RPA2 (RPA32) - the middle subunit of replication protein A, whose disordered N-terminus is the phospho-switch of the replication-stress response and whose 43-171 region binds menin

No three-dimensional model is available for this protein, so the map above is the whole picture. That is itself informative: AlphaFold does not model chains this long, and the parts that have been crystallised are held under the entries for the individual domains.

• RPA2 is essential and the RPA complex is required in every dividing cell; nothing in the menin or the AP-MS evidence speaks to a therapeutic window. Do not read this entry as target validation.
• RPA2 phosphorylation is this field's replication-stress readout, so a drug that raises it looks like it is 'working' on a mechanism whether or not the mechanism is the intended one. Any combination trial reading out phospho-RPA2 needs a second, mechanistically independent endpoint.
• If a menin inhibitor frees RPA2 from menin, and if free RPA2 is the species that relieves menin's repression of NF-kB (PMID 28007956), the inhibitor could de-repress NF-kB targets as an off-mechanism consequence. This is speculative - the breast-line axis has never been tested in a leukaemia - but it is the kind of bite-back that a transcriptional inhibitor acquires from a partner's second job.

1. Does menin bind RPA2 in a KMT2A-rearranged blast, and is the contact displaced by revumenib? The menin-binding determinants on menin (N-terminal 1-40 plus 286-448, PMID 12509449) are not obviously the KMT2A-binding pocket the menin inhibitors occupy, so the prediction is NO displacement - which makes it a clean, falsifiable test with an informative negative.
2. The menin-binding region of RPA2, aa 43-171, OVERLAPS its OB fold at 74-148, yet menin had no detectable effect on RPA-DNA binding in vitro (PMID 12509449). Does menin instead compete with an RPA2 PARTNER that uses the same surface - and does it affect the 8-10 nt to 30 nt mode switch, which was not assayed in 2003 and could be now?
3. Which RPA2 phosphosite is the readout in this corpus's proposed resection experiments? S4/S8, T21 and S33 have different kinases and different meanings, and an entry that says 'RPA2 phosphorylation' has not specified an experiment. Rule 74: name the antibody clone.
4. Is RPA2 an E2F1 target in a KMT2A-rearranged cell as it is in breast lines (PMID 28007956)? If so, RPA2 abundance would track proliferation in this disease and any RPA2 level difference between fusion-positive and fusion-negative cells would be confounded by cycling rate.

1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   RPA2 is UniProt P15927 (RFA2_HUMAN), primary gene symbol RPA2, 270 aa, 'Replication protein A 32 kDa subunit', with an OB fold at 74-148, a disordered region at 21-41, a C-terminal region 187-270 annotated 'Interaction with RAD52, TIPIN, UNG and XPA', and six annotated phosphosites - S4, S8, T21 and S33 by PRKDC, S23 by CDK2 and S29 by CDK1 - annotated to Nucleus on five ECO:0000269 attributions (PubMed 10982866, 12814551, 17596542, 20154705, 21504906) and to nuclear PML body on one (PubMed 12814551).

   UniProtKB P15927 (RFA2_HUMAN), Replication protein A 32 kDa subunit - accession, primary gene name, sequence, feature table, PTM and subcellular-location comments, fetched live 2026-09-24 · UniProt Consortium, EMBL-EBI/SIB/PIR 2026
   ▸What this evidence cannot show

   Resolved by accession with the primary gene name read back; sequence md5 BC44D5D5F9A66A44ECF84D8E048C0D9C recorded. EVERY phosphosite position was verified by string match against the fetched sequence - position 4 S, 8 S, 21 T, 23 S, 29 S, 33 S - so the coordinates are asserted against this accession and this release, not from recall. Rule 91: five ECO:0000269 localisation citations mean a curator saw experiments in five papers, not that any is the founding study; one of the five (PubMed 12814551, Curr Biol 2003, ATR-kinase-dependent intranuclear translocation of ATR and RPA after ionising radiation) was checked by title and IS a localisation study, and it is the PML-body attribution. Rule 93 on the symbol: RPA2 is a curated HGNC alias of POLR1B (RNA polymerase I subunit B) and RPA34 is a curated alias of POLR1G, so two of this protein's names point at RNA polymerase I subunits. The kinase attributions (PRKDC for S33) are UniProt's; the literature also attributes S33 to ATR, and this entry does not adjudicate that.

   ▸Evidence that argues against this (1)

   • The PTM list is a curated union across papers and cell systems and must not be read as a phosphorylation state. Which of S4/S8/T21/S23/S29/S33 is occupied is condition-dependent - CDK sites cycle, DNA-PK sites appear on damage - and no single experiment reports all six. A reader using 'RPA2 phosphorylation' as a readout is measuring a specific antibody's epitope, not this list, and rule 74 applies: the clone is the identifier. UniProt Consortium 2026
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   Human Protein Atlas immunofluorescence localises RPA2 to Nucleoplasm and Nuclear bodies with reliability 'Supported'; the same query returns POLR1B carrying RPA2 as a gene synonym, with a different localisation (Nucleoli fibrillar center, Cytosol).

   Human Protein Atlas search_download API, query 'RPA2', fetched 2026-09-24; RPA2 = ENSG00000117748, Reliability (IH) Supported, Subcellular location [Nucleoplasm, Nuclear bodies]; second row POLR1B = ENSG00000125630, gene synonyms including RPA2 and RPA135, Subcellular location [Nucleoli fibrillar center, Cytosol] · Human Protein Atlas, proteinatlas.org 2026
   ▸What this evidence cannot show

   Antibody-based immunofluorescence in fixed cultured lines, not a haematopoietic cell; 'Supported' is HPA's middle reliability tier. The antibody id is not returned by this column set (rule 74). 'Nuclear bodies' in HPA's vocabulary is not the same call as UniProt's 'PML body' and the two should not be merged - HPA's term covers any discrete nuclear focus. The Ensembl id is HPA's own join and was not independently resolved.

   ▸Evidence that argues against this (1)

   • The POLR1B row in the same response is the counter-case and it is the reason this item exists: a search-by-symbol lookup for 'RPA2' returns a nucleolar RNA-polymerase-I subunit as a legitimate hit, with a compartment (nucleoli fibrillar centre) that a reader would find biologically interesting and that belongs to a different protein. This corpus has a nucleolar thread, which makes the confusion live rather than theoretical. HUGO Gene Nomenclature Committee, genenames.org 2026
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Menin interacts with RPA2: found in a Ras-reconstitution yeast two-hybrid system, confirmed in a conventional two-hybrid system and by direct interaction between purified proteins; the interacting regions map to the N-terminal portion of menin and to amino acids 43-171 of RPA2, a region containing RPA2's low-affinity ssDNA-binding site; menin had no detectable effect on RPA-DNA binding in vitro; menin bound preferentially to free RPA2 rather than the heterotrimer or an RPA2-RPA3 subcomplex; RPA2 bound poorly to menin lacking amino acids 1 to 40 but efficiently to menin C-terminally truncated at 448, while removing a further 162 C-terminal residues abolished binding, implicating menin residues between 286 and 448 as well; of fifteen MEN1 missense mutations tested, H139D, A160P and A176P abolished the interaction, P12L, L22R, H139Y, F144V, F159C, L286P, A309P and T344R impaired it, and A164D, W183S, W436R and F447S had little or no effect; and menin showed no detectable mobilisation into replication foci even when punctate RPA2 prereplication foci were visible in cells synchronised at the G1/S border.

   The 32-kilodalton subunit of replication protein A interacts with menin, the product of the MEN1 tumor suppressor gene. · Sukhodolets KE, Hickman AB, Agarwal SK, Sukhodolets MV, Obungu VH, Novotny EA, Crabtree JS, Chandrasekharappa SC, Collins FS, Spiegel AM, Burns AL, Marx SJ · Molecular and Cellular Biology 23(2):493-509 2003
   PMID 12509449DOI
   ▸What this evidence cannot show

   Abstract from PubMed efetch plus a targeted read of the PMC open-access full text for the deletion mapping, the mutation panel and the foci statement; the figures themselves were not inspected and no quantitation was extracted. Two-hybrid plus purified-protein pulldowns plus HeLa co-IP - no KMT2A-rearranged system, no haematopoietic cell, no primary tissue. 2003, before menin's structure was solved, so the deletion mapping is not a structural epitope. The menin residue numbers are asserted against UniProt O00255 canonical (610 aa) because all fourteen distinct wild-type residues string-match that sequence [E4]; the paper does not itself state which isoform numbering it uses, so that match is a reconstruction and is labelled as one. 'Little or no effect' is the paper's own grading and no binding constants are reported for any variant.

   ▸Evidence that argues against this (1)

   • The paper contains its own strongest limitation and it points away from replication: menin does not mobilise into replication foci, so the contact is not a fork event. And the two directions of the species question are unresolved within the paper - free RPA2 preferred in vitro, an RPA1-containing species co-precipitated in cells. Searched for a paper disputing the menin-RPA2 interaction itself and found none; in 23 years it has been restated in reviews (PMID 15153434) and extended functionally (PMID 28007956), never contradicted. That is weak corroboration rather than replication, because nobody appears to have retested the binding independently. Agarwal SK et al. 2004
4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   All fourteen distinct menin residue positions named in the Sukhodolets 2003 mutation panel string-match the wild-type residue in the fetched UniProt O00255 canonical sequence (610 aa, isoform 2 displayed): P12, L22, H139, F144, F159, A160, A164, A176, W183, L286, A309, T344, W436 and F447.

   UniProtKB O00255 (MEN1_HUMAN) fetched live 2026-09-24 by Q192; sequence length 610, isoform O00255-2 displayed, isoforms O00255-1 and O00255-3 described; each of the fourteen positions indexed into the fetched sequence string and the residue printed and compared to the published mutation nomenclature · UniProt Consortium / this repository 2026
   ▸What this evidence cannot show

   A string match on fourteen positions confirms the numbering convention is consistent with this accession; it does NOT prove the paper used this convention, because a 14-of-14 match could in principle arise under a shifted numbering if the shifted residues happened to agree, which is unlikely but not impossible. Menin has a 610-aa displayed isoform and a longer described isoform, and the corpus's ana-q36-menin-residue-axis is the entry that holds the numbering question; read it before doing any further menin residue arithmetic. This check does not validate the BIOLOGY of any mutation, only that the residue identities line up.

   ▸Evidence that argues against this (1)

   • A successful string match is the weakest form of coordinate verification available and is being reported as such. It cannot detect an offset that preserves residue identity, it says nothing about whether the paper's construct matched the canonical isoform, and it is not a substitute for the paper stating its reference sequence - which it does not. The honest status is 'consistent with O00255 canonical numbering', not 'confirmed'. This repository 2026
5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   RPA2 appears as an MLL-AF4-specific prey row in the only published MLL-fusion AP-MS dataset - p = 0.0221, spectral counts 3|3, negative control 0|0, rank 55 of 291, recovered under MLL-AF4 and none of the other six baits - and this corpus has already adjudicated that row and DECLINED it as evidence of a KMT2A::AFF1-RPA2 contact.

   MLL-fusion-driven leukemia requires SETD2 to safeguard genomic integrity. · Skucha A, Ebner J, Schmoellerl J, Roth M, et al. (Grebien F senior author) · Nature Communications 9:1983 2018
   PMID 29777171DOI
   ▸What this evidence cannot show

   The numbers are quoted from this corpus's own reading of Supplementary Table 5 as recorded by Q180, not re-extracted by this beat; that entry states the column map was asserted against the sheet's header row. This item exists to point at the adjudication, not to re-litigate it: interface-kmt2a-rpa2-declined-an-abundant-nuclear-ssdna-binder-on-six-spectra-but-the-resection-thread-is-live is the entry of record. AF4-specificity in this deposit is substantially a statistical-permissiveness artefact - MLL-AF4 has the most permissive list of the seven baits (max_p 0.175) and 110 singletons, Spearman 0.8214 between a bait's max_p and its singleton count, permutation p = 0.0341 over 5,040 orderings, and only 5.45% of AF4 singletons reach p<0.05 with a spectral mean of 3.37 against 18.02 for pan-fusion prey. RPA2's p = 0.0221 puts it in the top ~5% of that tail and its 0|0 control helps, but two replicates at 3 spectra each is low and this item must not be read as an established index-fusion interaction.

   ▸Evidence that argues against this (1)

   • The decline's decisive argument is contradicted by E3 and a reader should see both. Q180 declined partly because RPA is an obligate heterotrimer and a genuine complex association should have brought RPA1 - but Sukhodolets 2003 reports menin binding PREFERENTIALLY TO FREE RPA2, not to the heterotrimer. Since menin is the documented bridge between a KMT2A fusion and anything, single-subunit recovery is what a menin-bridged association predicts. This does not rescue six spectra, and the decline still stands on count, nuclease treatment and RPA2's abundance - but the subunit-coherence leg is weaker than the decline states. Sukhodolets KE et al. 2003
6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   RPA2 is upregulated in breast cancer tissues and cell lines; ectopic RPA2 in MCF7 and MDA-MB-231 promoted proliferation, adhesion, migration, invasion and EMT while ablation in MDA-MB-231 induced apoptosis and suppressed colony formation, EMT and invasion; binding assays indicated menin interacts with RPA2; ectopic RPA2 inhibited formation of the menin-NF-kB p65 complex and repressed menin's inhibitory effect on NF-kB-regulated genes, knockdown did the reverse, and RPA2 expression was induced via an E2F1-dependent mechanism on treatment with TNF-alpha or LPS.

   Upregulation of RPA2 promotes NF-kappaB activation in breast cancer by relieving the antagonistic function of menin on NF-kappaB-regulated transcription. · Chen CC, Juan CW, Chen KY, Chang YC, Lee JC, Chang MC · Carcinogenesis 38(2):196-206 2017
   PMID 28007956DOI
   ▸What this evidence cannot show

   Abstract level. MCF7 and MDA-MB-231 breast cancer lines with overexpression and knockdown rather than endogenous perturbation; not haematopoietic and no fusion. This claim is ALREADY IN THE CORPUS as evidence E5 of gene-rela and is repeated here only because RPA2 now has an entry of its own that must carry it; gene-rela remains the entry of record for the NF-kB axis. 'Binding assays indicated' is the abstract's own hedge and no method is named in it.

   ▸Evidence that argues against this (1)

   • Direction of causation is not established by this design: RPA2 is an E2F1 target and therefore rises with proliferation, so 'RPA2 up, NF-kB up, tumour worse' is equally consistent with RPA2 being a proliferation marker that the overexpression experiment then forces into a functional-looking role. And a breast epithelial NF-kB axis is a long way from infant B-ALL - this corpus's own menin-NF-kB entries are careful about that distance and this item should inherit their caution rather than borrow their confidence. This repository 2026
7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   No paper names RPA2 together with KMT2A or with MLL in a title or abstract, while RPA2[tiab] AND menin[tiab] returns six records, of which three are on topic: the founding interaction paper (PMID 12509449), the breast-cancer functional paper (PMID 28007956) and an in-silico docking study of the menin-RPA2 binding site (PMID 37291166); the other three are a MEN1 expression-profiling paper and a MEN1 review that list RPA2 among menin's partners, and a 2026 menin-HR paper naming the RPA2/Rad51 pathway.

   PubMed esearch via scripts/pubmed_safe.py, run 2026-09-24 by Q192, one [tiab] tag per term (rule 94), every querytranslation read back. RPA2[tiab] AND KMT2A[tiab] -> 0 (genuine zero). RPA2[tiab] AND MLL[tiab] -> 0 (genuine zero, recorded by Q180 2026-09-24 and re-confirmed). RPA2[tiab] AND menin[tiab] -> 6, querytranslation "RPA2"[Title/Abstract] AND "menin"[Title/Abstract], ids 42578372, 37291166, 28007956, 15691381, 15153434, 12509449, ALL SIX title-checked by efetch. RPA32[tiab] AND menin[tiab] -> 0. · NCBI PubMed E-utilities 2026
   ▸What this evidence cannot show

   Bounded by titles and abstracts (rule 80). The asymmetry is the finding: the same six-record menin list is returned by RPA2[tiab] AND MEN1[tiab], and the narrower phrase '"replication protein A"[tiab] AND menin[tiab]' returns only 3 of the 6 - so the subunit symbol is the MORE sensitive query for this axis and the complex name is the more sensitive query for the KMT2A-locus axis (rule 64 cuts both ways). RPA32[tiab] AND menin[tiab] returns 0, so the RPA32 spelling would have missed this entire literature. Name lists tried are recorded in full above; NOT tried on the RPA side: RPA34, REPA2, RFA2.

   ▸Evidence that argues against this (1)

   • A count of six with three on topic is a small literature and should not be presented as a well-replicated axis. Two of the six are a review and an expression-profiling paper that merely LIST RPA2 among menin partners, which is citation inheritance rather than evidence; one is pure computation. Stripped to independent experimental work, the menin-RPA2 interaction rests on TWO papers, fourteen years apart, from unrelated groups, in HeLa and in breast lines - and no group has retested the binding itself. Kaur G et al. 2023

• part of →The replication protein A heterotrimer (RPA1/RPA2/RPA3) - the ssDNA clamp that every resection and replication-stress claim in this corpus silently depends on, and which had no entry until cycle 160

  32 kDa subunit; contributes DBD-D (OB fold 74-148)

• ← bindsRPA1 (RPA70) - the large subunit of replication protein A, which carries most of the ssDNA-binding folds and the N-terminal fold that hands the lesion to ATR

  heterotrimer partners; menin antibodies co-precipitate both from HeLa

• binds →RPA3 (RPA14) - the 121-residue subunit of replication protein A, which has no annotated domain, no HPA localisation call and no connection to this disease at all

  RPA2-RPA3 subcomplex, to which menin binds poorly

• ← bindsMEN1 / menin — the adaptor the fusion cannot work without

  menin aa 1-40 plus 286-448 to RPA2 aa 43-171; H139D/A160P/A176P abolish it

• ← part ofMenin binds RPA2 directly, the contact maps to RPA2 aa 43-171 across its low-affinity ssDNA site, and three MEN1 patient missense mutations abolish it - a 2003 finding that this graph had never cited despite menin being its central node

  interface entry for RPA2's menin contact

• related to →Replication stress and interstrand-crosslink repair as a cell state - an ATR dependency measured in mouse MLL-rearranged AML, and the same stress implicated in breaking the MLL locus in the first place

  RPA2 N-terminal phosphorylation is the canonical replication-stress readout

• related to →SETD2 (Q9BYW2) - the H3K36 trimethyltransferase that binds all seven MLL fusions, and the only one of them with a published KMT2A::AFF1 dependency test

  both are prey in the Skucha MLL-fusion AP-MS; SETD2 is pan-fusion (all seven baits), RPA2 is a single declined AF4-specific row

• ← related toRPA1 and RPA2 are curated HGNC aliases of POLR1A and POLR1B, and RPA34 - a UniProt synonym of RPA2 - is an alias of POLR1G, so three of the ssDNA clamp's names point at RNA polymerase I subunits; the hazard is in symbol-based DATABASE lookups, not in PubMed, and the measured contamination says which

  symbol-collision record for RPA2 / POLR1B and RPA34 / POLR1G

• ← related toWhere RPA actually is: nucleoplasmic and diffuse until there is single-stranded DNA, then punctate prereplication foci at the G1/S border, damage-induced foci with RAD51, and a PML-body pool - and the subunits differ in how well any of this is measured

  RPA2 carries the most coded localisation evidence of the three subunits, plus the prereplication-foci observation

• related to →PSIP1 / LEDGF p75 — the tether with the best therapeutic index in the whole interactome

  LEDGF reads H3K36me3 and recruits CtIP in the menin-scaffolded cascade whose terminal step is the RPA2/RAD51 pathway