GSDMC (Q9BYG8) review notes

Why this gene was selected

GSDMC is a contested-function case with an almost empty GOA. Three mutually incompatible
molecular pictures of the protein have been published, and the curated record carries
essentially none of them: five IBA rows propagated from the gasdermin family node, two
UniProt subcellular-location IEAs, one EXP plasma-membrane row, and one IDA cytoplasm row.
There is no GO annotation at all for pore formation, for vesicle targeting, or for anything
nuclear. So this is a curation gap layered on a live scientific dispute.

The three positions

Position A - intracellular (Rab7+ vesicle) permeabilization, not plasma-membrane pyroptosis

Pandey et al., Immunity 2025 (PMID:40701157, full text available) is the strongest
single body of work on GSDMC in its main tissue, the intestinal epithelium. It opens by
setting itself against the family paradigm:
PMID:40701157

Their central claims:

• Cathepsin S is the activating protease in intestinal epithelial cells, and lysis is not
  the point: PMID:40701157
  This is the mechanistically important observation: GSDMC's beta1-beta2 lipid-binding
  motif is one residue short relative to the other gasdermins, and completing it converts
  GSDMC into a better killer. On this reading GSDMC is a deliberately poor pore-former.
• The target membrane is intracellular, not the plasma membrane:
  PMID:40701157 and, at the level
  of the primary imaging, PMID:40701157.
• The downstream consequence is lipid-droplet/lipid-turnover control feeding type 2
  immunity: PMID:40701157
• They separate the two models pharmacologically: PMID:40701157
• And they state the generalization explicitly: PMID:40701157

Note the important asymmetry: most of the mechanistic imaging is on murine GSDMC2/GSDMC4
(mouse has four Gsdmc paralogs) in HeLa cells and C. elegans, though human GSDMC is
also cleaved by CTSS and the human N-terminal fragment was the one imaged in HeLa.

Position B - nuclear chromatin scaffold

Ren et al., Cell Rep 2026 (PMID:42176271, abstract only in our cache). The title is about
metabolic reprogramming, so the molecular claim is easy to miss; it is in the abstract:
PMID:42176271 and
PMID:42176271

This asserts a molecular function with no membrane component at all. It is a single
laboratory, one paper, not independently replicated, and we have only the abstract. It is
worth noting that the original cloning paper already flagged nuclear-targeting sequence
features: PMID:11223543 - although that same paper named
the protein extranuclear factor (MLZE = melanoma-derived leucine zipper-containing
extranuclear factor), so the older literature points the other way on localization.
I did not find independent corroboration of nuclear GSDMC acting as a chromatin scaffold.

Position C - classical pyroptotic executioner

The founding result is Hou et al., Nat Cell Biol 2020 (PMID:32929201, full text):
PMID:32929201
This is not merely a cell-biology inference; there is direct reconstitution:
PMID:32929201 and
PMID:32929201.
Independently, Zhang et al., Cell Res 2021 (PMID:34012073) reported the metabolite
alpha-ketoglutarate driving DR6/caspase-8-dependent PMID:34012073.

Wu et al., Signal Transduct Target Ther 2025 (PMID:41407678, full text) adds a third
protease: PMID:41407678 and
PMID:41407678.

The DdBIC paper is weaker support for the physiological model than it looks. It shows
GSDMC can be pushed into forming a lytic pore by a synthetic Nur77 ligand acting through
a long mito-ROS / OMA1 / OPA1 / PERK / ISR cascade. That establishes pore competence, not
that pore formation is what GSDMC normally does - which is exactly what Position A denies.
It also further destabilizes the canonical model in a second way, by making the activating
protease granzyme B rather than caspase-8. In other words, the three pyroptosis papers
agree that GSDMC can kill but disagree on what cleaves it (caspase-8 vs caspase-6 vs
granzyme B) and under what trigger.

The field has named the problem

PMID:41092892, a
2025 Immunity commentary on the Pandey paper, which nonetheless still frames GSDMC in
membrane terms: PMID:41092892

Adjudication taken in this review

1. Pore-forming (wide pore channel) activity is retained as the core molecular
   function. It is the only one of the three positions with direct, reconstituted,
   protein-level evidence (liposome binding, EM-visible pores, dye leakage; PMID:32929201),
   and it is reproduced across independent labs and independent activating proteases.
   Position A does not contradict pore competence - it accepts it and shows that
   completing the lipid-binding motif enhances it.
2. What is contested, and recorded as contested, is the target membrane and the
   physiological consequence. In the tissue where GSDMC is actually expressed at
   steady state (intestinal epithelium), the evidence favours permeabilization of Rab7+
   late endosomal vesicles with lipid-droplet consequences over plasma-membrane lysis.
   Both plasma-membrane annotations are therefore kept but demoted to non-core, and a
   NEW late endosome location is proposed from PMID:40701157.
3. Pyroptosis is kept as non-core, not accepted as core. It is real in caspase-8-driven
   cancer-cell settings and under the DdBIC/granzyme-B trigger, but the primary-tissue data
   and the incomplete lipid-binding motif argue it is not the default behaviour of this
   paralog. This is a deliberate demotion, not a rejection.
4. Position B is recorded but not annotated. One paper, abstract only, no independent
   replication, and a molecular claim (chromatin scaffold) that would be a major departure
   for a gasdermin. It goes into suggested_questions and suggested_experiments rather
   than into existing_annotations, because asserting chromatin binding or a
   nucleus location on this basis would put a machine-readable claim into the record
   that the evidence does not yet carry.

The family-level IBAs

All five IBA rows trace to PANTHER node PTN000419132. I resolved the donor identifiers
(they are not inspectable as a PAINT tree from inside this repository, so I did not
attempt structured propagation_review metadata):

Two observations that matter for the review:

• The three phospholipid-binding IBAs (GO:0005546, GO:0070273, GO:0001786) are seeded
  only by GSDMD (mouse and/or human). They assert exactly the inner-leaflet
  phosphoinositide/phosphatidylserine recognition paradigm that Position A argues GSDMC
  does not follow, given its one-residue-short beta1-beta2 lipid-binding motif. The only
  GSDMC-specific lipid data is a bulk liposome assay whose mix contained PC, PE, PS, PI and
  PI(4,5)P2 together (PMID:32929201 methods), which cannot resolve headgroup specificity.
  These are kept as non-core, with the doubt written into reason, rather than removed:
  per project guidance an IBA is a considered phylogenetic judgement and overturning one
  requires arguing with the node placement, which I could not inspect.
• GSDMC itself (UniProtKB:Q9BYG8) appears in the WITH/FROM of the GO:0070269 IBA. Per
  project guidance this is not circular: it marks that the target carries its own
  experimental grounding for the term, and the IBA then adds the claim that the function is
  inherited rather than lineage-specific.

GO:0042742 defense response to bacterium is the weakest of the five. Its donors are
GSDMD, GSDMA and GSDMB - each of which has direct antibacterial evidence - while GSDMC's
own documented in-vivo immune role is anti-helminth type 2 immunity, not antibacterial.
Kept as non-core and raised as a question for PAINT.

Localization annotations

• GO:0005737 cytoplasm IDA from PMID:11223543. Our cache of that paper is abstract-only
  and the abstract does not state the localization; the curator read the full text. Per
  project rules this is accepted, not challenged. It is also biologically uncontroversial:
  full-length GSDMC is an autoinhibited cytosolic zymogen.
• GO:0005886 plasma membrane appears twice (IEA from UniProt SubCell SL-0039, and EXP
  from PMID:32929201). Both must carry the same action. The EXP row is genuine for the
  caspase-8/cancer-cell context but is precisely what PMID:40701157 reports not to happen
  for the physiologically cleaved intestinal form, so both are kept as non-core.