ERF original sources and database provenance

The database assertions concern human ERF/P50548. The canonical protein has 548 residues. The reported 473-residue product deletes canonical residues 1–75, overlapping the ETS domain; the evidence below does not establish equivalent activity of the two products.

Source	Operative evidence and boundaries
PAINT PTHR11849 tree, alignment, current IBD records	Version 19 human leaf PTN002483866 descends PTN000218930. The 548-residue sequence matches exactly. All four asserted functions occur at that ancestor; no target-path NOT occurs. Human self-WITH evidence is legitimate. The differentiation record includes mouse Erf MGI:109637/P70459, whose 551-residue alignment also matches exactly.
InterPro P50548	IPR000418 maps an ETS domain at residues 26–111 to transcription-factor activity, sequence-specific DNA binding and transcription regulation. IPR046328 maps the ETS-family region at 11–112 to transcription-factor activity and Pol II transcription regulation. The shorter product loses part of these regions.
ARBA00026779	Actual condition set 0 requires IPR000418, IPR036388 and Primates; all match the canonical target. Its sequence-specific DNA-binding output is independently consistent with ERF function.
TFClass author-hosted classification	The recovered October 5, 2018 draft explicitly links ERF/P50548 and reported isoforms under Ets-related family 3.5.2. The current application timed out. The historical classification demonstrates membership, not an independent chromatin assay or equal isoform function.
HPA ERF, structured record	Supported overall nucleoplasm annotation. HPA058532 antigen 465–515 and HPA067952 antigen 168–232 occur in both products. HPA072144 antigen 11–36 is absent from the shorter product and supports nucleoplasm/nucleoli in tested cells; it has approved ICC/IF and an explicit protein-array low-specificity qualification. Its selected A-549 image was actually inspected. Localization does not establish a nucleolar biochemical function.
Reactome ERF export	Input 3209166 and output 3209167 both reach reference entity 54440/P50548, human residues 1–548, in nucleoplasm and cytosol respectively. This is a participant-location assertion, not an export enzyme assignment.
Reactome ERF–ETS2 promoter	Input human ERF entity 2534321 and its component in complex 3209182 explicitly occupy nucleoplasm.
Reactome mixed-species phosphorylation	Human ERF entity 2534321 is the substrate in nucleoplasm; catalyst activity 3245929 uses rat Mapk1 complex 206582. The mixed-species event does contain human ERF. No kinase chemistry is assigned to ERF.
PMID:14729966 original	Methods “DNA constructs,” “Cell culture and transfection” and “Cell staining”; Results and Figures 1–5. Human ERF-cDNA GFP fusions and engineered phosphorylation/deletion mutants in Ref-1 cells support regulated nuclear/cytoplasmic shuttling. Multisite phosphorylation and a nuanced T526 effect are measured; T526 alone is not a sufficient export switch. Engineered deletions are not natural isoform 2. Actual main text/captions were read; figure pixels and supplements were not.
PMID:17502352 original	Targeting/TSC Methods and Results/Figures 5–7. Mouse Erf loss changes placental lineage markers and delays TSC differentiation after growth-factor withdrawal; four mutant lines were compared with two wild-type lines. No human product assay is implied. Actual main text/captions were read; figure pixels and supplements were not.
PMID:23354439 original	Results/Figure 3h and ChIP-Seq Methods use E13.5 mouse fibroblasts. Figure 4b and Promoter assays Methods use human ERF in HeLa reporter assays. Mouse Erf ChIP is not human ERF ChIP; the human RUNX2 dataset is a separate comparison. Actual main text/captions were read; supplements and figure pixels were not.

The three full originals were recovered from official PMC HTML even though Europe PMC full-text XML requests failed. Normal publication-cache text and availability metadata were not rewritten. Two additional normal caches (14729966 and 17502352) were obtained with the genuine fetch tool; 23354439 was already in the accepted acquisition.

Direct human assays and interaction records

Sgouras et al., 1995 original: cDNA-isolation Results, Functional analysis of ERF, Figure 2A–D and Libraries/plasmids, Cell lines/transfection and EMSA Methods. The actual PDF text and Figure 2 pixels were inspected. Human 548-residue ERF represses HeLa CAT reporters; ETS-site deletion and GAL4 tethering distinguish recognition and active repression. Figure 2D is a GAL4-binding control, not an ERF–H28 gel. The original reports its initial Southwestern result as data not shown. No natural shorter-product equivalence is established.
Luck et al., 2020 original: main Results, Generation and characterization of HuRI. Author data have twelve method records for four ERF pairs, using ERF clone 2592/P50548-1. Exact partners are Q08117-2, Q96EZ8-1 and Q9UGI0; MEOX2 clone 824 is labeled P50222 in the author deposit. IntAct records the exact GOA ERF–Q6FHY5 pair under PMID:32296183, with IM-25472-32649, IM-25472-7688 and IM-25472-118431. Current MEOX2 sequences have 303 versus 304 residues and are not declared identical. The observations are yeast two-hybrid interactions, not proof of an endogenous complex or partner-mediated ERF repression. The TLE5 partner-class refinement is explained below. Complete supplementary Methods and target-pair mammalian orthogonal validation were not recovered.
Jolma et al., 2013: metadata and the directly attributed JASPAR MA0760.1 dataset identify human ERF/P50548 and HT-SELEX. The exact position-frequency matrix was retrieved. The full Methods, construct boundaries and supplement were not recovered. The sequence-specific DNA-binding judgment also uses the independent 1995 original and does not claim a full-text audit of the 2013 study.

No new publication quotations were added and the normal source caches were not rewritten. The 473-residue ERF product deletes residues 1–75, including part of the mapped ETS domain; the experiments above do not establish its equivalent function.

Feedback specificity and source boundaries (2026-10-24)

The differentiation refinement uses GO:0060718 chorionic trophoblast cell differentiation, a current child of cell differentiation. It is grounded in the in-vivo immature chorion and Errβ/Gcm1 results of Papadaki et al., Results and Figures 5–7, with targeting, trophoblast-cell culture and in-situ Methods. Four null and two wild-type TSC lines supply complementary evidence about differentiation timing after growth-factor withdrawal. The cultures form giant cells but do not assay terminal syncytiotrophoblast differentiation. The judgment is an ortholog-based conserved contribution, not a human product assay or proof of every direct target. Initial trophectoderm specification is not substituted for this later chorionic process.
GO:0000978 is unrestricted and more specific than GO:0000977 for upstream cis-regulatory DNA recognition. The 1995 original, page 4782 cDNA-isolation/Functional analysis, Figure 2A–B and Libraries/plasmids/EMSA Methods, identifies ETS2 H1 selection and GATA1 upstream-site dependence. The HT-SELEX row remains a bounded motif-recognition assertion. The normal 1995 cache remains abstract-only, while the original PDF and Figure 2 were genuinely inspected during the accepted initial review; cache availability is not a description of reviewer access. Residue mapping is located in the original Figure 2C–D and following Results on page 4784, not inferred from the UniProt disorder feature.
The reviewed TLE5 record, FUNCTION and ALTERNATIVE PRODUCTS sections, identifies a transcriptional corepressor class and the exact AES-2/Q08117-2 partner. VSP_043527 replaces canonical residues 1–9 with a 76-residue N-terminal segment; canonical 10–197 remains, yielding 264 residues. The HuRI pair is clone 2592/P50548-1 with clone 53886/Q08117-2 in screen 8. GO:0001222 records binding to a corepressor-class protein, supported here by the direct pair plus the curated role. It does not demonstrate corepression by this splice product, recruitment through ERF's C terminus, or mediation of ERF repression. Context-dependent antagonism of other repressors is compatible with a context-dependent corepressor classification and is not erased. The remaining partners are MEOX2, MCRS1 and ZRANB1, with their original accessions preserved.
The core location is chromatin, following mouse Erf ChIP and human promoter-dependent repression. The HPA/Reactome nucleoplasmic pool and conditional nucleolar observation remain supported separately; neither antibody qualifications nor inert annotation relationships are used to discard positive localization. No nucleolar biochemical function is inferred. No new osteogenic process assertion is added: the direct targets and stage-specific regulatory mechanism remain an explicit question.

This feedback assessment reread the original text and captions indicated above, the accepted HuRI target records, and the current official term/product records. No additional figure pixels, supplements, or primary TLE5 functional full texts were inspected. The earlier documented 1995 Figure 2 inspection remains valid; normal cache text, source assertions and natural products are unchanged.