The 43 imported assertions are retained exactly. The review distinguishes catalytic DNA methylation from a sequence-authenticated, catalytically inactive accessory product. Human DNMT3B catalytic-domain assays show CpG preference with lower CpA/CpT activity, so the broad cytosine-5 methyltransferase function remains appropriate alongside the positive CpG annotation. The substrate comparisons are in PMID:32083663, Results and Figure 1. Its Methods start at residue 571 while Results say 575; the deposited 6KDA entity 1 sequence is GHM followed by canonical residues 571–853. The expression fragment is not a full-length natural product, and the SAH-containing crystal is distinct from the SAM-dependent catalytic assay.
The inactive DNMT3B3 in PMID:32968275(https://pmc.ncbi.nlm.nih.gov/articles/PMC7540737/) binds the nucleosome acidic patch and positions the active DNMT3A component toward linker DNA. 6PA7 entity 8 exactly matches the 770-residue Q9UBC3-3 sequence. Human proteins expressed in Sf9 cells and Xenopus histones were used in reconstitution. Acidic-patch anchoring mutants affect chromatin targeting, while activity of the complex on naked DNA is preserved. This supports an accessory targeting role rather than either autonomous DNMT3B3 methylation or biological inactivity. All eight official sequences were reconstructed from alternative-sequence features and matched to their own FASTAs. Names and lengths alone do not identify older constructs as modern products.
Positive transcriptional repression is distinct from the stricter current corepressor molecular-function definition. PMID:17303076(https://pubmed.ncbi.nlm.nih.gov/17303076/) reports ZHX1 interaction and enhanced repression by DNA-tethered DNMT3B. PMID:18567530(https://pubmed.ncbi.nlm.nih.gov/18567530/) reports hPc2-dependent FGFR3 repression independent of methyltransferase activity. Both complete abstracts were read, but full Methods remained inaccessible after bounded recovery attempts. The direction of recruitment through a DNA-binding transcription factor therefore remains unresolved, rather than disproven. The matched ARBA corepressor prediction carries the same uncertainty. No separate mechanistic core or new process assertion is added from these incomplete details.
The positive binding evidence is preserved under the scoped project curation instruction. Targeted Ubc9, EZH2 and G9a experiments support specific enzyme-class binding refinements; oxidative-stress co-immunoprecipitation and spoke-expanded complex records remain generic positive associations. The LANA original PMID:16983096, Results/Figures 3–4 and Methods explicitly assays DNMT3B despite its DNMT3A-focused title. Its preferential DNMT3A promoter recruitment is not transferred to DNMT3B. The UHRF1 original PMID:19798101(https://pubmed.ncbi.nlm.nih.gov/19798101/) is predominantly mouse work but explicitly reports endogenous human DNMT3B–UHRF1 co-immunoprecipitation. Human host cells alone do not establish that every expressed construct is human.
The correction to PMID:16357870 addresses omitted lanes and missing boxes in Figures 2a, 2c, 4a and S1. The substantive correction and corrected supplementary description were read; the authors state that the conclusions are unaffected. The corrected description supplies the DNMT3B/GST–EZH2 fragment context, but corrected figure pixels were not independently assessed. The original publisher supplementary Methods were recovered; main full text remained subscription-limited. This access and presentation history is retained rather than describing the original as an unqualified whole-paper verification.
The nuclear location is refined using supported, nonexclusive nucleoplasmic and chromatin-associated pools. The HPA target record reports supported nucleoplasm by ICC/IF; no independent pixel interpretation is claimed. The DNMT3B3 study establishes chromatin association. Current chromatin ancestry lacks a typed nucleus parent, but that graph fact does not invalidate the biological nuclear chromatin pool. No nuclear-matrix annotation is introduced to manufacture a descendant replacement. The original nuclear studies retain their actual spatial resolution, and neither pool is assigned to every natural product.
PAINT reconstruction finds the human target below the exact ancestral nodes for all four inherited assertions in PTHR23068 version 19. The canonical sequence matches the target row of the 499-sequence MSA. Mouse Dnmt3b and rodent Dnmt3c experiments supply genuine descendant grounding; donor count and target self-evidence are not used as reasons to weaken the ancestral placement. PMID:21378119(https://pmc.ncbi.nlm.nih.gov/articles/PMC3130282/) also directly tests human DNMT3B proteins against SAT2 duplex DNA. The mouse Ment study PMID:22133874(https://pubmed.ncbi.nlm.nih.gov/22133874/) separates tumor-context methylation maintenance from normal thymocyte behavior; its human correlations are not represented as human gene-deletion experiments.
Four saved Reactome paths reach the exact Q9UBC3 1–853 nucleoplasmic participant. The PRC2 model does not assign histone methylation to DNMT3B; the SUMOylation model treats DNMT3B as substrate. These location assertions do not create additional process annotations. InterPro and ARBA rules were traced to their literal domain/FunFam conditions rather than inferred from family names.
Genuine ordinary gene/publication fetching preceded manual review. Falcon authentication and the perplexity-lite fallback quota failed; their failures are retained and no provider report was fabricated. Existing publication and runtime bytes were preserved. Cache availability flags remain literal metadata: PMID:18413740 and PMID:22094255 have nominal full-text flags but lack the operative Methods/Results in their cached bodies. Separately recovered originals are identified in the source guide, not inserted into protected caches. Five concise quotations occur once each in the review YAML; the source-specific new-word totals are 13, 16, 15, 17 and 9, with no inherited quotation inventory or duplicate quotations in these notes.
The existing broad gene-expression-regulation annotation is refined to include GO:0045814, negative regulation of gene expression, epigenetic, alongside the already supported RNA polymerase II repression and positive BAG1 contexts. DNMT3B performs the methyl-transfer step rather than merely being required for an unrelated outcome. PMID:22133874(https://pmc.ncbi.nlm.nih.gov/articles/PMC3248285/) connects mouse Dnmt3b-dependent promoter methylation to Ment repression in MYC-driven lymphoma; inducible deletion supports the relationship, while the human lymphoma arm is correlative. Its Results explicitly leave physical occupancy of individual promoters unresolved. This supports the broad epigenetic mechanism without asserting a purified human promoter assay.
PMID:27153398(https://pmc.ncbi.nlm.nih.gov/articles/PMC4863565/) adds the human D4Z4/DUX4 evidence. The inspected inline Methods, Results and captions describe genotype-defined families and MyoD-converted primary fibroblasts, with GFP/FLAG controls. Repeat length and a permissive chromosome 4 haplotype condition DUX4 expression and disease penetrance; a hypomethylated long-repeat carrier was unaffected. The 2016 source uses FSHD1/FSHD2 terminology, whereas the current UniProt disease record labels the digenic DNMT3B condition FSHD4. This does not imply that every heterozygous variant, cell type or repeat configuration produces DUX4. The ordinary publication cache remains abstract-only. The original PMC main text was independently read; the supplement and figure pixels were not assessed. Failed EuropePMC XML and rate-limited BioC attempts are retained in the evidence record.
No NEW biological-process row is needed: the proposed specific processes and the adopted GO:0045814 descend from the already imported GO:0010468. Current GO:0006306 is obsolete. GO:0044027 specifically concerns chromosomal CpG-island methylation, and GO:0006346 concerns constitutive heterochromatin formation; neither is a generic substitute for every DNA-methylation result. Exact human comparator queries for DNMT1 (P26358), DNMT3A (Q9Y6K1) and DNMT3B (Q9UBC3) confirm annotations on the first two, but comparator presence does not prove DNMT3B's precise mechanism. These QuickGO queries were checked on 2026-10-18. The saved GO-CAM index contains no target match; absence was not treated as evidence of a missing function.
A single NEW molecular-function row records GO:0031491, nucleosome binding, with PMID:32968275 and Q9UBC3-3. The complete 770-residue PDB 6PA7 entity 8 sequence exactly matches the official Q9UBC3-3 product. The original study combines human DNMT proteins expressed in Sf9 with Xenopus core histones. Acidic-patch contacts, AlphaScreen/native-gel binding and anchoring mutants establish nucleosome recognition. Reduced chromatin association in human DKO8 cells is distinct from retained naked-DNA methylation by the DNMT3A-containing complexes; it does not make DNMT3B3 catalytically active. Current typed ancestry shows no ancestor/descendant redundancy with the existing DNA or protein-binding terms. The existing core quotation remains once, and the new row uses reference-only support.
The nuclear localization set remains nonexclusive nucleoplasm plus chromatin, with its documented ontology and product limits. Corepressor uncertainty remains an unresolved recruitment mechanism, not a conclusion that inaccessible full experiments lack it. Products 4 and 5 now explicitly report UniProt's probable nonfunctionality judgment from missing methyltransferase motifs as a prediction rather than direct proof of total inactivity. All eight product identities and sequence notes are unchanged.
All 43 imported annotation source objects and actions are preserved; the total is 44 rows after the single NEW molecular function. There are 33 references and two cores. No new publication quotation is introduced in this follow-up: all five existing excerpts remain exact and occur once each. Prior notes and protected machine-source bytes are preserved.