# VPS28 review notes

## Scope

VPS28 is reviewed in the PN autophagophore sealing branch as a core ESCRT-I subunit. PN entries without PMIDs were used as context only. The main curation issue is to separate direct VPS28 function in ESCRT-I assembly and membrane remodeling from complex-level ubiquitin recognition that is mediated mainly by TSG101.

## Evidence synthesis

VPS28 is a constitutive ESCRT-I component. UniProt describes VPS28 as a "Component of the ESCRT-I complex" and says the complex "consists of TSG101, VPS28, a VPS37 protein" with MVB12A/MVB12B or UBAP1 variants [file:human/VPS28/VPS28-uniprot.txt, "Component of the ESCRT-I complex"; file:human/VPS28/VPS28-uniprot.txt, "which consists of TSG101, VPS28, a VPS37"]. Human ESCRT-I papers similarly state that "All ESCRT-I complexes contain three common subunits (TSG101, VPS28, and VPS37)" and that human ESCRT-I is a heterotetramer of TSG101, Vps28, Vps37, and MVB12 [PMID:18005716, "All ESCRT-I complexes contain three common subunits"; PMID:20654576, "comprises a heterotetramer of TSG101 (Vps23), Vps28, Vps37 and MVB12 protein"]. These sources support accepting ESCRT-I complex membership and core endosomal ESCRT-I sorting annotations.

VPS28 participates in endosomal sorting and MVB-related trafficking as part of ESCRT-I. The original mammalian VPS28 paper states that VPS28 is a "cytosolic protein that interacts with TSG101/mammalian VPS23" and that TSG101/hVPS28 translocate to "the surface of these vacuoles" when dominant-negative VPS4 induces endosomal vacuolization [PMID:11134028, "cytosolic protein that interacts with TSG101/mammalian VPS23"; PMID:11134028, "the surface of these vacuoles"]. The receptor-sorting study reports that the "TSG101 and hVPS28" complex localizes to ubiquitin-rich EGF-positive endosomes and that "antibodies to VPS28 specifically impair EGF degradation" [PMID:11916981, "TSG101 and hVPS28 are localized to ubiquitin-rich endosomes"; PMID:11916981, "antibodies to VPS28 specifically impair EGF degradation"]. These lines support the endosome/endosome membrane, MVB sorting, and ubiquitin-dependent endocytic cargo sorting process rows.

The `ubiquitin binding` rows should not be accepted as direct VPS28 molecular function. The same paper reports that "Both TSG101 and hVPS28 bound specifically to ubiquitin-agarose" in cytosol, but then shows that "In vitro-translated hVPS28 did not bind to ubiquitin-agarose" and concludes that complex binding occurs through the TSG101 UEV domain [PMID:11916981, "Both TSG101 and hVPS28 bound specifically to ubiquitin-agarose"; PMID:11916981, "In vitro-translated hVPS28 did not bind to ubiquitin-agarose"; PMID:11916981, "complex binding to ubiquitin via the Ubc domain of TSG101"]. This supports marking VPS28 `ubiquitin binding` as over-annotated rather than core MF.

The PN-specific autophagy connection is now stronger than the older generic macroautophagy row. A phagophore-closure study reports that loss of VPS37A abrogates recruitment of "the ESCRT-I subunit VPS28 and CHMP2A" to phagophores [PMID:31519728, "the ESCRT-I subunit VPS28 and CHMP2A"]. A later ESCRT-I structural paper reports that mutation of VPS28 helical-interface residues blocks ESCRT-I filament formation and "autophagosome closure in cells", and explicitly states that the "ESCRT-I helical interface is required for autophagosome closure" [PMID:32424346, "blocks filament formation in vitro and autophagosome closure"; PMID:32424346, "ESCRT-I helical interface is required for autophagosome closure"]. Therefore the autophagy rows are better represented as `GO:0000045 autophagosome assembly` than broad `macroautophagy`.

Generic `protein binding` rows should be marked over-annotated. Many are from interaction screens or ESCRT subunit interactions, but the informative biology is ESCRT-I complex membership and ESCRT-I-mediated endosomal sorting, not a generic binding MF. Viral budding, extracellular exosome, plasma membrane, cytosol, and cytoplasm rows are true contexts or detection/localization observations but should not be treated as VPS28 core function in this PN review.

## Falcon

Falcon deep research was started for VPS28 on 2026-06-02 but timed out after 600 seconds and did not produce a usable report. The review therefore relies on UniProt, cached primary literature, and PN context rather than Falcon output.

## Description cleanup note

The YAML `description` field was revised to keep it as a standalone biological summary. Project-specific curation framing moved here instead.

- Moved out of the YAML description: the prior wording described phagophore/autophagosome closure as PN-relevant.