Mouse Mpo (P11247) — curation notes

Session scope: issue #4098,
the in-repo follow-ups left over from the hypochlorous acid (HOCl) GO obsoletion
project (projects/HYPOCHLOROUS_ACID_OBSOLETION.md). No deep-research provider
file was generated for this gene (no provider credentials in this environment);
these notes plus the cached UniProt record and publications are the evidence base.

Why this gene was reviewed now

GO obsoleted all three HOCl process terms. QuickGO (checked this session)
returns GO:0002148, GO:0002149 and GO:0002150 as obsolete, with the comment
"The reason for obsoletion is that this BP term represents a MF term". The only
experimental annotation in scope was the MGI IMP of mouse Mpo to GO:0002149
(PMID:10085024), with a rat ISO row copied from it.

That obsoletion rationale is the key to the review: the HOCl chemistry belongs
in the molecular-function aspect, and in this GOA set it is already there —
GO:0004601 peroxidase activity IDA and IMP, both from PMID:10085024, the same
paper that backed the obsoleted BP term. The freshly fetched GOA (85 data rows, 84
distinct annotations - the GO:0140644 ISO located_in row appears twice,
differing only in date) no longer contains GO:0002149 at all, so the migration has already happened
upstream; nothing in this repo had to be rewritten to drop it.

The MF-term question (issue item 2)

The issue text names RHEA:43232 as the myeloperoxidase reaction. That is
wrong, and the error is worth recording so it is not propagated: RHEA:43232 is
isatin + H2O = isatinate + H(+) (EBI search, this session). The
myeloperoxidase reaction is RHEA:28218, as the UniProt record itself states
[UniProtKB:P11247, "Reaction=chloride + H2O2 + H(+) = hypochlorous acid + H2O;
Xref=Rhea:RHEA:28218 ... EC=1.11.2.2"].

Searching GO for a matching MF term (QuickGO term search for "myeloperoxidase",
"hypochlorous", "hypochlorite"; plus an xref scan of the local go.db for
RHEA:28218, EC:1.11.2.2) found no term for this reaction:

• GO:0016691 chloride peroxidase activity — "Catalysis of the reaction: 2 R-H
• 2 chloride + H2O2 = 2 R-Cl + 2 H2O", xref EC:1.11.1.10. Organic
  chlorination, not HOCl formation. Confirmed as the issue suspected.
• GO:0140825 lactoperoxidase activity — "Catalysis of the reaction: 2 a
  phenolic donor + H2O2 = 2 a phenolic radical donor + 2 H2O", xref EC:1.11.1.7
  and RHEA:56136. Not the HOCl reaction; the issue is right to warn against it.
• GO:1901531 hypochlorite binding / GO:1901530 response to hypochlorite —
  the only other live HOCl-named terms; neither is a catalytic term.
• GO:0004601 peroxidase activity — carries "myeloperoxidase activity" as a
  narrow synonym and the Reactome xrefs R-HSA-6789126 "Myeloperoxidase (MPO)
  produces hypochlorous acid (HOCl)" and R-HSA-6789031 "Membrane-bound
  myeloperoxidase (MPO) produces hypochlorous acid (HOCl)" (local go.db
  axiom annotations). So GO:0004601 is where GO currently puts this activity.

Conclusion recorded in the review: GO:0004601 is the most precise existing term
and is accepted; a child term is proposed in proposed_new_terms
("hypochlorous acid-forming chloride peroxidase activity", parent GO:0004601,
skos:exactMatch RHEA:28218). The precedent for the shape is GO:0036393 thiocyanate peroxidase activity (xref RHEA:69416), which exists for the
pseudohalide version of the same reaction and is itself annotated to mouse Mpo
by the Rhea pipeline. Creating the term is an upstream GO job, not an in-repo one.

Phenotype evidence (PMID:10085024, full text cached)

• The enzyme and the reaction: "Myeloperoxidase (MPO) catalyzes the reaction of
  hydrogen peroxide with chloride ion to produce hypochlorous acid (HOCl), which
  is used for microbial killing by phagocytic cells."
• Pathway position: "During the respiratory burst, neutrophils generate O 2 −
  via the NADPH oxidase system. O 2 − is then reduced to H 2 O 2 , from which
  HOCl is generated by the MPO reaction." This is why GO:0042744 hydrogen peroxide catabolic process (annotated IDA + IMP from this paper) is the
  mechanistically correct process home, and why GO:0002679 respiratory burst involved in defense response and GO:0019430 removal of superoxide radicals
  are marked as over-annotations here.
• Knockout chemistry: "the neutrophils of homozygous mutant mice produced HOCl
  at an almost undetectable level"; wild-type MPO activity was 1.7
  nmol/min/10^6 cells and tracked gene dosage.
• Superoxide went the wrong way for a superoxide-removal role: "O 2 −
  generation was slightly increased in our MPO-deficient mice, consistent with
  the deficiency in humans".
• Fungal defence: "they showed increased susceptibility to pneumonia and death
  following intratracheal infection with Candida albicans"; "MPO is important
  for early host defense against fungal infection, and the inability to generate
  HOCl cannot be compensated for by other oxygen-dependent systems in vivo in
  mice."
• The NOT annotation: "Homozygous mutant mice did not exhibit a remarkable
  difference from wild-type mice in the rate of clearance of S. aureus from the
  peritoneal cavity", and "It is thus likely that reactive oxygen species other
  than HOCl, such as O 2 − , are sufficient for neutrophils to kill S. aureus ."
  GOA records this as NOT|acts_upstream_of_or_within GO:0050830; accepted.

Halogenation in vivo (PMID:11593004, abstract only)

"In a mouse model of polymicrobial sepsis, we observed that mice deficient in
myeloperoxidase were more likely than wild-type mice to die from infection";
"Levels of 3-chlorotyrosine did not rise in the septic myeloperoxidase-deficient
mice"; and bromide is also used, "even though the extracellular concentration of
bromide is at least 1,000-fold lower than that of chloride". This is the mouse
support for the bacterial-defence rows (GO:0042742 IBA, GO:0070944).

Enzymology and regulation (PMID:23306200, full text cached)

Granule and phagosome localisation: "It is packaged in the azurophilic granules
of neutrophils and released into phagosomes when they ingest and kill bacteria";
extracellular release: "Myeloperoxidase is also released into the extracellular
space as a result of incomplete phagosome closure and during chronic
inflammation". Ceruloplasmin inhibition: "Ceruloplasmin was a potent inhibitor
of purified myeloperoxidase, inhibiting production of hypochlorous acid by 50%
at 25 nm", and the mouse-genetic part, "myeloperoxidase oxidized 75% of the
ascorbate in plasma from ceruloplasmin knock-out mice".

The second function: chromatin decondensation for NETs

Curated in UniProt from the human ortholog: "Promotes NET formation by mediating
chromatin disassembly: translocates to the nucleus and specifically binds to
nucleosomes, both as monomer and homodimer, leading to nucleosome unstacking and
initial chromatin decondensation" and "Homodimers clash with one end of the
nucleosomal DNA, leading to DNA unwrapping, initiating complete disassembly of
nucleosomes and chromatin transformation into NETs in an ATP-independent manner"
[UniProtKB:P11247]. This is kept as a second core function with MF
GO:0140776 protein-containing complex destabilizing activity — the same MF the
cached human NET GO-CAM gomodel:68fac5ed00000028 uses for MPO's nuclear
activity (the model's other MPO activity is GO:0004601 occurring in
GO:0140644 neutrophil extracellular trap).

Rows argued against, and why (issue-relevant summary)

• GO:0005739 mitochondrion (HDA, PMID:14651853) — REMOVE. A single
  high-throughput hit in a mouse mitochondrial proteome survey of brain, heart,
  kidney and liver. Mpo has no mitochondrial targeting and its curated locations
  are granule/lysosome, phagosome, secreted and nucleus; leukocyte
  co-purification is the parsimonious explanation.
• GO:0044194 cytolytic granule (IEA from UniProt SubCell SL-0543) — MODIFY to
  GO:0042582 azurophil granule. GO:0044194 is defined as a secretory lysosome
  "present in cells with cytolytic capability such as cytotoxic T lymphocytes
  and natural killer cells", which is not the neutrophil azurophil granule, and
  azurophil granule is not a descendant of it (checked in go.db).
• GO:0005764 lysosome, GO:0030141 secretory granule — MODIFY to
  GO:0042582 azurophil granule, which is an is_a descendant of both and is
  already annotated to Mpo.
• The rat-derived "response to X" rows (mechanical stimulus, food,
  lipopolysaccharide, gold nanoparticle) and GO:0006979 response to oxidative stress — MARK_AS_OVER_ANNOTATED. They report tissue MPO activity or
  expression as inflammation readouts, and the oxidative-stress one inverts the
  relationship for an enzyme that makes oxidants.
• GO:1904109 positive regulation of cholesterol import —
  MARK_AS_OVER_ANNOTATED; the underlying chemistry (APOA1 chlorination and
  nitration, which "impairs cholesterol transport") is kept non-core as
  GO:0034374 low-density lipoprotein particle remodeling.
• Experimental IDA rows whose cached records are abstract-only (PMID:11792727,
  PMID:11980719) are ACCEPTed with the caveat recorded in
  reference_review: the claim (peroxidase activity) is independently
  established, and the curator read the full text.

Review bookkeeping

The review has 84 existing_annotations entries against 85 TSV data rows: the
GO:0140644 neutrophil extracellular trap ISO located_in row is duplicated in
the download (same term, evidence, reference and qualifier; different date), so
it is reviewed once.

All twelve GO:0004601 rows are ACCEPT. Three of them do resolve the specific
chloride-to-HOCl reaction - the IDA and IMP from PMID:10085024 (HOCl measured in
neutrophils; abolished in the knockout) and the IDA from PMID:23306200 (purified
enzyme, chloride + H2O2) - while the other nine (tissue MPO assay, MPO staining
as a myeloid marker, the peroxisome-paper peroxidase control, and the
IBA/IEA/ISO/ISS transfers) show only generic turnover. Splitting them, with
MODIFY to an NTR replacement on the specific three, was tried and reverted:
the review guidance is that rows sharing a term must take consistent actions,
and the validator flags the split (inconsistent_review_actions). The
distinction is not lost - those three rows are named in the justification of
the proposed_new_terms entry and two of their quotes are in its supported_by,
which is where a GO editor weighing the term request will look.

Follow-ups that are not in-repo

• Creating the HOCl-forming MF term is a GO ontology request
  (proposed_new_terms here is the proposal text).
• geneontology/go-annotation#6404 is not readable from this session (only
  ai4curation/ai-gene-review is configured for the GitHub tools, and the
  anonymous API path is blocked), so its current comment state could not be
  re-checked. The GOA download shows the mouse row already gone from
  GO:0002149, which is the outcome that ticket was tracking.
• Rat Mpo (A0A0G2K1A2): QuickGO returns 65 rows this session, all ISO from
  MGI:97137 / UniProtKB:P05164, and none to GO:0002149 — so the rat row
  followed the mouse change, exactly as the project page predicted. No residual
  RGD follow-up.
• Human MPO (P05164): still unreviewed. Worth a review on its own merits (it is
  the ortholog behind most of the ISS/ISO rows here and appears twice in the
  cached NET GO-CAM), but it is not blocked by, and does not block, the
  obsoletion. Recorded as a plan item on the project page rather than done here.