TMF1 (TATA element modulatory factor 1) — research notes

Context

This review was triggered by geneontology/go-annotation#6381 ("TMF1 annotation review"). Curators (V. Wood, H. Attrill, S. Munro group, L. Breuza, R. Lovering, P. Gaudet) flagged that the human TMF1 GO annotations are still dominated by transcription-related terms (transcription coactivator activity, androgen receptor signaling, positive regulation of transcription by RNA Pol II) that trace back to a small set of late-1990s/early-2000s overexpression studies, while the well-supported biology is now Golgi vesicle tethering (golgin) in retrograde and intra-Golgi vesicle traffic.

FlyBase already updated the D. melanogaster TMF1 ortholog to "vesicle membrane tethering activity" via P2GO (per @hattrill in #6381).

Summary of evidence

TMF1 is a golgin (vesicle tethering factor) on the trans-Golgi rim

PMID:17698061. Knockdown of TMF or Rab6 blocks retrograde transport of Shiga toxin from early/recycling endosomes to the TGN and causes Rab6-dependent displacement of GalNAc-T2 (but not GalT) from the Golgi. The authors conclude "critical roles for TMF in two Rab6-dependent retrograde transport processes: one from endosomes to the Golgi and the other from the Golgi to the ER."
PMID:23239882. Establishes TMF as the long coiled-coil tether that, together with the COG complex, mediates retrograde tethering of intra-Golgi vesicles.
PMID:28122620. Mitochondrial-relocation assay shows that TMF1 acts as a vesicle-capture (tethering) golgin. The first 36 residues of TMF are necessary and sufficient for intra-Golgi vesicle capture; the C-terminal coiled-coil with the Rab6-binding region also contributes capture activity.
PMID:36103876. Drosophila in vivo characterisation reaffirms TMF as an intra-Golgi vesicle-capturing golgin with conserved tissue-level functions.

Mechanism / structure

TMF is recruited to the Golgi membrane via a C-terminal Rab6-binding coiled-coil (UniProt SUBCELLULAR LOCATION; PMID:17698061; PMID:23239882; PMID:28122620). It contains predicted disordered N-terminus (~residues 1–400) followed by extensive coiled-coil (residues 439–922, 984–1092 per UniProt feature table).
The N-terminal region has a conserved M-S-W-L/F motif shared with the other intra-Golgi tethering golgins GMAP-210 and golgin-84 PMID:28122620.

Mouse phenotype consistent with Golgi/secretion role, not transcription

PMID:23000399. The endocrine phenotype is consistent with disrupted Leydig cell secretion / spermiogenesis acrosome biogenesis (a Golgi-derived organelle), not with loss of a transcription coactivator.
PMID:24639530. The colon mucus phenotype is consistent with disrupted Golgi-mediated mucin glycosylation/secretion.

PMID:1409643. The original cloning paper used HIV-1 LTR reporter and in vitro TBP assays; the protein was characterised in a viral context, not in regulation of endogenous cellular genes.
PMID:10428808. Hedged language ("might function"); transient overexpression in PC-3 cells; a Gal4-DNA binding domain fusion to TMF1 reportedly does not work as a dbTF (data not shown), inconsistent with a bona fide transcription coactivator activity (GO:0003713 sensu stricto).
PMID:12044884. Reports interaction with hbrm/hSNF2α and BRG-1/hSNF2β SWI/SNF subunits. The bulk of TMF1 is at the Golgi; nuclear pool is a minor fraction.
PMID:15467733. Reports a BC-box-mediated role in Stat3 ubiquitination; this is the basis of the regulation-of-proteasomal-protein-catabolic-process annotations.
The 2025 Pharos status for TMF1 is "Tbio" (proteins with virtually no functional/drug characterisation), reflecting the field's view that transcription coactivator function is not solidly established.

Implications for GO annotation

Existing annotation (Aspect)	Recommendation	Rationale
GO:0005794 Golgi apparatus, GO:0000139 Golgi membrane (CC)	ACCEPT — core	Multiple IDA/IBA studies; UniProt SUBCELLULAR LOCATION; canonical golgin localization.
GO:0005829 cytosol (CC, TAS Reactome)	KEEP_AS_NON_CORE	Reflects diffuse cytoplasmic pool described in PMID:15467733; TMF is concentrated at Golgi.
GO:0005634 nucleus (CC, IEA)	MARK_AS_OVER_ANNOTATED	A minor isoform-dependent nuclear pool was reported (PMID:12044884) but the protein's primary localization is Golgi; the SubCell-derived IEA over-emphasizes this fraction.
GO:0005737 cytoplasm (CC, IEA)	KEEP_AS_NON_CORE	Generic; subsumed by Golgi/cytosol.
GO:0005783 endoplasmic reticulum (IBA, IEA-Ensembl)	MARK_AS_OVER_ANNOTATED	TMF mediates retrograde traffic to the ER but is not a steady-state ER-resident protein.
GO:0006891 intra-Golgi vesicle-mediated transport (NEW)	NEW	Strongly supported by PMID:28122620, PMID:23239882, PMID:36103876.
GO:0099041 vesicle tethering to Golgi (NEW)	NEW	Best existing BP for the golgin tethering activity (PMID:23239882, PMID:28122620).
GO:0006895 retrograde transport, Golgi to endoplasmic reticulum (NEW)	NEW	Established by PMID:17698061 (Rab6-dependent retrograde traffic).
GO:0042147 retrograde transport, endosome to Golgi (NEW)	NEW	Established by PMID:17698061 (Shiga toxin retrograde trafficking).
GO:0003713 transcription coactivator activity (IDA, PMID:10428808)	MARK_AS_OVER_ANNOTATED	Single overexpression study with hedged conclusions; Gal4-DBD fusion fails to transactivate; primary biology is at the Golgi. Demote from core; do not delete because a low-level moonlighting role cannot be excluded.
GO:0030521 androgen receptor signaling pathway (IDA, PMID:10428808)	MARK_AS_OVER_ANNOTATED	Same single overexpression study; mouse KO testosterone phenotype is more compatible with disrupted Leydig/spermiogenesis secretion than with cell-autonomous loss of an AR coregulator.
GO:0045944 positive regulation of transcription by RNA Pol II (IDA, PMID:10428808)	MARK_AS_OVER_ANNOTATED	Inferred from the same single transient transfection assay; not validated by orthogonal assays.
GO:0050681 nuclear androgen receptor binding (IPI, PMID:10428808)	KEEP_AS_NON_CORE	The far-Western/co-IP interaction is real, but is a moonlighting feature, not a core function.
GO:0061136 regulation of proteasomal protein catabolic process (ISS/IEA)	KEEP_AS_NON_CORE	PMID:15467733 supports a BC-box/Elongin-C-mediated role in Stat3 ubiquitination; this is a discrete moonlighting activity at the Golgi, not the core function.
GO:0005515 protein binding (IPI x2)	REMOVE / MARK_AS_OVER_ANNOTATED	Uninformative per project guideline.

Suggested new term

A molecular-function counterpart of GO:0099022 ("vesicle tethering" — currently a BP) would let us annotate golgins like TMF1, GMAP-210, golgin-84 and the GRIP-domain golgins to a meaningful MF beyond "protein binding". ValWood (#6381) referred to a prospective "vesicle membrane tethering activity" term (not yet present in GO at time of writing — GO:7770062 returned 404 from the GO API on 2026-05-02). Until such a term is added, GO:0099041 (vesicle tethering to Golgi) is the most informative existing BP, and we record the missing MF in proposed_new_terms.

Update 2026-09-26: GO:7770062 minted

GO:7770062 vesicle membrane tethering activity now exists in GO (it is used as the MF of GARP, exocyst and Dsl1/NZR annotons in cached production GO-CAMs, e.g. gocams/index.tsv), and GO:0099041 vesicle tethering to Golgi was obsoleted in the 2026-07-26 release. The review now uses GO:7770062 as TMF1's core MF and the proposed_new_terms entry has been retired (PR #3237). The 404 noted above is historical.