LRRC8A (SWELL1) — review journal

UniProt: Q8IWT6. HGNC: LRRC8A. Synonyms: SWELL1, LRRC8, KIAA1437.

1. What the protein is

LRRC8A is a four-transmembrane protein with a large cytosolic leucine-rich-repeat (LRR)
domain. It is the obligatory subunit of the volume-regulated anion channel (VRAC, also
called VSOAC / I(Cl,swell)), a hexameric channel of the plasma membrane that opens on cell
swelling and on lowering of cytoplasmic ionic strength.

Founding genetic evidence (two independent screens, 2014):

• PMID:24725410
• PMID:24725410
• PMID:24725410
• PMID:24790029
• PMID:24790029

The last quote is the key fact for the subunit question: in cells LRRC8A alone does
not reconstitute VRAC current — it must heteromerize with LRRC8B–E, and the isoform
combination sets inactivation kinetics and substrate preference
PMID:24790029.

Reconstitution and structure nuance the picture: purified LRRC8 complexes form
osmolality-gated anion channels in bilayers
PMID:26824658,
and a homo-hexameric LRRC8A channel conducts
PMID:29769723.
Cryo-EM confirms the homo-hexamer and, importantly, that a residue at the narrowest
constriction of the homomeric channel is a pore determinant of the heteromeric channel
PMID:30095067.

So LRRC8A is pore-forming, not merely a chaperone or accessory subunit — but the
physiological channel is a heteromer. Curation consequence: in core_functions I recorded
the channel activity with contributes_to_molecular_function plus in_complex
(GO:0034702), rather than asserting molecular_function outright, and stated the
homohexamer caveat in the description.

2. Substrates: it is a broadly permeable large-pore channel

VRAC is not a chloride-specific channel. Beyond Cl-/I-, it passes organic osmolytes and
signalling molecules:

• taurine and regulatory volume decrease PMID:24790029
• PMID:28193731
• substrate spectrum is subunit-dependent PMID:28193731

UniProt records Rhea reactions for chloride, iodide, taurine, L-aspartate, L-glutamate,
myo-inositol and 2',3'-cGAMP.

There is also an intracellular pool: LRRC8 proteins on lysosomal membranes carry Lyso-VRAC
currents PMID:33139539.

3. The contested question: is cGAMP transport real, or an overexpression artifact?

Evidence FOR (and it is substantial, and largely not overexpression-based)

• Lahey et al. 2020, the GOA IDA source:
  PMID:33171122,
  PMID:33171122,
  PMID:33171122,
  and a subunit-specific negative control
  PMID:33171122.
  A subunit-composition dependence with a suppressive paralog is hard to explain as a
  nonspecific overexpression leak.
• Zhou et al. 2020 (Immunity), independent group, with in vivo loss-of-function:
  LRRC8A/LRRC8E-containing VRACs transport cGAMP and cyclic dinucleotides across the plasma
  membrane, and Lrrc8e-deficient mice show impaired IFN responses and compromised immunity
  to HSV-1 (PMID:32277911; abstract-only in cache, so no verbatim quote is asserted in the
  YAML).
• Concepcion et al. 2022 (Nat Immunol), third group, T-cell-intrinsic in vivo phenotype
  attributable to LRRC8C-dependent cGAMP uptake and downstream STING signalling
  (PMID:35105987).
• Zheng et al. 2025 (Mol Cell), the PSA/NPEPPS paper, treats VRAC-mediated cGAMP import as
  an established, tunable function and uses endogenous-level manipulations (PSA knockout /
  re-expression), not VRAC overexpression:
  PMID:41371222,
  PMID:41371222.
  They are explicit about their own limits:
  PMID:41371222

Evidence AGAINST — read carefully, because it does not say what the title suggests

Thöne et al. 2026 (J Biol Chem, Jentsch lab) is titled "A protective cGAMP-mediated
anti-tumor immune response can proceed without LRRC8/VRAC channels" and its abstract states
PMID:41419196 and
PMID:41419196.

But the same paper, in its own results, confirms the molecular function. Using
Lrrc8a-disrupted MC38 cells at endogenous expression:
PMID:41419196
and
PMID:41419196.
Their own summary of scope is
PMID:41419196, and they explicitly accept the T-cell result of others
PMID:41419196.
They also note cell-type dependence: B16-F10 melanoma cells, unlike MC38, barely use VRAC
PMID:41419196.

Adjudication

The logical shape matters. A knockout showing that a physiological outcome (tumour growth,
antitumour immunity in MC38/B16-F10 syngeneic grafts) proceeds normally without the protein
does not show that the protein cannot transport the substrate. Thöne et al. refute a
requirement for VRAC in one in vivo process; in the same experiments they positively
demonstrate the transport. Nothing in the 2026 paper supports an "overexpression artifact"
reading, and no paper I found claims one.

Curation call:
- GO:0140360 cyclic-GMP-AMP transmembrane transporter activity (IDA + IEA) — not
removed, not an artifact. Marked KEEP_AS_NON_CORE.
- GO:0140361 cyclic-GMP-AMP transmembrane import across plasma membrane (IBA + IEA + IDA)
— same treatment, KEEP_AS_NON_CORE.

Non-core rather than core because (i) it is one permeant among many for a large-pore,
broadly selective channel, (ii) it requires LRRC8C or LRRC8E and is suppressed by LRRC8D,
so it is a property of particular heteromers rather than of LRRC8A as such, and (iii) its in
vivo weight is demonstrably context-dependent — dominant in MC38, negligible in B16-F10,
pivotal in irradiated tumours and in T cells, dispensable for the MC38 antitumour response.
Removing or demoting the IDA would also violate the project rule against overruling an
experimental annotation from an abstract; here I have the full text of both contested papers
and they agree with the annotation at the molecular-function level.

4. Other annotation notes

• GO:0005515 protein binding (4 IPI rows, PMIDs 24790029, 28514442, 33961781, 40205054) —
  MARK_AS_OVER_ANNOTATED per project guidance. The biology behind them (LRRC8A–LRRC8B/C/D/E
  heteromerization) is already captured by GO:0034702 and GO:0034214.
• GO:0042802 identical protein binding — kept as non-core. Unlike bare protein binding this
  is informative (LRRC8A homohexamer, PMID:30095067, PMID:30127360), but it is an assembly
  property, not the channel function.
• GO:0005253 monoatomic anion channel activity (IMP, contributes_to) — MODIFY to
  GO:0005225. The contributes_to qualifier is apt; the term is simply the generic parent of
  the volume-sensitive term the same experiments established.
• GO:0006820 monoatomic anion transport (IMP x2) — MODIFY to GO:0098656 (transmembrane).
• GO:0016020 membrane (IDA, HDA) — MODIFY to GO:0005886.
• GO:0005737 cytoplasm (IBA, is_active_in) — kept as non-core, not corrected. Defensible
  in the trivial sense (the LRR domains are cytosolic and the Lyso-VRAC pool lies within the
  cytoplasm) but uninformative next to plasma membrane / lysosomal membrane. I did not
  inspect the PAINT tree, so per project rules I did not issue a corrective IBA action with
  fabricated propagation metadata.
• Organismal/tissue phenotypes transferred from mouse or rat (GO:0001678 glucose
  homeostasis, GO:0032024 insulin secretion, GO:0002329 pre-B cell differentiation,
  GO:0007283 spermatogenesis, GO:0045663 myoblast differentiation) — all
  KEEP_AS_NON_CORE. They are real downstream consequences of VRAC activity, well documented
  (PMID:29371604; PMID:14660746) but they are not the molecular activity.
• GO:0015734 taurine and GO:0015810 aspartate transmembrane transport — accepted; these
  are the classic VSOAC organic-osmolyte fluxes (PMID:24790029, PMID:28193731).

5. Verification log

PMIDs confirmed via PubMed MCP (title/journal/year) or the cached PubMed record header:
24725410 (Cell 2014), 24790029 (Science 2014), 26824658 (Cell 2016), 29769723 (Nature 2018),
30095067 (eLife 2018), 30127360 (Nat Struct Mol Biol 2018), 33139539 (PNAS 2020),
33171122 (Mol Cell 2020), 14660746 (J Clin Invest 2003), 24782309 (J Biol Chem 2014),
19946888 (J Mass Spectrom 2010), 28514442, 33961781, 40205054, 28193731 (J Cell Sci 2017),
29371604 (Nat Commun 2018), 32277911 (Immunity 2020), 35105987 (Nat Immunol 2022),
41371222 (Mol Cell 2025), 41419196 (J Biol Chem 2026).