G6PC2

Summary: Recombinant IGRP hydrolyses glucose 6-phosphate with Km 0.29-0.45 mM, and a selective small-molecule inhibitor distinguishes it from the liver enzyme. This is the paper UniProt ties the human entry's catalytic-activity line to.

Reason: Direct enzymology, and for human the catalytic claim is now established on the human protein itself rather than only transferred. Mutating one residue in each of the six regions of the predicted PAP2 motif of human G6PC2, including the four predicted catalytic residues Arg79, His115, Arg168 and His174, significantly reduces activity (PMID:38095063), and the human open reading frame gives measurable microsomal G6Pase activity in islet-derived 832/13 cells, where several nonsynonymous variants reduce it further (PMID:34954144). The abstract of this paper does not state the species of the construct, but UniProt attributes the human CATALYTIC ACTIVITY line to it with ECO:0000269, and the species-selective inhibitor VU0945627 blocks human but not mouse G6PC2 (PMID:38431189), which fits a human construct. Judged against the promiscuous-activity tiers: turnover is far below G6PC1 (PMID:34954144, PMID:38431189), but G6P hydrolysis is this protein's own reaction, G6PC2 owns the reaction in islets (human G6PC1 is absent from sampled beta cells and only weakly present in islets, file:human/G6PC2/G6PC2-bioinformatics/RESULTS.md), and human coding variants that reduce the protein or its activity shift fasting glucose. The low rate is what lets G6PC2 modulate rather than abolish glucokinase flux, so this is a core evolved function.

Summary: Glucose-6-phosphatase activity inherited from the PTHR12591 root node PTN000971890; the catalysis is shared by all three G6PC paralog clades and is experimentally established for human G6PC2 itself.

Reason: The IBD for GO:0004346 sits at the eumetazoan G6PC node and is seeded by experimental annotations of G6PC1, G6PC2 and G6PC3 from human, mouse and rat (interpro/panther/PTHR12591/PTHR12591-paint.tsv). Human G6PC2 (UniProtKB:Q9NQR9) appearing among the seeds is expected and is a marker that experimental grounding exists on the target: its own Reactome EXP row is one of the descendant evidences the PAINT curator used. The node placement holds for this gene on independent grounds: G6PC2 retains the PAP2 catalytic machinery (UniProt active sites His115 and His174, substrate-binding Arg79 and Arg168), mutating any of the six PAP2 regions of the human protein reduces activity (PMID:38095063), and the curated catalytic-activity line is tied to experimental evidence. `just paint-losses human G6PC2` reports no IRD or IKR on the lineage, so the row is current.

Summary: Combined electronic assignment from the glucose-6-phosphatase family signature (IPR016275), the Rhea reaction, EC 3.1.3.9 and the mouse ortholog.

Reason: The electronic mapping agrees with the experimental evidence for this protein, so the same claim as the accepted EXP and IBA rows. The ARBA component and the EC number are not themselves evidence, but the family assignment is: G6PC2 matches the glucose-6-phosphatase family signature (InterPro IPR016275, PIRSF000905, Pfam PF01569 PAP2, PANTHER PTHR12591:SF1), and the curated UniProt catalytic-activity line, tied to PMID:14722102 by ECO:0000269, gives the same reaction as the Rhea mapping. Human structure-function work confirms the predicted catalytic machinery is required (PMID:38095063).

Summary: ER location inherited from the G6PC family node; all three paralogs are ER-resident integral membrane enzymes, and the human protein was localised directly.

Reason: Root organelle term, which is informative and stays; it is not downgraded or replaced by the sub-compartment. Checking what the IBA rests on: the IBD at PTN000971890 is seeded by mouse G6pc1, G6pc2 and G6pc3 and human G6PC3, so the row stops at the organelle because that is what the seed annotations assert. The organelle claim holds for human G6PC2 on its own evidence: PMID:15044018 localised the human protein to the ER and showed it is held there by nine transmembrane domains, UniProt records the location with ECO:0000269 from that paper, Asn92 glycosylation was demonstrated by mutagenesis (a lumenal modification), and the C-terminal motif prevents secretion from the ER. The ER membrane sub-compartment is already carried by the IEA and TAS rows below, so no NEW sub-compartment row is needed.

Summary: Combined electronic assignment of ER location, from an ARBA rule and the mouse ortholog; correct for this protein.

Reason: Root organelle term, kept. Neither component is evidence in itself: the ARBA rule is an unvetted prediction and the mouse ortholog's own ER annotation is partly by-similarity (UniProt gives Q9Z186's location as ECO:0000250). The organelle claim is nevertheless independently established for the human protein by direct localisation and topology (PMID:15044018), which UniProt records as ECO:0000269 for this entry. Same claim as the accepted IBA row; the more specific ER membrane term is carried separately.

Summary: UniProt subcellular-location mapping to ER membrane; correct, experimentally based for human, and the right granularity.

Reason: Unlike the mouse entry, where this location is by similarity, the human UniProt statement carries ECO:0000269 from PMID:15044018, a topology and glycosylation study of the human protein. Several independent lines agree: nine predicted transmembrane helices with a lumenal N-terminus, the C-terminal ER-retention motif, Asn92 glycosylation shown by mutagenesis, and activity measured in isolated microsomal membranes (PMID:34954144). This is where the catalytic activity acts, with the active site facing the lumen.

Summary: Reactome's islet G6PC2 reaction places the catalysis on the ER membrane, with the ER lumen as the compartment the products are released into.

Reason: The curated statement matches the experimental evidence: the reaction is annotated to the ER membrane and ER lumen compartments and cites the enzymology paper (PMID:14722102), and the location is independently shown for the human protein (PMID:15044018). The lumenal compartment is the consequence of the topology, which is why substrate must be imported from the cytosol. Only this reaction's placement inside the Gluconeogenesis pathway is contested, in the GO:0006094 TAS row below; the location it asserts is sound.

Summary: Gluconeogenesis propagated from the root G6PC node to the islet paralog. Human G6PC2 is a pancreatic islet enzyme that cycles glucokinase-derived glucose 6-phosphate back to glucose to set the glucose sensitivity of insulin and glucagon secretion; it does not take part in making glucose from non-carbohydrate precursors.

Reason: The IBD for gluconeogenesis was placed at PTN000971890, the eumetazoan node ancestral to G6PC1, G6PC2 and G6PC3, and is seeded only by G6PC1 (human, mouse, rat), G6pc3 (mouse, rat) and a Drosophila member; no G6PC2 is a seed (interpro/panther/PTHR12591/PTHR12591-paint.tsv, dated 20241120). `just paint-losses human G6PC2` finds no IRD or IKR on the lineage, so the row is current and the disagreement is with the node reaching the G6PC2 clade (the path runs PTN000971890 to PTN002650241 to PTN000287500 to human PTN002488967, recorded in projects/BGEEDB_EXPRESSION/data/panther_leaves.tsv). Target-specific evidence shows that the process role diverged in that clade while the catalytic machinery was retained. (1) Human G6PC2 expression is predominantly pancreatic, with the protein in most islet cells, lower levels in testis, and low transcript in the other tissues Bgee samples (duodenum and kidney cortex score 38-51 against 99 in the islet); the cloning paper that established the islet pattern concluded that it points to a role in islet metabolism and insulin secretion. (2) In islet cells its substrate is G6P made from glucose by glucokinase, so its activity forms a futile substrate cycle that lowers glycolytic flux, not a route from pyruvate, lactate, glycerol or amino acids to glucose, which is what GO:0006094 denotes. Islet cells do not supply blood glucose. (3) The demonstrated consequences of the activity are hormonal set points: beta-cell deletion alone lowers fasting glucose in mouse (PMID:32213654), alpha-cell deletion lowers glucagon (PMID:39742505), and in human, coding variants that reduce the protein or its activity track with fasting glucose (PMID:34954144). (4) Bgee expression, used here only as a paralog-controlled consistency check, agrees but is weaker than in mouse: in the gluconeogenic tissues (liver, kidney cortex, small intestine) human G6PC2 has eight present calls and no absent calls, all low-scoring (best duodenum 50.5), with 2 of its top 10 tissues in that context, whereas G6PC1 is present at 97-98 in liver with 5 of its top 10 tissues there. Because human has no absent calls, expression cannot carry this decision at all and is cited as corroboration only. (5) The one direct measurement bearing on glucose production is human and nominal: among Danish elderly twins, carriers of the fasting-glucose-raising rs560887 G allele had higher basal hepatic glucose production (p = 0.04; PMID:19669124). It deserves more weight in a human review than in the mouse one, and it still does not establish participation: it is an unreplicated subgroup of 159 twins at a locus that also contains ABCB11, the same allele raised early insulin responses, and higher basal hepatic glucose output is the expected downstream consequence of a higher defended glucose level and lower insulin secretion. Pound et al. weighed exactly this alternative and judged it less likely because G6PC2 is expressed in islets and not liver (PMID:23274894). (6) A 2025 review co-authored by the O'Brien group describes G6PC1 and G6PC2 together as regulating glycolysis, gluconeogenesis and glycogenolysis (PMID:39426930); that sentence lumps the two paralogs, and the same review's point, like the same group's primary data, is that they have distinct tissue distributions, kinetics and physiological functions. No measurement of gluconeogenic flux attributes any of it to G6PC2. Taken together the inherited process term is an over-propagation to a paralog co-opted for islet glucose sensing; the UniPathway IEA and the Reactome TAS are removed for the same reason.

Summary: Electronic mapping from the UniProt PATHWAY line "Carbohydrate biosynthesis; gluconeogenesis", a family-level carry-over from G6PC1.

Reason: Same claim as the IBA row and removed for the same biological reasons. The mapped PATHWAY line carries no evidence tag, and the FUNCTION text it accompanies on Q9NQR9 is an entirely by-similarity statement inherited from G6PC1 ("May hydrolyze glucose-6-phosphate to glucose in the endoplasmic reticulum. May be responsible for glucose production through glycogenolysis and gluconeogenesis", ECO:0000250), which is now out of date for human: the hydrolysis is experimentally established on the human protein (PMID:38095063, PMID:34954144), and the physiology it serves is the islet glucose cycle, not glucose production. G6PC2 is not expressed at meaningful levels in the gluconeogenic tissues and nothing attributes glucose production there to it (PMID:23274894, PMID:32213654).

Summary: Reactome places its islet G6PC2 reaction inside the hepatic and renal Gluconeogenesis pathway, although the pathway's own summary confines gluconeogenesis to liver and kidney and the reaction's own summary places the enzyme in islet cells.

Reason: This row is the one gluconeogenesis assertion the mouse review did not have, and it is a curated statement, so it was examined on its own terms rather than dismissed as an electronic mapping. The assertion rests on pathway placement, not on any measurement: R-HSA-3266566, "G6PC2 hydrolyzes G6P to form Glc and Pi (islet)", is a contained event of R-HSA-70263 Gluconeogenesis (confirmed through the Reactome ContentService, which returns R-HSA-70263 for that reaction and lists the liver G6PC, ubiquitous G6PC3 and islet G6PC2 hydrolysis reactions side by side), and its only literature reference is the enzymology paper PMID:14722102, which measures the reaction and says nothing about gluconeogenesis. Reactome's own text contradicts the placement twice over: the pathway summary states that gluconeogenesis is confined to liver and kidney cells, and the reaction summary states that this enzyme is expressed in islet cells. The underlying biology is the same as for the IBA row: in islet cells the substrate is glucokinase-derived G6P, the reaction closes a futile cycle rather than completing a route from non-carbohydrate precursors, and islet cells do not export glucose to the blood. So the reaction is real and correctly curated, but its membership of this pathway is not, and GOA inherits the error as a gluconeogenesis annotation. Removed, consistently with the IBA and IEA rows, and recorded as a source defect for Reactome. Note the contrast with the human G6PC3 review, which keeps the equivalent Reactome row as non-core: there the ubiquitous paralog has its own experimental glucose-production evidence (PMID:15661744 in astrocytes), whereas no study places G6PC2 activity in glucose production in any tissue.

Summary: Generic membrane location inherited from the G6PC node; true but less specific than the supported ER membrane location.

Reason: The membrane claim is correct (nine transmembrane helices), and GO:0016020 is a generic membrane term rather than a root organelle term, so the "correct but general is MODIFY" rule applies. The evidence supports the specific membrane: the human protein was shown to be embedded in the ER membrane and held there by its nine transmembrane domains (PMID:15044018), UniProt records ER membrane and multi-pass with ECO:0000269 from that paper, and the gene already carries GO:0005789 by IEA and TAS. Refine to that descendant.

Proposed replacements: endoplasmic reticulum membrane

Summary: G6PC2 variants modulate fasting plasma glucose in a 6,822-participant Chinese study; G6PC2 sets the fasting blood glucose level through the islet glucose/G6P cycle.

Reason: The cited study is a population genetic association rather than a perturbation of the protein, which is weak evidence on its own, but the claim it makes is one of the best-supported things about this gene in human. Coding variants that reduce G6PC2 protein abundance or catalytic activity are associated with fasting blood glucose, and most nonsynonymous G6PC2 variants tested reduce expression or activity (PMID:34954144); promoter variants tie the fasting-glucose-raising allele to higher G6PC2 promoter activity, in the direction the knockout predicts (PMID:20622168); and UniProt records the locus as the fasting plasma glucose quantitative trait locus FGQTL1. The mechanism is the gene's own activity: by cycling G6P back to glucose in islet cells, G6PC2 sets the glucose concentration at which insulin and glucagon output change, and hence the level at which blood glucose settles (PMID:23274894, PMID:39742505). GO:0042593 has exactly one child, intracellular glucose homeostasis (checked in OLS), which does not apply, so this term is as precise as warranted.

Summary: Correct, but less specific than the evidence supports: G6PC2 negatively regulates insulin secretion in response to glucose, and the human association carries that direction.

Reason: The direction is in the cited study itself: the G6PC2 rs16856187 allele that raises fasting glucose is associated with lower first-phase insulin secretion (beta -0.0245, SE 0.0096, p = 0.0108) and with lower second-phase secretion under a dominant model, so more G6PC2 goes with less glucose-stimulated insulin release. A second human cohort associates G6PC2 variants with beta-cell glucose sensitivity and insulin-secretion parameters (PMID:34128214), and the promoter work shows the fasting-glucose-raising allele raises G6PC2 expression (PMID:20622168), closing the loop. The sign and the stimulus are established mechanistically in beta cells: deletion shifts the glucose dose-response of insulin secretion leftward and raises secretion about two-fold at submaximal but not maximal glucose (PMID:23274894), and raises glycolytic and citric acid cycle flux (PMID:35176280). G6PC2 does part of the work: its own G6P hydrolysis opposes glucokinase at the entry to glycolysis, which is the step that sets the glucose sensitivity of insulin release. Refine to GO:0061179 negative regulation of insulin secretion involved in cellular response to glucose stimulus (verified in OLS; an is_a descendant of GO:0050796 via GO:0046676 and GO:0061178), the same replacement as in the mouse review.

Proposed replacements: negative regulation of insulin secretion involved in cellular response to glucose stimulus

Summary: G6PC2 is present in human alpha cells and sustains glucagon output there: knocking it down in primary human alpha cells makes glucagon secretion suppressible at lower glucose and lowers maximal secretion, and alpha-cell-specific deletion in mice lowers glucagon in vivo and ex vivo.

Reason: No existing row covers the alpha-cell arm of this gene's function, and the evidence is direct and human as well as mouse. Bahl et al. 2025 showed that the G6PC2 protein is detectable in human alpha cells by immunofluorescence, that the glucose/G6P cycle runs there (2-NBDG efflux from human alpha cells falls on G6PC2 knockdown), and that shRNA knockdown (86% mRNA reduction) in sorted primary human alpha cells aggregated into over 90% alpha-cell pseudo-islets makes glucagon secretion significantly more glucose-sensitive at 1, 3, 5 and 9 mM glucose, with unchanged total glucagon content. In mice an inducible alpha-cell-specific deletion (88% of alpha cells) shifted the glucose set point for suppression of amino-acid-stimulated glucagon secretion leftward, with no difference at zero glucose but significantly less glucagon at 0.5-5 mM, and lowered fasting and glucose-suppressed plasma glucagon with insulin unchanged; islet composition, hormone content and proliferation were unaffected, and phloridzin and insulin-tolerance experiments excluded a paracrine insulin explanation. Sign: loss of G6PC2 lowers glucagon output at a given glucose, so G6PC2 is a positive regulator of glucagon secretion, and GO:0070094 (verified in OLS; is_a GO:0070092 and GO:0090277, no children) is the correctly signed term. There is no "involved in cellular response to glucose stimulus" child for glucagon as there is for insulin, so the set-point detail is stated in this summary rather than in the term. Participation: the regulated step is alpha-cell glycolytic flux and G6PC2's own G6P hydrolysis is what lowers it, opposing glucokinase; this is the gene product doing the work, not merely being required for the process. Comparator check: cell-intrinsic regulators of alpha-cell glucagon release do carry these terms, so there is no convention against annotating an intracellular islet enzyme here (QuickGO annotation search by reference, re-run 2026-11-10; all four hits are MGI rows on mouse genes: Pask GO:0070092 IMP PMID:21181396, Kcnk16 GO:0070092 IMP PMID:29402588, Syt7 GO:0070092 IMP PMID:19171650, Ffar4 GO:0070094 IMP PMID:24742677). Glucokinase itself carries no glucagon term despite the alpha-cell GCK literature, which reads as a curation lag rather than a convention, and in any case this row rests on direct evidence rather than on pathway completeness. The same row is added to the mouse review, where the conditional knockout was done.