MSP1 (Plasmodium falciparum 3D7, Q8I0U8, PF3D7_0930300) — curation notes

Provenance note: automated deep research was not available in this environment
(Falcon authentication failed and the OpenAI key was invalid), so no
*-deep-research-*.md file was generated. The notes below are my own manual
literature review, built from the UniProt flat file, the cached publications in
publications/, and additional PMIDs I fetched into that cache
(uv run ai-gene-review fetch-pmid). Every assertion carries an inline citation with a
verbatim quote where possible.

1. What the protein is

MSP1 is the most abundant GPI-anchored protein of the P. falciparum merozoite surface
coat, made as a ~190–200 kDa precursor (1720 aa in 3D7) during schizogony
[file:PLAF7/MSP1/MSP1-uniprot.txt "RecName: Full=Merozoite surface protein 1"].
It has no orthologues outside Plasmodium
PMID:26468747,
so electronic family-based transfer has little to offer here — the annotation base is
almost entirely direct experimental work on the parasite itself.

GPI anchoring is established biochemically rather than merely predicted
PMID:8992312.

2. Proteolytic maturation defines its states

Two ordered processing steps generate the functional forms:

• Primary processing by SUB1, in the parasitophorous vacuole minutes before egress,
  cuts the precursor into p83/p30/p38/p42, which stay non-covalently associated on the
  merozoite surface
  PMID:19214190.
  The order of cleavage matters for viability
  PMID:20735778.
• Secondary processing by SUB2 during invasion cleaves p42 into p33 + p19; the whole
  complex is shed except GPI-anchored p19
  PMID:16322767
  PMID:1694225.

Note for review purposes: in the SUB1/SUB2 papers MSP1 is the substrate, not the
protease — those references support MSP1's location (surface, then shed/extracellular),
not any peptidase activity of MSP1.

3. Core function 1 — egress via host spectrin

The best-defined mechanistic role. SUB1 processing converts MSP1 into a spectrin-binding
form, and loss of surface MSP1 blocks egress:

• PMID:26468747
• PMID:26468747
• PMID:26468747
• Direct binding, with affinity, confirmed structurally:
  PMID:34078606
  and PMID:34078606

So MSP1 itself does work in egress (it binds and perturbs the host cytoskeletal lattice);
it is not merely required for it. This passes the participation test for a process term,
and the host-directed nature of the act makes the symbiont/host branch
(GO:0052039, GO:0035891) the right place for it.

4. Core function 2 — adhesion to host erythrocyte during invasion

• MSP1-42/19 binds the exofacial 5ABC segment of host band 3 (SLC4A1), sialic
  acid-independently
  PMID:12692305
  PMID:12692305
• MSP9 and MSP1-42 form a co-ligand complex on the same receptor
  PMID:14630931
• The N-terminus binds host glycophorin A
  PMID:25778531
• Band 3 engagement also involves rhoptry RhopH3 in complex with MSP1-19
  PMID:25157665
• Anti-p19 antibodies block invasion
  PMID:36202833

These justify a host-directed molecular function term, GO:0046789 host cell surface receptor binding, which GOA currently lacks for this gene (it only has bare
GO:0005515).

5. Core function 3 — scaffold for the merozoite surface coat

MSP1 nucleates a large peripheral surface complex:

• PMID:16940297
• PMID:25074930
  — note MSP1 itself did not bind erythrocytes in that assay
  PMID:25074930,
  which is a genuine tension with the band 3 / GPA binding literature (different
  fragments, different assay formats) and worth flagging to experts.
• PMID:29511044
• Rhoptry proteins co-purify with MSP1 on the merozoite surface
  PMID:21175202

6. Accessory activities

• p33 sequesters host S100P:
  PMID:22431641
  — host-directed immune dampening; the appropriate branch is
  GO:0052170 symbiont-mediated suppression of host innate immune response. Evidence is
  in vitro with recombinant protein, so it is a defensible but non-core addition.
• p19 marks and may build the food (digestive) vacuole after invasion:
  PMID:18769730
  PMID:18769730
  UniProt is careful here — "May play a role in the biogenesis and/or function of the food
  vacuole" is flagged ECO:0000305 (Probable). So the location (vacuolar membrane) is
  solid; the process (food vacuole biogenesis) is not, and I do not propose a process
  term for it.
• PFD6 (prefoldin subunit 6) interaction is reported
  PMID:33145997
  but UniProt records it as "May interact", and it is not in GOA; not proposed.

7. Protist-specific checks performed

• No animal/plant-only terms present in the GOA set — nothing to strip on taxon grounds.
• Organelles: the only organellar term is GO:0005774 vacuolar membrane, which here
  means the food/digestive vacuole membrane (PMID:18769730), a real apicomplexan
  compartment. GO has GO:0020020 food vacuole but no "food vacuole membrane" term —
  recorded as a proposed_new_terms entry rather than forcing a bad id. No apicoplast,
  rhoptry or microneme claim is made for MSP1 (the rhoptry proteins RhopH3/RAP1 are
  partners, not evidence that MSP1 is in a rhoptry).
• Host vs symbiont: all host-directed activity is annotated to the symbiont/host
  branch (GO:0085017, and proposed GO:0046789, GO:0035891, GO:0052039,
  GO:0052170), never to the plain host-cell-biology terms.
• Paralog family: MSP1 is a single-copy, Plasmodium-restricted gene; the MSP3/MSP7
  "families" are distinct genes, not MSP1 paralogs, so there is no paralog-family
  over-annotation risk. The single IEA from InterPro (IPR010901, GO:0016020 membrane)
  is merely uninformative, not wrong.

8. Curation decisions taken

(Revised after coordinator review; the earlier draft of this section is superseded.)

• All four GO:0005515 protein binding rows are MODIFY, never kept as bare binding:
  the PMID:21175202 (RhopH3/RAP1/CLAG9) and PMID:25074930 (MSPDBL1) rows to
  GO:0140378 protein complex scaffold activity; the PMID:25157665 row (MSP1-19 with
  RhopH3 on the band 3 5ABC segment) to GO:0046789 host cell surface receptor binding;
  the PMID:22431641 row to GO:0044548 S100 protein binding.
• GO:0005886 plasma membrane accepted on all seven rows (core location); the extra
  specificity from the GPI-anchor biochemistry (PMID:8992312) is added as a separate NEW
  row for GO:0031362 anchored component of external side of plasma membrane.
• GO:0005576 extracellular region kept but non-core: the shed complex is real but not
  where MSP1 does its work.
• Both GO:0005774 vacuolar membrane rows downgraded to KEEP_AS_NON_CORE. The
  localisation is directly observed by two methods, so it is not an artefact to remove or
  mark over-annotated, but it is the post-invasion fate of the GPI-anchored p19 remnant,
  carried there by endocytic uptake of the surface coat — the paper offers a functional
  role only as a possibility ("a potential role", "may play a significant role") and
  UniProt flags it ECO:0000305 Probable. Not a core location.
• The proposed new term "food vacuole membrane" has been dropped: with this
  localisation judged non-core and possibly a trafficking endpoint, MSP1 is not a good
  basis for proposing the term, and GO:0005774 is adequate for the claim being made.
• GO:0016020 membrane (IEA/InterPro) marked TOO_GENERAL — entailed by the accepted
  plasma membrane term.
• Three NEW rows only, all passing the participation test: GO:0031362 (location),
  GO:0030507 spectrin binding (two labs, orthogonal methods) and
  GO:0035891 exit from host cell (MSP1 itself binds and engages the host spectrin lattice
  that must give way; loss of surface MSP1 gives a severe egress defect).
• Dropped from the earlier draft: GO:0052170 symbiont-mediated suppression of host innate immune response (rests on a single in vitro study — now a suggested_question,
  with the binding itself retained as GO:0044548) and
  GO:0052039 symbiont-mediated perturbation of host cytoskeleton (the perturbation is
  inferred from binding plus the rupture phenotype, never measured — now a
  suggested_question paired with the membrane-mechanics experiment). The GO:0046789
  and GO:0140378 NEW rows were also dropped as redundant once they became
  proposed_replacement_terms on the protein-binding rows.
• No REMOVE anywhere: nothing in the GOA set is contradicted by the literature.
• Validation (just validate PLAF7 MSP1) passes with no errors and no protein-binding
  warnings.