AIGR Gene Hypothesis Deep Research — LPGAT1 (Q92604)

Hypothesis slug: secondary-acylation-capacities-and-lipid-pathway-scope
Focus type: function_assignment
Gene: LPGAT1 / LPLAT7 (Homo sapiens, NCBITaxon:9606), UniProt Q92604

Executive Judgment

Verdict: Partially supported.

Human LPGAT1's demonstrable, reproducible primary molecular function is that of an endoplasmic-reticulum, sn-1-specific, stearoyl-CoA-preferring lysophospholipid acyltransferase acting on lysophosphatidylethanolamine (LPE) and lysophosphatidylcholine (LPC). This is established by three independent groups using different constructs, organisms, and readouts (recombinant human LPLAT7 screen, bacterially expressed murine LPGAT1, and skeletal-muscle knockout lipidomics). Against this core, the seed hypothesis proposes three "secondary" capacities: (1) monoacylglycerol acyltransferase (MGAT)/triacylglycerol (TAG) biosynthesis, (2) PG/PI remodeling, and (3) acylation of 1-acyl-lysophosphatidylserine (LPS). Assessed separately with attention to donor/acceptor and sn-position, these secondary activities are real-but-re-ranked, context-specific, in-vivo-only, or paralog-confounded — not co-equal primary functions.

The seed's central question — "does newer predominant positional specificity exclude all older secondary activities, or do construct/assay/organism differences explain them?" — is answered the latter, with important nuance. The newer sn-1 LPE/LPC specificity re-ranks rather than erases the older activities. The phrase "poor is not zero" from PMID:36049524 is correct and matters for curation: LPI/LPG are weak but detectable intrinsic substrates, and the same paper's experimental evidence underlies the curated LPS reactions. However, several older annotations rest on weaker foundations than the sn-1 LPE/LPC core: MGAT is by-similarity to mouse (single human study), TAG "biosynthesis" is a downstream pathway-flux phenotype (knockout raises TAG), and PI/LPI remodeling has no curated UniProt reaction and is better attributed to the same-family paralog ALCAT1/LCLAT1.

Most important caveats: (i) sn-position assignments for lyso-substrates are confounded by non-enzymatic acyl migration, so "sn-1 vs sn-2" claims must be read cautiously; (ii) the seed's exact phrase "acylates 1-acyl-lysophosphatidylserine" maps to a curated RHEA reaction (RHEA:37403) but with weak experimental backing; (iii) in-vivo lipidomic phenotypes reflect pathway flux and abundance, not necessarily direct intrinsic chemistry, and must not be curated as MF evidence.

Key Findings

Finding 1 — The core function is an ER sn-1-specific LPE/LPC acyltransferase, stearoyl-CoA preferring

Three independent lines of evidence converge on a single, tightly specified molecular function.

In vitro recombinant human screen (PMID:36049524(https://pubmed.ncbi.nlm.nih.gov/36049524/)). In a systematic siRNA + recombinant screen of 14 human LPLATs, LPLAT7/LPGAT1 was identified as an sn-1-specific lysophospholipid acyltransferase. The verified snippet is unambiguous: LPLAT7 "mainly incorporated several fatty acids into the sn-1 position of lysophosphatidylcholine (LPC) and lysophosphatidylethanolamine (LPE), with weak activities toward other lyso-PLs." In cells and knockout tissue, only C18:0 (stearoyl) PC/PE species were reduced, with a reciprocal rise in C16:0/C18:1 species — a signature of loss of a stearate-preferring sn-1 remodeling enzyme.

Independent recombinant murine assay (PMID:35131264(https://pubmed.ncbi.nlm.nih.gov/35131264/)). Bacterially expressed murine LPGAT1 "transferred saturated acyl-CoAs specifically into the sn-1 position of lysophosphatidylethanolamine (LPE) rather than lysophosphatidylglycerol and preferred stearoyl-CoA over palmitoyl-CoA as the substrate." Knockout abolished 1-LPE:stearoyl-CoA activity and shifted PE/PC composition from stearate toward palmitate. This directly demonstrates sn-1 LPE specificity, stearoyl-CoA preference, and — critically for the PG hypothesis — activity into LPE rather than LPG.

In vivo/ex vivo confirmation (PMID:37217003(https://pubmed.ncbi.nlm.nih.gov/37217003/)). In murine skeletal muscle, LPGAT1 sets the sn-1 18:0/16:0 (stearate/palmitate) profile of PC and PE; "LPGAT1 knockout decreased the incorporation of stearate into PC and PE."

Structural/database orientation. UniProt Q92604 curates the enzyme as an ER-membrane, PlsC/AGPAT-family acyltransferase (Pfam PF01553/PF16076; InterPro IPR002123) with the catalytic HX4D motif (residues ~101–106) and two transmembrane helices. The primary curated catalytic reaction is 2-acyl-LPE + stearoyl-CoA → 1-stearoyl-2-acyl-PE.

Together these establish the core MF (sn-1 LPE/LPC O-acyltransferase, stearoyl-CoA-preferring), core BP (PE/PC sn-1 acyl-chain remodeling), and core CC (ER membrane).

Finding 2 — The secondary activities (MGAT, PG, PI, PS, TAG) are real but weak, context-specific, or in-vivo-only

MGAT/TAG. PMID:20018982(https://pubmed.ncbi.nlm.nih.gov/20018982/) identified mouse LPGAT1 as an acyl-CoA:monoacylglycerol acyltransferase; hepatic MGAT activity fell with LPGAT1 shRNA and serum TAG dropped ("we identified a novel MGAT gene, which is identical with lysophosphatidylglycerol acyltransferase1 (LPGAT1)"). This is direct but restricted to mouse liver, from a single group, and not reproduced by the later sn-1-focused studies. Crucially, PMID:42173283(https://pubmed.ncbi.nlm.nih.gov/42173283/) shows that LPGAT1/Lplat7 knockout increases triglyceride and lipid-droplet accumulation — the opposite of what a direct, rate-limiting TAG synthase would produce on deletion. TAG "biosynthesis" is therefore best read as a downstream pathway-flux/abundance phenotype, not intrinsic chemistry.

PG remodeling. The original human enzyme characterization (PMID:15485873(https://pubmed.ncbi.nlm.nih.gov/15485873/)) demonstrated LPG acyltransferase activity, and knockout produces aberrant PG acyl composition and a MEGDEL-like disease (PMID:37917582(https://pubmed.ncbi.nlm.nih.gov/37917582/)). Yet LPG is a poor intrinsic substrate relative to LPE/LPC (PMID:36049524), and PMID:35131264 explicitly shows acylation into LPE "rather than" LPG. So PG remodeling is experimentally real in human but minor in intrinsic ranking, and is shared with ALCAT1/LCLAT1 and CRLS1.

PI remodeling. LPI is a poor substrate (PMID:36049524), and the 2004 human assay found no LPI activity: "no significant acyltransferase activities were detected against glycerol 3-phosphate or a variety of lysophospholipids, including lysophosphatidylcholine, lysophosphatidylethanolamine, lysophosphatidylinositol, and lysophosphatidylserine." (Note this early assay also missed LPC/LPE, so absence there reflects assay limitations — but the PI negative is consistent with later "poor substrate" data.) The same-family paralog ALCAT1/LCLAT1 carries robust LPIAT activity (PMID:19075029(https://pubmed.ncbi.nlm.nih.gov/19075029/)), creating a paralog-attribution risk for the PI-remodeling annotation (GO:0036149, IBA only, no curated UniProt reaction).

LPS acylation. No isolated in vitro human assay demonstrates lyso-PS acylation as a strong activity; the RHEA-curated PS reactions are attributed to PMID:36049524's "weak activities toward other lyso-PLs." In vivo support comes from PMID:42173283: Lplat7 knockout "reduced the abundance of 1-stearoyl-2-unsaturated species of phosphatidylcholine, phosphatidylethanolamine, and phosphatidylserine" — supporting an sn-1 PS remodeling contribution in vivo, alongside PC and PE.

Finding 3 — RHEA/UniProt evidence codes refine sn-position and provenance of the secondary activities

A programmatic pull of UniProt Q92604 catalytic-activity comments with RHEA IDs and ECO codes sharpens the picture in ways that matter directly for curation:

• PS acylation is curated as EXPERIMENTAL (ECO:0000269, PubMed:36049524), not merely electronic. Two reactions exist: RHEA:74815 (sn-1 acylation of 2-acyl-LPS, adding palmitoyl) and RHEA:37403 (sn-2 acylation of 1-oleoyl-LPS, adding stearoyl). The seed's exact phrase "acylates 1-acyl-lysophosphatidylserine" maps to RHEA:37403 (sn-2 addition onto a 1-acyl acceptor) — an important distinction from the enzyme's sn-1 preference on LPE/LPC.
• MGAT/DAG reactions (RHEA:16741/65096/65100, 2-acylglycerol → 1,2-diacylglycerol) are ECO:0000250 by-similarity to MOUSE Lpgat1 (UniProtKB:Q91YX5) — i.e., no human experimental support in UniProt.
• All 9 PG reactions are ECO:0000269 PubMed:15485873 (human experimental, 2004).
• LPE/LPC sn-1 reactions are ECO:0000269 PubMed:36049524.
• UniProt curates NO lysophosphatidylinositol (LPI) reaction for LPGAT1, despite GO:0036149 (PI remodeling) existing as an IBA annotation.

Paralog check. LPGAT1 (PANTHER PTHR10983:SF2) and LCLAT1/ALCAT1 (Q6UWP7, PTHR10983:SF16) are same-family AGPAT/PlsC paralogs sharing Pfam PF01553+PF16076 and InterPro IPR002123/IPR032098. CRLS1 (Q9UJA2) is a distinct family (PTHR14269, PF01066). This supports the interpretation that anionic-lysophospholipid (LPI/LPG) activities historically attributed to LPGAT1 may partly belong to ALCAT1.

Mechanistic Model / Interpretation

 Lands cycle (sn-1 remodeling), ER membrane
 ============================================
  2-acyl-LPE  +  stearoyl-CoA  --LPGAT1-->  1-stearoyl-2-acyl-PE   [CORE, experimental x3]
  2-acyl-LPC  +  stearoyl-CoA  --LPGAT1-->  1-stearoyl-2-acyl-PC   [CORE, experimental]

  Preference:  stearoyl-CoA (18:0) > palmitoyl-CoA (16:0)
  Position:    sn-1 specific on LPE/LPC
  Phenotype:   KO shifts PC/PE from 18:0 -> 16:0/18:1 (reciprocal)

 Secondary / re-ranked capacities
 ================================
  2-acyl-LPS  --LPGAT1-->  PS (RHEA:74815, sn-1)   [WEAK experimental, PMID:36049524]
  1-acyl-LPS  --LPGAT1-->  PS (RHEA:37403, sn-2)   [<= seed's exact phrase; WEAK]
  LPG         --LPGAT1-->  PG (RHEA x9)            [human experimental but POOR substrate;
                                    shared with ALCAT1/CRLS1]
  2-MAG       --LPGAT1-->  DAG -> TAG              [MOUSE by-similarity only; TAG = flux]
  LPI         --??-->      PI                      [NO curated reaction; likely ALCAT1]

  Downstream / abundance phenotypes (NOT intrinsic chemistry):
    - KO INCREASES TAG / lipid droplets (PMID:42173283) -> opposite of a TAG synthase
    - KO causes MEGDEL-like disease via PG remodeling + mito transport (PMID:37917582)

The unifying model: LPGAT1 is a Lands-cycle sn-1 acyl remodeler whose bona-fide, reproducible substrates are LPE and LPC with a stearoyl-CoA preference. Its promiscuity toward anionic and other lysophospholipids (LPG, LPI, LPS) is genuine but intrinsically weak; whether a given weak activity produces a physiological phenotype depends on tissue context, substrate availability, and the presence of overlapping paralogs (ALCAT1, CRLS1). The dramatic in-vivo phenotypes (MEGDEL syndrome, TAG accumulation, reproductive defects) are systems-level consequences of perturbing membrane acyl composition and mitochondrial lipid supply, not direct evidence that PG/TAG/PS synthesis is the enzyme's primary chemistry.

Evidence Base / Evidence Matrix

GO Curation Implications

The following are leads requiring curator verification.

Retain as core:
- MF: sn-1-specific lysophospholipid O-acyltransferase acting on LPE/LPC (e.g., 1-acylglycerophosphoethanolamine O-acyltransferase; stearoyl-CoA preferring). Evidence: PMID:36049524, PMID:35131264 (both ECO:0000269-grade).
- BP: phosphatidylethanolamine/phosphatidylcholine acyl-chain remodeling (Lands cycle, sn-1). Evidence: PMID:37217003, PMID:35131264.
- CC: endoplasmic reticulum membrane. Evidence: UniProt Q92604, PMID:15485873.

Retain with a weak/secondary qualifier:
- MF: 1-acylglycerophosphoserine O-acyltransferase (GO:0106263 / RHEA:37403,74815). Backed by experimental evidence (ECO:0000269, PMID:36049524) but explicitly the "weak activities toward other lyso-PLs." Curator should note it is secondary and that the seed's "1-acyl-LPS" phrasing corresponds to the sn-2 reaction RHEA:37403.

Demote to non-core (retain but flag as by-similarity / context-specific):
- MGAT (GO:0003846) and TAG biosynthesis (GO:0019432). MGAT reactions are ECO:0000250 by-similarity to mouse only; TAG "biosynthesis" is a downstream flux phenotype (KO raises TAG). Neither should be presented as a primary human function.

Flag for removal or generalization:
- PI remodeling (GO:0036149, IBA only). No curated UniProt reaction; LPI is a poor substrate; ALCAT1/LCLAT1 is a strong LPIAT paralog. High risk of paralog carry-over. Recommend removal or generalization to "glycerophospholipid remodeling."

PG remodeling: retain (human experimental, PMID:15485873) but annotate as minor/shared, given LPG is a poor substrate relative to LPE/LPC and the activity overlaps with ALCAT1/CRLS1.

Avoid "protein binding" as a final recommendation — the informative MF terms above are supported. Note: LPGAT1's interaction with the prohibitin/TIMM14 import motor (PMID:37917582) could justify a specific protein-binding annotation with a named partner, but that is orthogonal to the acyltransferase function under review.

Mechanistic Scope

The immediate molecular function under test is acyl-CoA–dependent O-acyltransferase chemistry on lysophospholipid acceptors. The direct, gene-product-intrinsic activity is transfer of a (preferentially stearoyl) acyl chain onto the sn-1 position of LPE and LPC. Everything else in the seed hypothesis must be separated from this:

• Direct chemistry (intrinsic): sn-1 LPE/LPC acylation (strong); LPG, LPI, LPS acylation (weak, "poor is not zero").
• Pathway consequences / flux: PG remodeling feeding cardiolipin synthesis and mitochondrial cristae; TAG abundance changes.
• Downstream phenotypes / disease: MEGDEL-like syndrome, dilated cardiomyopathy, 3-methylglutaconic aciduria (PMID:37917582); hepatic lipid accumulation (PMID:41507908); LUAD proliferation (PMID:37258473); zebrafish reproductive/developmental defects (PMID:35508627); BMI/body-fat associations (PMID:23505186) and plasma 18:0 GWAS signals (PMID:23362303).

Only the first category constitutes MF evidence. The second and third support BP and disease-association annotations but must not be curated as direct enzymatic function.

Conflicts and Alternatives

1. Positional-specificity re-ranking vs exclusion. The newer sn-1 LPE/LPC data (PMID:36049524, PMID:35131264) do not exclude older secondary activities; they re-rank them. "Poor is not zero" is the correct reading. Construct (bacterial murine vs recombinant human vs endogenous), assay (in vitro vs cellular vs KO lipidomics), and organism differences explain most apparent contradictions.

2. Paralog confusion (ALCAT1/LCLAT1; CRLS1). ALCAT1 robustly carries LPIAT and LPGAT activities (PMID:19075029) and is a same-family AGPAT/PlsC paralog; CRLS1/hCLS1 has strong LPG acyltransferase activity (PMID:20025994). Anionic-lysophospholipid activities historically credited to LPGAT1 (especially PI and part of PG) may belong to these paralogs.

3. Organism/tissue specificity. MGAT/TAG evidence is mouse-liver (PMID:20018982); sn-1 PC/PE evidence spans human recombinant, murine recombinant, and murine muscle. Human-experimental support for MGAT is limited.

4. sn-position confound (acyl migration). Lyso-phospholipid substrates undergo non-enzymatic acyl migration between sn-1 and sn-2, so "sn-1 vs sn-2" assignments (including the seed's "1-acyl-LPS" → RHEA:37403 sn-2 mapping) carry inherent uncertainty.

5. Loss-of-function ≠ direct chemistry. The MEGDEL-like disease (PMID:37917582), TAG accumulation (PMID:42173283), reproductive/developmental defects (zebrafish, PMID:35508627), and cancer proliferation phenotypes (PMID:37258473) are downstream/pleiotropic consequences, not evidence of primary MGAT/TAG/PG chemistry.

6. Era-limited negative results. PMID:15485873 found no LPC/LPE activity — later shown to be the core function — so its negative LPI/LPS results should be weighted as assay-limited, not definitive.

Limitations and Knowledge Gaps

This report relied on the supplied knowledge outline (17 papers, 3 confirmed findings) and public database/RHEA provenance; no new wet-lab data were generated. sn-position claims inherit acyl-migration uncertainty. Several key phenotypic studies are murine or single-group. UniProt/RHEA evidence codes were used as provenance orientation and specific reaction IDs should be re-verified against the current UniProt release before curation.

Proposed Follow-up Experiments / Discriminating Tests

1. Purified human LPGAT1 kinetic panel: measure V_max/K_m for LPE, LPC, LPG, LPI, LPS (both 1-acyl and 2-acyl regioisomers) and 2-MAG with matched acyl-CoA donors, including acyl-migration controls, to rank intrinsic activities on a common scale.
2. Paralog side-by-side: express human LPGAT1 vs ALCAT1/LCLAT1 (and CRLS1) in the same system and assay LPIAT/LPGAT to resolve PI/PG attribution.
3. Isoform/organism control: compare human vs mouse enzyme MGAT activity under identical conditions to test whether MGAT is species-specific.
4. In vivo flux vs abundance: isotope-tracing lipidomics in an LPGAT1-specific KO to separate direct acylation flux from steady-state abundance changes (especially for TAG and PS).
5. Structural docking / active-site modeling: use the AGPAT/PlsC fold and HX4D motif to predict headgroup selectivity and rationalize weak anionic-lysophospholipid activity.

Curation Leads (require curator verification)

Candidate references + exact snippets to verify:
- PMID:36049524 — "mainly incorporated several fatty acids into the sn-1 position of lysophosphatidylcholine (LPC) and lysophosphatidylethanolamine (LPE), with weak activities toward other lyso-PLs" → supports core MF + "poor is not zero."
- PMID:35131264 — "transferred saturated acyl-CoAs specifically into the sn-1 position of lysophosphatidylethanolamine (LPE) rather than lysophosphatidylglycerol and preferred stearoyl-CoA over palmitoyl-CoA" → supports sn-1 LPE core; refutes strong PG.
- PMID:20018982 — "we identified a novel MGAT gene, which is identical with lysophosphatidylglycerol acyltransferase1 (LPGAT1)" → MGAT, but mouse liver / by-similarity in UniProt.
- PMID:15485873 — "no significant acyltransferase activities were detected against ... lysophosphatidylinositol, and lysophosphatidylserine" → historical negative for LPI/LPS.
- PMID:42173283 — "reduced the abundance of 1-stearoyl-2-unsaturated species of phosphatidylcholine, phosphatidylethanolamine, and phosphatidylserine" → in-vivo PS support; note KO raises TAG.
- PMID:19075029 — "HEK293 cells overexpressing human ALCAT1 demonstrated significant increases in LPI acyltransferase (LPIAT) and LPG acyltransferase (LPGAT) activities" → paralog alternative.

Candidate GO actions:
- Keep: sn-1 LPE/LPC acyltransferase MF; PE/PC remodeling BP; ER membrane CC.
- Keep with weak qualifier: 1-acylglycerophosphoserine O-acyltransferase (GO:0106263).
- Demote to non-core: MGAT (GO:0003846), TAG biosynthesis (GO:0019432).
- Flag for removal/generalization: PI remodeling (GO:0036149, IBA).
- Retain but mark minor/shared: PG remodeling.

Suggested curator questions:
- Is any human-experimental MGAT activity available beyond PMID:20018982?
- Should GO:0036149 (PI) be reassigned to ALCAT1/LCLAT1?
- Does the RHEA:37403 (sn-2 on 1-acyl-LPS) mapping justify the seed's "acylates 1-acyl-lysophosphatidylserine" as a curated MF, given only "weak activity" evidence?

Suggested experiments: the kinetic panel and paralog side-by-side assays under Discriminating Tests.

Conclusion

The seed hypothesis is partially supported. LPGAT1's primary, reproducible function is sn-1-specific LPE/LPC acylation with stearoyl-CoA preference at the ER. The proposed secondary capacities are genuinely present in the literature but are re-ranked, not erased, by the newer positional-specificity data: LPS acylation is weakly experimental (retain with qualifier), MGAT/TAG are by-similarity/downstream (demote to non-core), PG remodeling is minor and shared, and PI remodeling is best attributed to the ALCAT1 paralog (flag for removal). Construct, assay, and organism differences — not mutual exclusion — explain the apparent tension between older and newer studies.

Artifacts

• OpenScientist final report
• OpenScientist final report