repair-restore" chromatin context, which is out of scope; so are the deubiquitinases USP7 (via UVSSA) and USP44/USP24 (stabilising DDB2), PARP1/PARP2 recruitment to DDB2, and the post-repair restart of transcription. (7) Substrate versus participant: RNA polymerase II is a participant, not merely the ubiquitylation substrate of CRL4-CSA, because its own translocation detects the lesion and the arrested complex is the platform the TC-NER factors assemble on (the stalled polymerase is the sensor in RGD's diagram and in PMID:9734359). XPC is ubiquitylated by UV-DDB but is likewise a participant through its own sensor activity. (8) The prokaryotic analogue is modules/bacterial_nucleotide_excision_repair (UvrA-UvrB-UvrC with Mfd for transcription coupling). The two systems share the logic of recognition, verification by a helicase, dual incision and resynthesis but no homologous core proteins, so it is not referenced as a module_ref. GO-CAM links. gomodel:6796b94c00002938 (TC-NER by ERCC6, ERCC8, ELOF1 and UVSSA) grounds the TC-NER variant activity by activity. gomodel:6348a65d00002206 (USP44 stabilisation of DDB2) grounds the DDB2 and XPC annotons; its USP44 activity is not modelled here. The seed's best_gocam is this USP44 model (overlap 2), and it is relevant. In gomodel:617b481400000521 (KAT5 acetylation of XPF) the XPF activities are enabled by the non-displayed isoform Q92889-2 rather than the canonical sequence; the module links only the endonuclease activity. Observations for upstream review: in gomodel:6796b94c00002938 CSB is typed chromatin-protein adaptor activity whereas its gene review's core activity is DNA translocase activity, and UVSSA is typed RNA polymerase inhibitor activity rather than the TFIIH-recruiting chromatin-protein adaptor activity that carries its IDA in human GOA (PMID:32355176); the KAT5 model's isoform-level enabled_by for XPF may be unintended. PANTHER family ids and PTNs are deliberately not asserted.