CBFB is the non-DNA-contacting beta subunit of core-binding factor. Its principal molecular role is to bind RUNX proteins, stabilize their Runt-domain conformation and support promoter-dependent transcription. RUNX supplies sequence recognition and DNA contact. The same general cofactor can participate in activating and repressive complexes; the direction depends on the regulatory site and cellular context. The review retains one coactivator core with nucleoplasmic localization and membership in core-binding factor, without inventing a separate DNA-binding or enzyme activity for CBFB.
The reviewed canonical protein is human UniProt Q13951 (PEBB_HUMAN), 182 amino acids, HGNC:1539. Both normal alternative products, Q13951-1 and Q13951-2, remain exactly as seeded, including the second product's VSP_036044 sequence note. No activity is asserted to be unique to either product. All 136 machine-sourced assertions retain their original terms, labels, evidence, references, supporting entities, ordering and source flags. The archived GOA and UniProt records and all publication and Reactome cache versions remain immutable.
The frozen ClinGen cohort records autosomal dominant cleidocranial dysplasia 2 as Definitive. This review does not change that association, date or classification. The clinical literature is used to describe the protein and distinguish variant mechanisms, rather than to recalculate the frozen gene–disease assessment.
PMID:10856244 establishes a human RUNX1 Runt-domain–CBFB fragment interface. In addition to the complete cached abstract, selected complete original text was read at the official PMC article. The proteins were recombinant human fragments expressed in bacteria, so the expression host is not the species of the protein constructs. The 2.6-angstrom structure supports heterodimer formation and allosteric stabilization; it does not make CBFB the DNA-contacting subunit or establish every full-length cellular assembly. No crystallographic image, raw density or supplement was reanalysed.
PMID:12217689 was read as complete metadata and abstract, with official primary identity confirmation. The Runt-domain conformational comparison and biochemical context support the influence of CBFB on RUNX DNA binding. Full primary body access remained unavailable. Ionic effects described for the Runt-domain system are not transferred to CBFB as ion binding or catalysis.
PMID:17431401 contains a target-relevant experiment despite its CDK6-focused title. Selected actual Results and Methods show Flag-Runx1–CBFβ co-immunoprecipitation in 293T cells. The Methods specify mouse Runx1 cDNA; the experiment is not represented as an exclusively human recombinant pair. CDK6 affects RUNX1 DNA binding while leaving the RUNX1–CBFB association intact. This supports the conserved transcription-factor association, not a CBFB kinase function.
PMID:20483917 combines Runt-domain/CBFB biochemistry, response-element perturbation and reporter work. Human-locus transgenic-mouse T-cell and mast-cell experiments were distinguished from human CEM/Jurkat reporter assays. The study supports transcription by the RUNX–CBFB complex, while the DNA-recognition surface belongs to RUNX. Selected complete Methods and Results were read from the normal cache; no independent image or raw-data analysis is claimed.
PMID:20591170 supplies especially useful endogenous human controls. In MDA-MB-231 cells, RUNX2 immunoprecipitation recovers CBFB, with IgG and input controls. Nuclear-extract controls support nuclear CBFB. Two CBFB hairpins reduce CBFB without lowering RUNX2, and selected target transcripts respond. Galectin-3 remains CBFB-independent despite RUNX2 promoter occupancy. Thus neither chromatin association nor CBFB binding means that every RUNX target is activated. Cancer-cell invasion is a contextual phenotype, not an additional intrinsic molecular activity. Complete metadata/abstract and selected unique Methods, Results and Discussion passages were read.
PMID:25678665 was assessed using selected actual normal-protein and fusion controls. In human ME-1 cells, the inhibitor preferentially disrupts RUNX1 association with oligomeric CBFB-SMMHC, while affecting the wild-type CBFB association only modestly. The paper therefore does assay normal CBFB; the fusion-focused title is not grounds for rejecting the interaction. The fusion's leukemia and drug-response phenotypes are not assigned to the normal protein.
PMID:39696918 independently supports RUNX3 association. Human HEK293T/U2OS experiments include Runt-region and interface-mutant tests, co-immunoprecipitation, controlled proximity assays, CBFB depletion/rescue and RUNX3 stability measurements. Proximity ligation alone is not a resolved physical interface. RUNX3 ADP-ribosylation and repair outcomes do not make CBFB an ADP-ribosyltransferase. Selected complete Results and Methods were read; the review does not add a new repair-process assertion from necessity alone.
Five generic binding assertions with source-specific RUNX structure or targeted association evidence are refined to GO:0140297, DNA-binding transcription factor binding. Each refinement carries a short exact source anchor. Generic screen associations remain qualified non-core where the preserved partner is independently corroborated by the archived UniProt interaction record and, for RUNX partners, separate targeted evidence. The original screen's bait orientation, fragment boundaries and quantitative controls are not invented from its overall validation rate. No screen establishes an isoform-exclusive interaction merely because the source partner accession names an isoform.
This treatment follows the standing ClinGen project instruction, which takes precedence over general skill guidance to remove uninformative generic binding. Supported associations can remain non-core, a demonstrated finer molecular function can be used, and genuinely unresolved evidence remains undecided. There is no required removal quota. The MAP2 and INTS13 screen associations do not establish microtubule function, Integrator catalysis or constitutive membership in another complex.
The electronic, unqualified GO:0003677 DNA-binding assertion is marked over-annotated because the established CBFB mechanism is a non-DNA-contacting cofactor interaction. By contrast, the three sequence-specific DNA-binding assertions explicitly carry contributes_to, which accurately records CBFB's contribution to the RUNX complex. That recognized qualifier is retained and considered. Other relationship-column wording is not used to manufacture distinctions in biological activity.
PMID:2845103 remains an explicit uncertainty. Its cached abstract describes bacterial adenylate kinase, but that alone does not prove how an experimental interaction record acquired the citation. The exact official IntAct publication record was inaccessible. Independent PDB 1E50 identifies a human CBFB–RUNX1 complex with PMID:10856244, corroborating the pair without establishing a replacement for the preserved source PMID. The annotation remains UNDECIDED; no wrong-gene, MISCITED, WRONG_IDENTIFIER or identifier-replacement claim is made. The separate bounded provenance consultation reached the same access limit.
The mouse donor underlying the developmental transfers was traced to PMID:18258917. The normal cache remains abstract-only. Publisher/official body retrieval was unavailable, but selected original author-uploaded text and captions were read from the matching paper, with authors and DOI 10.1126/science.1151844 matched. Conditional Cbfb deletion, Cbfb re-expression, Th-POK expression and anti-Cbfβ chromatin-association passages support a mouse T-cell lineage mechanism. Re-expression suppresses Th-POK in the tested CD8-intermediate population, rather than uniformly in every T-cell state. These transfers remain non-core developmental outputs, not direct human differentiation experiments or autonomous CBFB DNA-binding repression. No complete-paper, supplement or image-reading claim is made.
PMID:22190036 explicitly reports endogenous human CBFB dependence of Vif–CUL5 assembly, APOBEC3G polyubiquitination and degradation while Vif–APOBEC association remains detectable. PMID:24402281 provides structural corroboration of the virus-recruited assembly, and PMID:22190034 provides the original interaction-landscape context. Their complete available abstracts and identities were read, with the full-body limits retained. CBFB participates through assembly and stabilization, rather than serving as the ligase catalytic subunit. The viral co-option is retained non-core and is not generalized to every CUL5 substrate or complex.
All 91 cited Reactome event identities and complete chosen summaries were read separately. Each corresponding source assertion is nucleoplasm, not the event's enzyme activity or every process in its title. Their compartment evidence is the preserved curated GOA projection, corroborated independently by nuclear extracts and RUNX-associated transcriptional biology. The summaries themselves are not described as 91 independent localization assays. Each reference assessment records its actual event context and the limitation that the full component graph and underlying primary experiments were not reanalysed.
The exact non-HSA identifier Reactome:R-NUL-8939977 is preserved. Its summary combines human AKT1/CBFB with mouse Runx2-5 and is not silently converted to a wholly human event. Neither that mixed-species event nor other pathway summaries transfer AKT, KMT2, PRMT, HIPK, HDAC or ubiquitin-ligase catalysis to CBFB. Likewise, proximity to an ELF1-containing complex does not establish a direct CBFB–ELF1 interface.
The PMID:19946888 membrane-proteomics assertion remains UNDECIDED. The complete accessible abstract describes YTS membrane fractions and includes proteins that can associate transiently with membranes. The exact CBFB peptide, fraction and target controls were not recovered. A mainly nuclear role neither proves the membrane observation nor refutes a peripheral membrane pool. The precise access gap is retained instead of asserting a wrong-gene error from the abstract.
Two additional references were obtained through normal source acquisition and read from the resulting immutable XML caches. PMID:36241386 reports eight affected individuals from five families, with RNA examined in two subjects. The discussion's proposed truncated or neomorphic products are distinguished from the demonstrated RNA observations; one molecular mechanism is not assigned to every allele. PMID:39894570 reports a mother and daughter with a deletion-inversion-deletion involving exons 1–2, breakpoint confirmation and reduced expression supporting dosage loss for that allele. Their phenotype does not reproduce every feature of the earlier cohort. The latter paper's introductory 187-amino-acid statement is not substituted for the reviewed 182-amino-acid canonical UniProt protein. Selected complete Methods, Results and Discussion were read; raw sequencing, images and all supplementary tables were not independently reanalysed.
All 119 references have an individual manual assessment. A full_text_available cache flag records availability, not a claim of complete body or supplementary-table reading. In particular, the chosen PMID:33961781 cache is partial HTML, and PMID:35914814 remains the preserved abstract-only version despite a fuller recovered artifact. Separate external primary access for PMID:10856244 and PMID:18258917 does not change their normal cache flags. References marked VERIFIED have supporting identity/source facts; the exact citation provenance of PMID:2845103 and target-specific membrane evidence from PMID:19946888 remain UNVERIFIED.
The normal isolated deep-research attempt failed with a connection error and produced no provider report; the fallback also produced none. No provider-branded research text was fabricated. These notes document the actual source reading and reasoning instead. Source-access and provenance consultations are bounded evidence contributions, not substitutes for whole-review peer assessment.
The final scientific proposal has 109 ACCEPT, 19 KEEP_AS_NON_CORE, five MODIFY, two UNDECIDED and one MARK_AS_OVER_ANNOTATED decisions, with no NEW annotations. Both products and all machine-source objects are preserved. Canonical application, standard CREATE history, final rendering and publication require normal candidate checks, distinct whole-science review and ROOT acceptance. Those workflow checks do not convert the two unresolved assertions into certain biological claims.
The required review issue concerns the legacy TAS assertion of GO:0006366 transcription by RNA polymerase II. This process describes Pol II RNA synthesis. The independently supported activity of CBFB is RUNX-cofactor regulation, including human enhancer and target-gene experiments in PMID:20483917 and PMID:20591170. That positive regulatory evidence remains accepted. It does not, on its own, resolve the precise scope of the historical broad-process assertion.
The original PMID:8351518 PubMed record and complete cached abstract were rechecked. They describe the CBFB–MYH11 fusion and identify normal CBFB as a transcription-factor subunit regulating T-cell genes. The official publisher body was not accessible, and the full original article was not read. Its exact support for the legacy process assignment remains unverified. This assertion is now UNDECIDED, with the original term, TAS evidence and reference unchanged. No wrong-gene, MISCITED or invalid-reference claim is made from an abstract.
Lack of intrinsic RNA polymerase activity or direct DNA contact is not a universal exclusion rule for participation in a biological process. Cofactors and structural components can do part of a process without catalyzing its central chemistry. Conversely, regulatory effects do not automatically establish every regulated process. The uncertainty here concerns that boundary and the inaccessible original evidence. A previous AI review of another cofactor is not primary evidence for the CBFB decision. No redundant MODIFY to the already accepted GO:0006357 or new process annotation is proposed.
The description now separates the RUNX Runt-domain conformational mechanism from the RUNX3 protein-stability result in PMID:39696918. The latter is supported by the previously inspected human-cell perturbation and stability experiments, and is not generalized to every RUNX product or cell state. No NEW stabilization annotation is added. The coactivator core remains specific to supported positive transcriptional activity; promoter-dependent repression remains explicitly acknowledged. A broader coregulator label is possible, but is not required to erase the demonstrated coactivator activity or to imply identical outputs at every locus.
The clinical biological question now names PMID:36241386 and PMID:39894570 where their allele evidence is used. Two source-access tasks have been moved from biological questions into this journal:
- Which original IntAct record links the preserved PMID:2845103 annotation to the human CBFB–RUNX1 pair, and can its citation provenance be resolved without replacing the machine source speculatively?
- Which CBFB peptides, fractions and controls support the membrane-proteomics observation in PMID:19946888, and does it represent a transient peripheral association?
All 136 machine assertions, both alternative products, 119 reference identities, original source caches and prior histories remain exact. The other 135 annotation reviews and all core fields are unchanged. Individual Reactome context remains in each row, including the mixed-species R-NUL event, rather than being bulk-rewritten for prose brevity. Existing evidence anchors remain unchanged; one short cache-verbatim anchor identifies the limited original abstract context. Current totals are 108 ACCEPT, 19 KEEP_AS_NON_CORE, five MODIFY, three UNDECIDED and one MARK_AS_OVER_ANNOTATED. No NEW annotation or source replacement is added.