mPGES-2 (PTGES2 / Q9H7Z7) — biology, expression & disease dossier

Foundation: identity, biochemistry, mechanism, and the physiological controversy

mPGES-2 (PTGES2 / Q9H7Z7) is a scientifically contested enzyme. It was discovered and named as a prostaglandin E2 synthase, but whether it acts as a physiological PGE2 synthase is actively disputed: mice lacking the gene show no measurable drop in PGE2 in any tissue ¹, and UniProt records its biological function as "still under debate" ². The modern, biochemically grounded picture is that mPGES-2 is a glutathione-dependent heme protein whose behaviour is governed by whether heme is bound (see 10_heme_configurations.md). This foundation section establishes what is firmly known about its identity, fold, maturation, in-vitro catalysis, structures, baseline expression, and the controversy itself.

Evidence levels used throughout. (a) Direct PTGES2 protein evidence — experiments on the mPGES-2/PTGES2 protein or gene. (b) PGE2-pathway inference — reasoning downstream of the prostaglandin the enzyme is claimed to make. (c) Analogy to related heme/GSH proteins. Statements are labelled Predicted: or Hypothesis: where they rest on (b) or (c) rather than direct measurement. Where no direct study exists, that absence is stated explicitly rather than filled in.

1. Identity and nomenclature

mPGES-2 was first purified from the microsomal fraction of bovine heart and cloned from monkey/human cDNA; the recombinant ~33 kDa protein carries the consensus region shared by glutaredoxin and thioredoxin, and shows PGE synthase activity in vitro ⁵. (Direct PTGES2 evidence.)

Not to be confused with — isoform disambiguation

The prostaglandin-E-synthase field contains four different genes that are constantly conflated in the literature. mPGES-2 is PTGES2 / Q9H7Z7 only. Do not import claims about the following into the mPGES-2 record:

• PTGES = mPGES-1 = O14684 — a different protein and fold (MAPEG family, membrane-integral, no heme), inducible and NF-κB-driven, and the classic oncogenic PGE2 drug target. Most papers titled "mPGES inhibitor," and most NF-κB/inflammation/EMT prostaglandin-oncology work, are about mPGES-1, not mPGES-2 ⁶. Example confounder: a PDAC study of "PTGES" is mPGES-1, not our gene ⁷.
• PTGES3 = cPGES = p23 — a cytosolic, structurally unrelated prostaglandin E synthase 3 / Hsp90 co-chaperone ⁸.
• PTGS2 = COX-2 ("prostaglandin G/H synthase 2" / "prostaglandin- endoperoxide synthase 2") — an entirely different upstream enzyme that makes PGH2. A finding of "low prostaglandin-endoperoxide synthase-2 in colorectal carcinoma predicts poorer survival" is about COX-2, and is the likely origin of any mistaken "mPGES-2 is down in CRC" claim ^{9 10}.

Any source that attributes an O14684 (mPGES-1), PTGES3 (cPGES), or PTGS2 (COX-2) property to mPGES-2 is a mis-attribution and is excluded here.

2. Fold and catalytic residue

mPGES-2 adopts a thioredoxin/glutaredoxin fold within the GST superfamily and carries the redoxin-type Cys-x-x-Cys motif at Cys110-x-x-Cys113 ^{5 4}.

• Cys110 is the essential catalytic residue. Mutating Cys110→Ser (or the Cys110/Cys113 double mutant) abolishes PGE synthase activity, whereas the single Cys113→Ser mutation does not affect activity — so Cys110, not Cys113, plays the catalytic role, analogous to the reactive cysteine of thioredoxin/ glutaredoxin ⁴. (Direct PTGES2 evidence.)
• Cys110 is NOT the heme axial ligand. In the heme-bound form the iron is coordinated by the thiolate of bound glutathione, not by a protein cysteine — a point developed in 10_heme_configurations.md ¹¹.
• The enzyme is activated by a broad range of sulfhydryl reagents (dithiothreitol, dihydrolipoic acid, glutathione, β-mercaptoethanol), which distinguishes it from mPGES-1 and cPGES that require GSH specifically ^{5 4 12}.

3. Maturation and quaternary structure

mPGES-2 is synthesised as a Golgi-membrane-anchored precursor (single-pass membrane protein) and is then processed by proteolytic removal of its N-terminal hydrophobic domain to yield the predominant mature, soluble cytosolic enzyme, which is enriched in the perinuclear region ^{2 13}.

• The mature recombinant protein runs at ~33 kDa and is catalytically active; intact and N-terminally truncated forms have similar activity ^{5 11}.
• The functional unit is a homodimer ¹⁴.
• Because the precursor is membrane-associated, mPGES-2 is often described as "microsomal/membrane-associated," but the mature, physiologically relevant form is cytosolic ^{2 13}. (Direct PTGES2 evidence — do not read "microsomal" as "integral membrane enzyme.")

4. In-vitro catalysis: PGH2 → PGE2 isomerase, heme not required

In vitro, the heme-free (apo) enzyme catalyses isomerisation of the cyclooxygenase product PGH2 to PGE2 ^{5 12}. For the recombinant enzyme, reported kinetics are Vmax ≈ 3.3 µmol·min⁻¹·mg⁻¹ and Km(PGH2) ≈ 28 µM ⁵. Cellular over-expression increases PGE2 formation downstream of both COX-1 and COX-2 ¹³.

• Heme is NOT required for the PGE synthase activity. The colourless heme-free enzyme is an active PGES; the heme-bound form can be converted back to an active PGES by dithiothreitol, which dissociates the bound heme in vitro ^{15 12}. Claiming heme is required for PGES activity is incorrect. (Direct PTGES2 evidence; see 10_heme_configurations.md for the two-state detail.)

5. Crystal structures

The 2PBJ structure is the structural basis for the "Fe-on-GSH-thiolate" coordination: the heme is complexed with the bound glutathione, not directly ligated by a protein residue ¹¹.

6. The contested physiological-PGES status

This is the throughline of the whole dossier and must be stated as contested, not settled:

• Knockout mice: loss of mPges-2 did not measurably decrease PGE2 in any tissue or cell type of healthy mice; the authors conclude the data "does not substantiate the contention that mPGES-2 is a PGE2 synthase" ¹. (Direct PTGES2 evidence.)
• UniProt: the biological function and GSH-dependent property "is still under debate"; in vivo the protein "would not participate in PGE2 synthesis" but would degrade PGH2 ².
• Dual-function model: the GSH–heme (holo) form catalyses degradation of PGH2 to 12(S)-hydroxy-heptadecatrienoic acid (12-HHT) + malondialdehyde (MDA) rather than isomerising it to PGE2 — described as "the first example of a dual-function enzyme" (isomerase when apo, lyase when holo) ^{11 12}.

Separation of evidence levels for the controversy: the in-vitro isomerase activity (apo) and the in-vitro lyase/degradation activity (holo) are both directly measured ^{5 11 12 15}. What is disputed is which activity operates in vivo — and the in-vivo KO evidence argues against a physiological PGE2-synthase role ¹. Much of the strongest recent in-vivo mechanism (heme-sensing nuclear receptors, ferroptosis) comes from a single research group and is not yet independently replicated — flagged where cited in 10_heme_configurations.md.

7. Baseline expression

mPGES-2 is broadly and constitutively expressed, in contrast to the inducible mPGES-1 ^{5 2}:

• The Human Protein Atlas classifies PTGES2 RNA as "tissue enhanced" and "detected in all" tissues, with the highest transcript level in skeletal muscle (nTPM ≈ 131) ³.
• The original characterisation found highest mRNA in heart and brain (and it was first purified from bovine heart microsomes); UniProt records wide expression across heart, brain regions, kidney, lymph nodes and other tissues ^{5 2}.
• HPA immunohistochemistry shows broad cytoplasmic staining rather than tumour-restricted expression; tissue- and tumour-specific expression detail is covered in the expression section (20_expression.md).

So the baseline is a constitutive, house-keeping-like distribution (highest in heart / skeletal muscle, present everywhere), which is consistent with a redox/heme-handling role rather than a tightly regulated inflammatory PGE2 source.

8. Foundation summary

• Direct, well-established: identity as PTGES2/Q9H7Z7; GST-superfamily thioredoxin/glutaredoxin fold; Cys110-x-x-Cys113 with Cys110 catalytic; Golgi-precursor → cleaved → mature cytosolic ~33 kDa homodimer; in-vitro PGH2→PGE2 isomerase; heme NOT required for PGES activity; apo structure 1Z9H and GSH–heme structure 2PBJ (Fe on GSH thiolate); broad constitutive expression.
• Contested / not settled: whether mPGES-2 is a physiological PGE2 synthase (KO PGE2 unchanged; UniProt "under debate"; holo form degrades PGH2 to 12-HHT + MDA).
• Absence stated: there is no validated selective mPGES-2 inhibitor, and the strongest in-vivo disease mechanism rests on a single, not-yet-replicated group (developed in later sections).

The functional consequence of the apo↔holo switch — the true engine of mPGES-2 biology — is the subject of 10_heme_configurations.md.

Heme-bound vs heme-free: two configurations, opposite consequences

The central mechanistic fact about mPGES-2 (PTGES2 / Q9H7Z7) is that it exists in two functionally opposite configurations, and which one dominates depends on heme occupancy. The apo (heme-free) enzyme is the classical — and disputed — PGH2→PGE2 isomerase; the holo (GSH–heme-bound) enzyme is a PGH2-degrading dual-function enzyme and a heme-sequestering redox/nuclear-receptor modulator. This section owns the heme-form biochemistry; the canonical ferroptosis axis is owned by 50_cell_death.md and is cross-referenced, not re-derived, here.

Evidence levels. (a) Direct PTGES2 protein evidence; (b) PGE2-pathway / downstream inference; (c) analogy. Predicted/inferred statements are labelled Predicted: / Hypothesis:. Where no direct study exists, that is stated.

0. The two configurations at a glance

Both forms and their opposing consequences are documented below.

1. APO (heme-free) — the classical, disputed PGH2 → PGE2 isomerase

• The heme-free enzyme is the catalytically active prostaglandin E synthase in vitro: it isomerises PGH2 to PGE2, activated by thiol reagents (DTT, DHLA, GSH, β-mercaptoethanol) ^{5 12}. (Direct PTGES2 evidence.)
• Heme is not required for this PGES activity. When the heme-bound enzyme is purified in the presence of dithiothreitol, only PGE2 is produced and essentially no HHT is detected; native mPGES-2 "enzymatically catalyzes only the conversion of PGH2 to PGE2 … and heme is not involved in this reaction" ¹². (Direct PTGES2 evidence.)
• DTT dissociates the bound heme to regenerate active PGES. Oxidation by H2O2 or reduction by GSH or 2-mercaptoethanol does not dissociate the heme; only DTT — described explicitly as an artificial reducing compound — does, producing active PGE2 synthase in vitro ¹⁵. This is the operational definition of the apo form. (Direct PTGES2 evidence.)
• Whether the apo form ever dominates in vivo is disputed. The same study measured a heme dissociation constant of ~2.9 nM and found substantial heme-bound enzyme in cell extracts, arguing that the protein is largely heme-bound in vivo ¹⁵. Combined with the knockout data (PGE2 unchanged, see 00_foundation.md ¹), the physiological reality of the apo-PGES function remains contested.

Hypothesis: the apo form functions as a local PGH2→PGE2 isomerase only where free-heme availability is low and reducing conditions strip the heme; this is a mechanistic inference from the in-vitro heme-loading data ^{15 12}, not a directly demonstrated in-vivo activity.

2. HOLO (GSH–heme-bound) — PGH2-degrading dual-function enzyme

• The recombinant enzyme grown with δ-aminolevulinate and Fe(III) is red and contains both glutathione and heme; grown in minimal medium it is colourless and heme-free. The red (holo) enzyme was characterised by mass, fluorescence and EPR spectroscopy and X-ray crystallography ¹¹. (Direct PTGES2 evidence.)
• Coordination geometry (PDB 2PBJ): glutathione binds in the active site via six hydrogen bonds; the heme is complexed with the bound GSH, forming an S–Fe coordination bond via the GSH thiolate, with no polar interaction between heme and the protein, and a large open pocket that could accommodate PGH2 ^{11 17}. The iron axial ligand is therefore the glutathione thiolate — not protein Cys110 (see 00_foundation.md).
• Reaction: PGH2 degradation, not isomerisation. The GSH–heme-bound enzyme (mPGES-2h) catalyses formation of 12(S)-hydroxy-5(Z),8(E),10(E)- heptadecatrienoic acid (12-HHT) and malondialdehyde (MDA) from PGH2, but not PGE2 ¹¹. Reported kinetics for the degradation: KM ≈ 56 µM, kcat ≈ 63 s⁻¹, kcat/KM ≈ 1.1 × 10⁶ M⁻¹s⁻¹ — i.e. significant catalytic efficiency ¹¹. This is the basis for calling mPGES-2 "the first example of a dual-function enzyme" — an isomerase when apo, a lyase when holo — which cannot be placed cleanly into a single EC class ¹¹.
• The two activities are independently thermolabile. PGE synthase activity is completely lost at 50 °C (5 min), whereas HHT-synthase (lyase) activity survives even 100 °C (5 min) — direct evidence that they are distinct activities of distinct forms, and that free heme itself can drive HHT formation ^{12 2}.
• Tissue context. Because the holo form dominates where heme is abundant, the authors propose mPGES-2 in heme-rich liver is mostly mPGES-2h and may act like a cytochrome-P450-type degradation enzyme ¹¹.

3. The switch: heme loading ↔ DTT dissociation

The apo↔holo interconversion is the control point of the whole system:

• Loading (apo → holo): heme binds mPGES-2 only in the presence of glutathione; heme content rises with extracellular Fe(III), and heme-free enzyme is converted to the heme-bound form simply by mixing with liver extract — so the protein readily acquires heme in mammalian cells ^{15 11}. Reported heme affinities are high (Kd ≈ 0.53 µM in the first crystallographic study ¹¹; ≈ 2.9 nM re-determined in vivo/in vitro ¹⁵).
• Dissociation (holo → apo): only DTT strips the heme (not H2O2, GSH or 2-mercaptoethanol), regenerating active PGES in vitro ^{15 12}.

So the same polypeptide toggles between a (disputed) PGE2-making isomerase and a PGE2-forgoing, heme-carrying degradation enzyme depending on heme occupancy — and, critically, the holo form removes free heme from the cellular pool.

4. Disease-relevant role of the HOLO form: heme sequestration

The modern, in-vivo relevance of mPGES-2 is that its holo form sequesters heme, so losing mPGES-2 liberates heme to act on heme-sensing effectors. This is the "PGE2-independent" biology and is the reason the enzyme matters despite the PGES controversy. Provenance caveat: the studies in §4.1–§4.2 come from a single research group (Xuzhou Medical University / Nanjing; Zhong, Sun, Jia and colleagues) and are not yet independently replicated; they are direct PTGES2 gene-manipulation studies but in kidney and liver, not tumours.

4.1 Heme-sensing nuclear receptors — Rev-Erbα / NR1D1

• NASH / fatty liver. mPGES-2 "was originally discovered as a PGE2 synthase; however, it does not produce PGE2 in the liver." Whole-body or hepatocyte-specific mPGES-2 deficiency reduced hepatic lipid accumulation and liver injury, and the protection was dependent on the heme receptor NR1D1 (Rev-Erbα) — "heme regulated the increased NR1D1 activity mediated by mPGES-2 deficiency," not PGE2 ¹⁸. Mechanistically, removing mPGES-2 frees heme, which activates NR1D1 (down-regulating CYP4A14, up-regulating ACOT4). (Direct PTGES2 evidence; single-group; liver.)
• Diabetic kidney disease. Genetic or pharmacologic blockade of mPGES-2 attenuated renal lipotoxicity and diabetic kidney disease by targeting Rev-Erbα/FABP5 signalling ²⁰. (Direct PTGES2 evidence; single-group; kidney.)

These are the clearest demonstrations that the holo (heme-bound) configuration is the disease-relevant one: the enzyme's job is to hold heme away from Rev-Erbα/NR1D1, and blocking it re-mobilises heme onto those receptors.

4.2 Ferroptosis / redox axis (cross-reference — owned by 50_cell_death.md)

• In acute kidney injury, mPGES-2 (which "can metabolize PGH2 to malondialdehyde by forming a complex with heme") promotes ferroptosis; genetic or pharmacologic blockade (research tool SZ0232) inhibited ferroptosis via the heme-dependent regulation of the p53/SLC7A11/GPX4 axis ¹⁹. (Direct PTGES2 evidence; single-group; kidney, not tumour.)
• The heme dependence is the link to this section: it is the holo form and its handling of heme (and the MDA it generates from PGH2) that couples mPGES-2 to lipid-peroxide/ferroptosis chemistry — consistent with the GSH–heme biochemistry above ^{11 19}. The full ferroptosis mechanism, the SZ0232 selectivity caveat, and cell-death coverage are in 50_cell_death.md and are not re-derived here.
• Predicted: the same heme-form/ferroptosis axis operates in tumour cells. This is a plausible extrapolation from the heme + GSH biochemistry and the renal data, but no direct PTGES2 tumour-ferroptosis study was found — the tumour relevance is inference (level b), not direct evidence. See 50_cell_death.md.

5. Summary — both directions, and what is and isn't known

• Apo (heme-free): classical (disputed) PGH2→PGE2 isomerase; heme not required; DTT dissociates heme to restore PGES in vitro ^{5 12 15}.
• Holo (GSH–heme): dual-function lyase — degrades PGH2 to 12-HHT + MDA; Fe coordinated by the GSH thiolate (PDB 2PBJ); the disease-relevant, heme-sequestering form controlling Rev-Erbα/NR1D1 and the ferroptosis/redox axis ^{11 12 17 18 19 20}.
• Interconversion: GSH-dependent heme loading vs DTT-driven heme dissociation sets which activity dominates ^{15 11}.
• Absence / caveats stated: the in-vivo disease mechanism is single-group and not yet independently replicated; it is shown in kidney/liver, not tumours; tumour relevance of the heme-form axis is Predicted, not demonstrated; there is no validated selective mPGES-2 inhibitor (SZ0232 is a research tool of unestablished selectivity — see 60_inhibition_effects.md).

The take-home: heme-bound and heme-free mPGES-2 are not functionally equivalent — they catalyse opposite fates for PGH2 and have opposite consequences for cellular heme availability, and it is the heme-bound form that carries the credible modern, PGE2-independent biology.

Expression in healthy tissue vs. tumors

Evidence density: MODERATE, tissue-mixed. There is no dedicated PTGES2 pan-cancer expression study. The direct tumor evidence rests on one functional HCC paper with TCGA + tissue-microarray validation ²¹, one colorectal immunohistochemistry cohort ²², a handful of corroborating cell/pathway reports, and the Human Protein Atlas antibody/prognostic survey ³. Read the picture below as cancer-type-specific and sometimes contradictory — high mPGES-2 is unfavorable in some tumors and favorable in others. Do not read it as a uniform "up in cancer" or "down in cancer" signal.

Baseline expression in healthy tissue

mPGES-2/PTGES2 is a constitutive, broadly expressed enzyme, not an inducible one. UniProt records it as "widely expressed," highest in heart (apex, septum, atria and ventricles) and also present in skeletal muscle, several brain regions (cerebellum; occipital, frontal and parietal lobes), kidney, lymph nodes and trachea, but absent from aorta, thymus and lung ². The Human Protein Atlas classifies it as detected in all tissues with low tissue specificity ³. Its identity, maturation and the disputed-PGE2-synthase framing are established in the Foundation section (see 00_foundation.md); this section only contrasts that constitutive baseline against tumors.

Per-cancer direction and prognosis

Hepatocellular carcinoma (HCC / TCGA "LIHC") — UP, unfavorable [direct PTGES2 evidence]. In HCC, PTGES2 is associated with poor patient outcome in both TCGA data and a tissue microarray, and it acts functionally: overexpression promotes, and knockdown inhibits, HCC cell-line proliferation and migration ²¹. Independently, the Human Protein Atlas lists PTGES2 as a potential prognostic, unfavorable marker in liver cancer, high expression associating with worse survival (p<0.001, HPA-sourced) ³. This is the single best-supported tumor context for mPGES-2.

Colorectal cancer (CRC) — UP, unfavorable [direct PTGES2 evidence]. In 155 surgical resections, mPGES-2 immunoreactivity was significantly more pronounced in cancer cells than in adjacent normal epithelium in 36% of cases, and mPGES-2 (like mPGES-1) correlated with significantly worse prognosis in stage I–III patients ²². This is corroborated by an independent report that mPGES-1 and mPGES-2 are overexpressed in CRC (Caco-2 / HCT-116 cell context) ²³, and by UniProt's tissue-expression annotation, which explicitly notes PTGES2 is "overexpressed in colorectal cancer" ². A metabolomic CRC study also reported that "PGE2 synthases" are upregulated and correlate with CD68+ macrophage density ²⁴ — but that study measured PGE2 synthases collectively and did not resolve the PTGES2 isoform, so it is cited only as pathway-level corroboration, not as isoform-specific PTGES2 evidence. Note also that the Human Protein Atlas does not list CRC among PTGES2's prognostic cancers ³; the CRC prognostic signal comes from Seo's protein-level cohort, not from HPA.

Renal cancer (TCGA "KIRC" and "KIRP") — high expression is FAVORABLE [HPA-sourced]. In both clear-cell (KIRC) and papillary (KIRP) renal cell carcinoma, the Human Protein Atlas classifies PTGES2 as a potential prognostic, favorable marker — high expression associates with better survival (p<0.001, HPA-sourced) ³. This is the opposite prognostic direction from HCC and CRC and is the clearest illustration of the tissue-mixed reality: mPGES-2 is not uniformly pro-tumor.

Urothelial / bladder cancer (TCGA "BLCA") — unfavorable [HPA-sourced]. In urothelial cancer, the Human Protein Atlas classifies PTGES2 as a potential prognostic, unfavorable marker; high expression associates with worse survival (p<0.001, HPA-sourced) ³.

Basal cell carcinoma (BCC) — genetic-risk biomarker, not a tumor-vs-normal expression finding [direct PTGES2 evidence, different modality]. A proteome-wide Mendelian-randomization, colocalization and MR-PheWAS study identified PTGES2 (with RNASET2) as a plasma-protein biomarker and candidate therapeutic target causally associated with BCC risk, with strong colocalization (posterior probability PP.H4 > 0.92), and found BCC to be the single most significant phenotype linked to PTGES2 across 2,408 phenotypes ²⁵. This is a genetic/plasma-protein causal-risk signal, not a measurement of tumor-versus-normal tissue expression — the distinction matters and is stated here deliberately.

Broad immunohistochemistry across cancers [HPA-sourced]. The Human Protein Atlas reports granular cytoplasmic staining of varying intensity (with occasional nucleolar positivity) in all cancers examined ³. mPGES-2 is therefore broadly present across tumor types, not tumor-restricted; presence alone is not a prognostic or driver signal, which is why the per-cancer prognostic directions above (not mere detection) carry the weight.

Cancers lacking PTGES2-specific expression data (explicit absence)

There is no PTGES2-specific expression or prognostic study for gastric cancer, glioma, lung cancer, prostate cancer, or leukemia. Any statement about mPGES-2 in these tumors would be unsupported. This absence is a genuine finding, not an omission: outside HCC, CRC, renal, bladder and the BCC genetic-risk signal, PTGES2 tumor biology is essentially uncharacterized.

Do not confuse: "mPGES-2 down in colorectal cancer" is unsupported / refuted

A claim that mPGES-2 is down-regulated in CRC and that low expression predicts poor survival is not supported for PTGES2 and is most likely a gene-identity confusion. The "low expression → poorer survival in CRC" finding belongs to PTGS2 (cyclooxygenase-2 / COX-2), a different gene: that study's own text states "prostaglandin-endoperoxide synthase-2 (ptgs2), otherwise called Cyclooxygenase 2" and reports low ptgs2 transcripts associate with poorer survival ¹⁰ — this is COX-2, not mPGES-2, and is cited here only as a do-not-confuse marker. The direct PTGES2 evidence points the other way: mPGES-2 is overexpressed in CRC and its high expression correlates with worse prognosis ^{22 23 2}. (See the identity guards in 00_foundation.md: PTGES2 = mPGES-2 = Q9H7Z7 is distinct from PTGS2/COX-2, from PTGES/mPGES-1 = O14684, and from PTGES3/cPGES.)

Reading limits

Direct tumor-expression evidence for PTGES2 exists for HCC and CRC (protein/functional level) plus HPA prognostic directions (LIHC, KIRC, KIRP, BLCA) and one BCC genetic-risk signal. There is no broad TCGA validation of PTGES2 across cancers beyond the HCC analysis and the HPA survey — statements should not imply pan-cancer transcriptomic validation that has not been published.

Co-expression with oncogenes, oncogenic-metabolism genes, and tumor promoters

Evidence density: THIN / largely ABSENT. This is the honest headline: with one exception, no study reports PTGES2 co-expression or correlation with classic oncogenes or oncogenic pathways. The one measured transcriptional relationship is a lipogenic regulator (SREBP2) acting on PTGES2 in HCC — not a co-expression correlation with an oncogene. Everything else in this area is unstudied, and it is stated as such rather than filled in by analogy.

Explicit absence: no reported oncogene co-expression

No published study reports PTGES2 (mPGES-2 / Q9H7Z7) co-expression, correlation, or coordinated regulation with any of the classic oncogenes or oncogenic programs commonly checked in tumor transcriptomics:

• c-MYC — no PTGES2 co-expression reported.
• KRAS (or other RAS-family) — no PTGES2 co-expression reported.
• CTNNB1 / β-catenin and Wnt target genes — no PTGES2 co-expression reported.
• PTGS2 / COX-2 — no PTGES2↔COX-2 co-expression correlation reported for PTGES2 specifically. (COX-2 is the upstream enzyme that produces PGH2, the mPGES-2 substrate, so a pathway relationship exists biochemically; but a measured co-expression correlation attributable to PTGES2 has not been published, and COX-2 must not be conflated with mPGES-2 — see the identity guards in 00_foundation.md.)
• HIF-1α / hypoxia target genes — no PTGES2 co-expression reported. The only signal linking PTGES2 to HIF is a non-human correlative panel (feline oral SCC) discussed under angiogenesis in 40_tumor_phenotype.md, and it does not establish co-expression in human tumors.

This is a real evidence gap, not an oversight. A proper PTGES2 co-expression analysis (e.g. TCGA correlation against MYC/KRAS/CTNNB1 modules) has not been published, so no oncogene-correlation numbers are asserted here. Inventing such correlations would violate the honesty rules of this dossier.

The one measured transcriptional regulator: SREBP2 → PTGES2 in HCC [direct PTGES2 evidence]

The single directly measured transcriptional relationship is that SREBP2 (sterol-regulatory- element-binding protein 2) activates PTGES2 transcription in hepatocellular carcinoma. When cholesterol de novo synthesis is inhibited (by atorvastatin) under high-fatty-acid conditions, SREBP2 is engaged and drives up PTGES2 expression, which in turn feeds arachidonic-acid metabolism and promotes HCC proliferation and migration; knockdown of PTGES2 reverses the statin-enhanced phenotype ²¹.

Two clarifications are load-bearing here:

1. This is regulation of PTGES2, not co-expression with an oncogene. SREBP2 is a lipogenic / cholesterol-homeostasis transcription factor acting upstream of PTGES2; the finding is a measured transcriptional-activation link, not a correlation between PTGES2 and a canonical oncogene. It must not be relabeled as "PTGES2–oncogene co-expression."
2. The direction is counter-intuitive. It is inhibition of cholesterol synthesis (a context created by statins under high fatty acid) that raises SREBP2-driven PTGES2 and thereby promotes tumor progression — this is the mechanistic axis, cross-referenced under tumor metabolism in 40_tumor_phenotype.md.

Prostanoid-network hubs are PTGES3 / PTGIS, not PTGES2 [direct, from systems review]

A systems-biology analysis of prostanoid signaling across 24 solid tumors (19 prostanoid-pathway genes, including PTGES2) identified the most connected / most post-translationally regulated prostanoid proteins as PTGES3 (cPGES) and PTGIS (prostacyclin synthase) — targeted by at least six ubiquitin ligases and eight protein kinases — and modeled regulation for PTGIS/PTGIR in lung and uterine cancers ²⁶. PTGES2 was not identified as a network hub in that analysis. This reinforces the picture that, within the tumor prostanoid network, the connected and heavily co-regulated nodes are other family members, and PTGES2 is peripheral rather than a co-expression center.

What is not established (gap statement)

Beyond the SREBP2→PTGES2 regulation in HCC, there is no evidence that PTGES2 co-expresses with tumor-promoter gene programs, oncogenic metabolic modules (glycolysis/Warburg, glutaminolysis), or immune-evasion signatures. The macrophage association reported in CRC is at the level of "PGE2 synthases" collectively and does not resolve to PTGES2 (see 20_expression.md, ²⁴). Any future co-expression claim for PTGES2 would require a dedicated correlation analysis that does not yet exist.

Influence on metastasis, angiogenesis, tumor metabolism, and redox homeostasis

Evidence density: THIN for direct tumor phenotype. Only one tumor context (HCC) provides direct functional PTGES2 data, and only for proliferation and migration. Angiogenesis has no direct PTGES2 study. Tumor-metabolism evidence is confined to the HCC arachidonic-acid node, with the rest of the enzyme's metabolic biology coming from non-tumor disease models. The redox role is real and well-supported mechanistically, but the supporting experiments are in kidney injury, not cancer. Throughout this section each sub-topic is explicitly tagged as [direct PTGES2 evidence], [PGE2-pathway inference], or Predicted:, and the canonical mechanism statements are deferred to the sections that own them.

Metastasis and invasion

Direct PTGES2 evidence — HCC migration and proliferation only. In hepatocellular carcinoma cell lines, overexpression of PTGES2 promotes proliferation and migration, while knockdown inhibits both; high PTGES2 tracks with poor patient outcome ²¹. This is the only direct PTGES2 pro-tumor cellular phenotype in the literature.

Explicitly untested (absence). Beyond HCC cell migration, there is no PTGES2 evidence for: tissue-invasion (Transwell/Matrigel invasion) assays, epithelial–mesenchymal transition (EMT) marker regulation, or in vivo metastasis (experimental or spontaneous metastasis models). These are genuine gaps, not implied positives — "migration in a scratch/Transwell assay" is not the same as demonstrated invasion or metastasis, and that distinction is kept explicit here.

Do not confuse. Reports of prostaglandin-E-synthase-driven EMT / invasion (e.g. in pancreatic cancer) belong to mPGES-1 / PTGES / O14684, a different enzyme and fold, not to mPGES-2 (see identity guards in 00_foundation.md). Such findings must not be imported as PTGES2 metastasis evidence.

Angiogenesis

No direct PTGES2 angiogenesis study found (absence). There is no study in which PTGES2 was manipulated (knockdown/overexpression/inhibitor) and a tumor-angiogenesis readout (microvessel density, endothelial tube formation, VEGF secretion driven by PTGES2) was measured. This is an explicit absence.

Correlative, non-human [PGE2-pathway, animal]. A feline oral squamous-cell-carcinoma study profiled PGE2-synthase genes (PTGES1–3, so PTGES2 was in the panel), PGE2 receptors (EP1–4), HIF1A and VEGFA in cat tumor cell lines. Of the synthases, PTGES1 and PTGES3 (with EP4 and VEGFA) were serum-inducible, and exogenous PGE2 stimulated HIF1A and CD147; PTGES2 was part of the panel but was not the highlighted regulated gene, and the study is correlative and in cat (feline) cells ²⁷. It therefore supports at most a pathway-adjacent association, not a PTGES2→angiogenesis mechanism.

Predicted: the generic route PGE2 → EP receptors (mainly EP4/EP2) → VEGF / pro-angiogenic signaling is a downstream property of PGE2 in tumors — the same feline OSCC study shows exogenous PGE2 inducing HIF1A (and the pathway co-profiles VEGFA), which is the nearest real evidence this prediction is grounded in ²⁷ — so if mPGES-2 produced PGE2 in a given tumor it could in principle feed angiogenesis by that route. This is PGE2-pathway inference, not a PTGES2 finding, and it is doubly weak for mPGES-2 specifically because the enzyme's PGE2-productive role is contested — the apo (heme-free) form isomerizes PGH2→PGE2 in vitro, but the GSH-heme (holo) form instead degrades PGH2 (see 10_heme_configurations.md). Labelled Predicted and grounded only in general prostaglandin biology plus that correlative animal signal, not in any mPGES-2 angiogenesis experiment.

Tumor metabolism

Direct PTGES2 evidence — HCC arachidonic-acid node under SREBP2 / cholesterol rewiring. In HCC, PTGES2 is a key enzyme in arachidonic-acid (AA) metabolism that becomes a driver of progression when cholesterol de novo synthesis is inhibited under high-fatty-acid conditions: SREBP2 upregulates PTGES2, AA metabolism is the most significantly changed pathway, and the PTGES2-dependent metabolic rewiring promotes HCC proliferation/migration ²¹. This places mPGES-2 inside the lipid-metabolic circuitry of the tumor rather than in a glycolytic/Warburg axis (see the SREBP2 regulation in 30_oncogenic_coexpression.md).

Non-tumor lipid-metabolic roles (context, not cancer). The enzyme's broader lipid-metabolism engagement is documented outside cancer and is included for mechanistic context, clearly labelled non-tumor: in NASH/steatohepatitis, mPGES-2 deficiency reduces hepatic lipid accumulation via heme-dependent NR1D1 signaling (and explicitly not via PGE2 — "it does not produce PGE2 in the liver") ¹⁸; in diabetic kidney disease, mPGES-2 blockade attenuates renal lipotoxicity by competing with Rev-Erbα for heme to regulate FABP5 and lipid metabolism ²⁰. These establish that mPGES-2 modulates lipid handling through heme-sensing nuclear receptors, but they are kidney/liver disease models, not tumors.

Absence. There is no direct link between PTGES2 and glycolysis / Warburg metabolism, glutaminolysis, or OXPHOS in tumors. The metabolic story for PTGES2 in cancer is, at present, the HCC arachidonic-acid/lipid node only.

Redox homeostasis

Direct mechanism (non-tumor) — states the effect; canonical derivation deferred. The GSH-heme–bound (holo) form of mPGES-2 is redox-active: complexed with heme it metabolizes PGH2 to malondialdehyde (MDA), a lipid-peroxidation product, and mPGES-2 governs a heme-dependent p53 / SLC7A11 / GPX4 axis, such that loss or blockade of mPGES-2 suppresses ferroptosis ¹⁹. The enzyme also acts as a heme sink, sequestering heme and thereby limiting its availability to heme-sensing partners ¹⁸. The MDA-forming PGH2-degradation chemistry of the GSH-heme complex is the biochemical basis of this redox role ¹¹.

• The canonical statement of the p53/SLC7A11/GPX4 ferroptosis axis is owned by 50_cell_death.md (T4); the heme-bound vs heme-free biochemistry is owned by 10_heme_configurations.md (T8). This section states the redox effect in a tumor-phenotype frame and links to those sections — it does not re-derive the mechanism.

Tumor relevance is Predicted, not demonstrated. Critically, the p53/SLC7A11/GPX4 ferroptosis / redox axis and the MDA/heme-sequestration data come from acute kidney injury and metabolic-liver models, not cancer ^{19 18}. A tumor context for the mPGES-2 redox axis is Predicted: biologically plausible — a constitutive GSH-dependent heme enzyme sitting on the GPX4/SLC7A11 ferroptosis node is exactly the kind of factor expected to tune tumor redox and ferroptosis sensitivity — but it has not been tested in tumor cells. It is labelled a hypothesis and grounded in the demonstrated (non-tumor) heme/GSH redox biology, not asserted as a cancer finding. The consequences for tumor cell-death susceptibility are developed in 50_cell_death.md.

Effect on susceptibility to cell-death pathways (apoptosis, ferroptosis, autophagy, and others)

This section is the canonical statement of the heme-dependent p53/SLC7A11/GPX4 ferroptosis axis; other sections (tumor phenotype, inhibition effects, metabolism/transcription factors, heme configurations) cross-reference it rather than restate it. Read the evidence-level tags literally: the one death mode with direct mPGES-2 loss-of-function evidence is ferroptosis, and even that evidence is from acute kidney injury, not cancer. Everything else is thin, indirect, or absent, and is labelled as such.

A structural note that governs this whole section: mPGES-2 is a GSH-dependent heme protein whose heme-bound (holo) form degrades PGH₂ to 12(S)-HHT + malondialdehyde (MDA) rather than isomerising it to PGE₂ ^{11 15 17 2}. Its cell-death activity tracks this heme/redox chemistry (heme, MDA, GSH), not its contested PGE₂-synthase role — consistent with Ptges2-knockout mice showing unchanged PGE₂ in every tissue ¹.

Ferroptosis — DIRECT evidence (mPGES-2 promotes it; blockade inhibits it) — renal, not tumor

Evidence level: DIRECT PTGES2 loss-of-function, in acute kidney injury (AKI). In cisplatin- and renal ischemia/reperfusion–induced AKI, mPGES-2 promotes ferroptosis, and both genetic (global and tubule-specific Ptges2 deletion; mPGES-2 knockdown in HK-2 cells) and pharmacologic (the tool inhibitor SZ0232) blockade of mPGES-2 inhibits ferroptosis and reduces renal dysfunction and tubular damage ¹⁹. The mechanism is explicitly heme-dependent regulation of the p53/SLC7A11/GPX4 axis: loss of mPGES-2 activity relieves suppression of the cystine/glutamate antiporter subunit SLC7A11 and preserves GPX4, lowering lipid-peroxide accumulation and iron-dependent death ¹⁹. mPGES-2 overexpression drives the axis the other way. The direction is therefore unambiguous: more mPGES-2 → more ferroptosis; inhibiting mPGES-2 → less ferroptosis.

Mechanistically this is coherent with the enzyme's biochemistry: the holo (GSH-heme) form generates MDA and consumes reducing equivalents, tying mPGES-2 to the same lipid-peroxidation/GSH economy that governs ferroptosis ^{11 15}. This axis is the load-bearing modern biology of mPGES-2 and is referenced by the inhibition-effects and heme-configuration sections.

Predicted (tumor relevance): No study has tested the mPGES-2 → ferroptosis axis in cancer. The AKI finding makes a tumor role plausible — the p53/SLC7A11/GPX4 module is a central cancer-ferroptosis node — but this is inference, not evidence. The nearest supporting literature is PGE₂-pathway (not direct mPGES-2 loss-of-function) work showing that excess PGE₂ downstream of COX/PGE-synthases promotes ferroptosis in renal tubular cells, where mPGES-2 was noted to be upregulated alongside the response but the mechanism was probed through PGE₂ and its EP receptors, not through mPGES-2 itself ²⁸. Treat any "mPGES-2 sets tumor ferroptosis sensitivity" statement as a hypothesis grounded in ¹⁹ + ²⁸, to be tested — not as an established cancer finding.

Apoptosis and toxic cell death — THIN; ABSENT for cancer

Evidence level: no cancer PTGES2 apoptosis study found (ABSENT for cancer); one non-cancer toxic-death finding (THIN). There is no study of mPGES-2 in cancer-cell apoptosis (no reports of mPGES-2 modulating BAX/BCL-2, caspase activation, or intrinsic/extrinsic apoptosis in tumor cells). The p53 that appears in the mPGES-2 literature is engaged as part of the ferroptosis axis above, not a classical apoptotic program ¹⁹.

The one direct death-related finding outside the kidney is in drug-induced hepatocyte death, not cancer: mPGES-2-knockout mice are resistant to acetaminophen (APAP)-induced liver injury, with improved histology, lower liver-enzyme release, and fewer APAP-cysteine adducts; conversely, hepatic mPGES-2 overexpression aggravates the injury ²⁹. The protective mechanism is again redox, not PGE₂: mPGES-2 deletion suppressed MDA production and raised glutathione (GSH), and the elevated basal PGE₂ in the knockout was shown not to be the protective factor ²⁹. This is a toxic/oxidative death context that reinforces the heme/MDA/GSH theme, but it is a single hepatotoxicity model and must not be generalised to tumor apoptosis.

Autophagy, necroptosis, and pyroptosis — ABSENT (no direct PTGES2 evidence)

Evidence level: ABSENT. A targeted search of the PTGES2/mPGES-2 literature returned no direct study linking mPGES-2 to any of the following, in cancer or elsewhere:

• Autophagy — no study reports mPGES-2 regulating autophagic flux, LC3/ATG machinery, or autophagy-dependent death. No direct evidence.
• Necroptosis — no study reports mPGES-2 acting through RIPK1/RIPK3/MLKL or otherwise modulating necroptosis. No direct evidence.
• Pyroptosis — no study reports mPGES-2 acting through the inflammasome/caspase-1/gasdermin axis. No direct evidence.

Any mechanistic claim about mPGES-2 and these three death modes would be fabrication; we state the absence explicitly rather than infer a link from the enzyme's redox biology.

Adjacent context — growth/proliferation (not a death pathway)

For completeness, and cross-referenced (not owned) here: in polycystic kidney disease, mPGES-2 blockade (Ptges2 knockout and SZ0232) impedes renal cyst growth by curbing proliferation via an MDA- and PGE₂–EP4–β-catenin/STAT3–c-Myc route ³⁰. This is an anti-proliferative (growth) effect, not a cell-death-pathway effect, and is included only so it is not mistaken for death-axis evidence.

Effect of inhibition/knockdown on drug & chemo sensitivity and metabolic markers (iron, pH, ROS, cell death)

This section answers "what happens when you inhibit or knock down mPGES-2?" The honest headline is twofold: (1) there is no validated selective mPGES-2 inhibitor, so almost all direct evidence comes from genetic tools, and (2) the one drug-sensitivity result is about protecting host tissue from chemotherapy toxicity, not about sensitising a tumor to chemotherapy. The metabolic markers that respond (iron, ROS, lipid peroxide, cell death) are the ferroptosis-axis readouts owned by the cell-death section; intracellular pH has no evidence and is offered only as a labelled prediction.

The inhibitor landscape — no validated selective mPGES-2 inhibitor

Evidence level: DIRECT (on tooling). There is no validated selective small-molecule inhibitor of mPGES-2. This matters because the "mPGES inhibitor" drug class in the literature and clinic targets a different enzyme, mPGES-1 (PTGES / O14684) — an inducible, MAPEG-fold, non-heme PGE₂ synthase that is the classic anti-inflammatory drug target, not our constitutive heme/GSH enzyme mPGES-2 (PTGES2 / Q9H7Z7). Do not conflate them. Examples of the mPGES-1 programs (cited here as identity context, not as mPGES-2 evidence):

• PF-9184 — a preclinical selective mPGES-1 inhibitor (IC₅₀ ≈ 16.5 nM; >6500-fold over COX-1/COX-2), used to distinguish mPGES-1 inhibition from COX-2 inhibition ³¹.
• Vipoglanstat — a clinical mPGES-1 inhibitor evaluated in a phase-2 trial in systemic-sclerosis Raynaud's ³².

For mPGES-2 itself, the pharmacologic tool used in the primary literature is SZ0232, applied as an mPGES-2 inhibitor in AKI and polycystic-kidney-disease models ^{19 30}. Its selectivity has not been independently established — it is a research probe, and this dossier does not claim it is a validated selective mPGES-2 inhibitor. Because of this, the field's direct loss-of-function evidence rests mainly on genetic tools: global and tubule-specific Ptges2 knockouts plus mPGES-2 knockdown/overexpression in HK-2 cells ¹⁹, Ptges2⁻ᐟ⁻ crosses ³⁰, tissue-specific blockade in diabetic kidney disease ²⁰, and the original whole-body Ptges2-null line ¹. A ChEMBL target record exists for PTGES2 (CHEMBL4411) but reflects no bona fide medicinal-chemistry program ³³.

Drug / chemo sensitivity — toxicity-modulation, not tumor chemosensitization

Evidence level: DIRECT for chemo-toxicity protection (renal); ABSENT for tumor chemosensitization. The single drug-sensitivity result is that mPGES-2 blockade protects against cisplatin-induced acute kidney injury by suppressing ferroptosis: both genetic deletion and SZ0232 reduced cisplatin-driven (and ischemia/reperfusion-driven) renal dysfunction and tubular death through the heme-dependent p53/SLC7A11/GPX4 ferroptosis axis ¹⁹. Read this correctly — it means mPGES-2 inhibition lowers the host-tissue toxicity of a chemotherapeutic (cisplatin nephrotoxicity). It is a chemo-toxicity–modulation finding, framed around protecting normal kidney, and says nothing about making a tumor more or less responsive to chemotherapy.

There is no tumor chemosensitivity evidence for mPGES-2 (ABSENT). No study reports that inhibiting or knocking down mPGES-2 sensitises (or protects) cancer cells against cisplatin, doxorubicin, 5-FU, or any other chemotherapeutic. A tumor chemosensitisation role is therefore unestablished; any such statement would be inference from the ferroptosis biology, not data.

Metabolic markers under mPGES-2 blockade — iron, ROS, lipid peroxide, cell death

Evidence level: DIRECT (renal), via the ferroptosis axis. The metabolic markers that move on mPGES-2 blockade are exactly the ferroptosis readouts: inhibition/knockout lowers lipid-peroxide (MDA) accumulation, reduces iron (Fe²⁺)-dependent death, restores GSH and GPX4, and decreases reactive-oxygen/lipid-ROS–driven cell death ¹⁹; the parallel APAP-liver study shows the same directional signature — mPGES-2 loss suppresses MDA and raises GSH ²⁹. So the marker panel (iron ↓, ROS/lipid-peroxide ↓, cell death ↓, GSH ↑) all points one way when mPGES-2 is inhibited, consistent with removing a pro-oxidant heme/MDA-generating activity. These are cross-referenced to the cell-death section, which owns the axis.

Intracellular pH — no evidence; PREDICTED only. Predicted: no study has measured an effect of mPGES-2 inhibition on intracellular pH, so this is offered purely as a mechanistic hypothesis, not a finding. A pH effect is conceivable because mPGES-2 is a GSH-dependent heme enzyme whose activity consumes reducing equivalents and generates aldehyde/lipid-peroxidation products, and severe redox/ferroptotic stress can perturb membrane integrity and ion handling ^{5 15 11}. This remains a hypothesis to test (e.g. ratiometric pH imaging under mPGES-2 knockdown); it must not be presented as an observed effect.

Toxicity of inhibition — animal evidence and a reasoned prediction

No toxicology study of any mPGES-2 inhibitor exists (there is no validated selective one — see the inhibition-effects section). What we do have is strong genetic evidence that removing mPGES-2 is well tolerated in vivo, plus the enzyme's biochemistry, from which we give an explicitly-labelled prediction of what an inhibitor's on-target toxicity profile is likely to be. The prediction is not data and is marked as such.

On-target genetic loss is well tolerated — DIRECT evidence

Evidence level: DIRECT. Homozygous Ptges2-knockout mice are viable, fertile, and phenotypically normal, and — critically — PGE₂ levels are unchanged in every tissue and cell type examined from healthy animals; the authors concluded that mPGES-2 is not essential for in vivo PGE₂ biosynthesis ¹. This is the cleanest available toxicity signal: complete, lifelong, whole-body loss of mPGES-2 function produces no overt phenotype and no PGE₂ deficit, which argues that on-target inhibition of mPGES-2 should be well tolerated at the level of normal physiology.

This tolerability is reinforced by the disease-model literature, where global and tissue-specific Ptges2 deletions (and knockdown) have been carried into adulthood and challenged without reports of baseline toxicity: tubule-specific and global knockouts in acute kidney injury ¹⁹, Ptges2⁻ᐟ⁻ crosses in polycystic kidney disease ³⁰, and knockout lines in NASH and diabetic kidney disease ^{18 20}. Across these, loss of mPGES-2 was protective in the disease context rather than harmful.

No inhibitor toxicity data exist — ABSENT

Evidence level: ABSENT. Because there is no validated selective mPGES-2 inhibitor, there is no dedicated animal or clinical toxicology study of mPGES-2 inhibition. The tool compound SZ0232 has been used in short-term rodent efficacy experiments ^{19 30} without a reported systemic-toxicity readout, but no toxicokinetic, safety-pharmacology, or repeat-dose tox study has been performed, and its selectivity is unestablished (see inhibition-effects). Any statement that an mPGES-2 inhibitor "was safe/toxic in animals" would be fabrication.

Predicted: on-target inhibitor toxicity — reasoned prediction (NOT observed data)

Predicted (hypothesis, grounded in cited biochemistry — not measured): The most likely on-target toxicity concerns for a future selective mPGES-2 inhibitor arise from the enzyme's heme/GSH chemistry and its broad constitutive expression, but the knockout evidence argues these are probably mild.

• Heme handling. mPGES-2's holo form binds heme (Fe coordinated by a GSH thiolate) and sequesters it; inhibiting or removing mPGES-2 liberates heme and activates heme-sensing nuclear receptors (e.g. Rev-Erbα/NR1D1) ^{11 18 17}. Predicted: a drug that shifts intracellular free-heme availability could perturb heme-dependent signalling/circadian and redox programs in heme-rich tissues — a class-plausible, unproven concern.
• GSH/redox coupling. The enzyme is glutathione-dependent and its activity intersects the GSH/lipid-peroxide economy ^{5 15}. Predicted: off-target or excessive inhibition might shift cellular redox balance; the direction seen so far (mPGES-2 loss raises GSH and lowers MDA) suggests any redox shift would tend to be protective rather than toxic ^{29 19}.
• Broad constitutive expression. mPGES-2 is expressed widely and constitutively — prominently in heart, and also in brain, skeletal muscle and kidney ^{2 3}. Predicted: wide expression means on-target engagement would occur in many tissues, so the safety margin depends more on selectivity (avoiding the related GST/thioredoxin-fold and heme proteins) than on tissue restriction.

Net predicted assessment: the decisive piece of real evidence — viable, fertile, phenotypically normal Ptges2-knockout mice with unchanged PGE₂ ¹ — argues that on-target mPGES-2 toxicity is likely low. The residual, unproven risk is off-target/selectivity liability, which cannot be estimated because no validated selective inhibitor exists to characterise. This paragraph is a prediction; only the knockout tolerability and biochemistry it rests on are observed facts.

Role in lipid & iron metabolism; relationship to transcription factors

Scope of this section. This part of the dossier covers three linked topics: (1) where mPGES-2 (PTGES2 / Q9H7Z7) sits in lipid metabolism, (2) its relationship to iron, which — for this enzyme — runs entirely through its heme cofactor and is indirect, and (3) its documented and undocumented links to transcription factors (SREBP2, Rev-Erbα/NR1D1, NR4A1, p53, HIF-1α, NRF2/KEAP1, NF-κB). Throughout, evidence is graded and the direction of causality is made explicit: some transcription factors regulate the PTGES2 gene, while others are modulated by the enzyme's activity (either by heme competition or by its product PGE2).

Two adjacent sections own overlapping mechanism and are cross-referenced rather than duplicated here: the iron-dependent ferroptosis mechanism is developed in 50_cell_death.md, and the detailed heme-bound (holo) vs heme-free (apo) biochemistry is developed in 10_heme_configurations.md. The recurring framing for the whole dossier applies: mPGES-2 is a contested PGE2 synthase (knockout mice show unchanged PGE2; UniProt lists the function as "under debate") ², and most of its credible modern biology is heme/redox-driven and largely PGE2-independent.

1. Lipid metabolism

1.1 Canonical position — the PGH2 → PGE2 node of the arachidonic-acid cascade

Direct PTGES2 evidence. mPGES-2 was originally characterised as a prostaglandin E synthase: a glutathione-requiring enzyme that isomerises the cyclooxygenase product prostaglandin H2 (PGH2) to prostaglandin E2 (PGE2), positioning it in the terminal step of the arachidonic-acid (eicosanoid) branch of lipid metabolism, downstream of both COX-1 and COX-2 ^{5 13}. In cellular systems it can couple to both cyclooxygenase-1 and cyclooxygenase-2 to produce PGE2, distinguishing it from mPGES-1, which preferentially couples to COX-2 ¹³.

Contested-synthase caveat (load-bearing). This "PGE synthase" role is disputed. Multiple lines of evidence indicate the isomerase activity is a property of the heme-free (apo) enzyme measured in vitro, whereas the physiological heme-bound (holo) enzyme does not form PGE2 and instead degrades PGH2 ¹⁵. In the liver specifically, mPGES-2 "does not produce PGE2," and its disease role there is PGE2-independent ¹⁸. Tissue context matters: in ovarian granulosa cells, PTGES2 does contribute to PGE2 production and its loss impairs PGE2-dependent processes ³⁴. So the enzyme's lipid role spans a contested PGE2-synthase function (apo, tissue-dependent) and a PGH2-degrading / heme-redox function (holo) — see 10_heme_configurations.md for the apo/holo switch.

1.2 Hepatic lipid metabolism — NASH/NAFLD (PGE2-independent, heme/NR1D1-driven)

Direct PTGES2 evidence [STRONG]. In non-alcoholic steatohepatitis (NASH/NAFLD), whole-body or hepatocyte-specific mPGES-2-deficient mice on high-fat or methionine-choline-deficient diets show reduced hepatic lipid accumulation and ameliorated liver injury, inflammation, and fibrosis ¹⁸. Crucially, the protective effect was independent of PGE2 and instead ran through decreased CYP4A14 and increased acyl-CoA thioesterase 4, controlled by the heme-sensing nuclear receptor NR1D1 (Rev-Erbα); heme regulated the increased NR1D1 activity seen on mPGES-2 loss ¹⁸. This makes mPGES-2 a regulator of hepatic lipid handling via a heme/nuclear- receptor axis, not primarily via prostaglandin output. The pharmacologic tool SZ0232 reproduced the benefit ¹⁸ (inhibitor selectivity is not independently established — see 60_inhibition_effects.md).

1.3 Renal lipid metabolism — diabetic kidney disease (FABP5 / Rev-Erbα)

Direct PTGES2 evidence [STRONG]. In diabetic kidney disease (DKD), enhanced mPGES-2 expression is associated with impaired renal lipid homeostasis and lipotoxicity. Global knockout or pharmacologic blockade of mPGES-2 — and podocyte- or tubule-specific deletion — reduced lipid accumulation and lipotoxicity, attenuating diabetic podocyte injury and tubulointerstitial fibrosis ²⁰. Mechanistically, mPGES-2 and Rev-Erbα competed for heme binding to control fatty-acid-binding protein 5 (FABP5) expression and renal lipid metabolism ²⁰. As in the liver, the operative lever is heme availability to a nuclear receptor, tying a lipid- metabolism phenotype to the enzyme's heme chemistry rather than to PGE2.

1.4 Cholesterol–arachidonic-acid rewiring in HCC — SREBP2 → PTGES2

Direct PTGES2 evidence [MODERATE, tumor]. In hepatocellular carcinoma (HCC), PTGES2 is a node of arachidonic-acid metabolism whose expression is coupled to cholesterol/lipid rewiring. Inhibiting cholesterol de-novo synthesis with atorvastatin, under high-fatty-acid conditions, upregulated PTGES2 via SREBP2-mediated transcription and thereby promoted HCC proliferation and migration; PTGES2 overexpression was pro-tumorigenic and knockdown reversed the statin-enhanced phenotype, and high PTGES2 tracked with poor patient outcome (TCGA + tissue microarray) ²¹. This is the one setting where a transcription factor (SREBP2) is shown to drive the PTGES2 gene (see §3.2), and it directly links sterol/lipid metabolism to eicosanoid output. (Expression details in 20_expression.md; tumor phenotype in 40_tumor_phenotype.md.)

1.5 PTGES2 as a lipid-modified protein — farnesylation (mevalonate/isoprenoid link)

Direct PTGES2 evidence [MODERATE]. PTGES2 is itself a farnesylated protein — a post-translational lipid modification that attaches a 15-carbon isoprenoid (from the mevalonate pathway, the same pathway that produces cholesterol) to the protein ³⁴. In aged ovarian granulosa cells, decreased PTGES2 farnesylation reduced its endoplasmic-reticulum localisation and impaired PGE2 production, compromising cumulus expansion and oocyte maturation; blocking farnesylation (FTI-277) recapitulated the aging phenotype and restoring it (farnesol) rescued the defect ³⁴. Two consequences for this section: (a) PTGES2's subcellular localisation and activity are lipid-modification-dependent, and (b) this connects PTGES2 to the mevalonate/cholesterol axis a second way — as a substrate of prenylation — complementing the SREBP2-transcription link in HCC (§1.4).

1.6 Lipid-metabolism summary and stated absences

Stated absence. No study reports mPGES-2 in de-novo fatty-acid synthesis (FASN/ACC), β-oxidation enzymology, glycerophospholipid/triglyceride synthesis, or a Warburg/glycolytic program as a direct enzymatic participant; its documented lipid roles are the eicosanoid step above plus the regulatory effects on lipid handling mediated by heme/nuclear receptors. Claims of a direct glycolytic ("Warburg") role are not supported for PTGES2.

2. Iron metabolism

Framing (load-bearing honesty). mPGES-2 has no demonstrated role in systemic or cellular iron homeostasis. Every iron connection it has is through its heme cofactor (heme = an iron–protoporphyrin-IX complex) and is therefore indirect. Presenting any of the following as a "direct iron-homeostasis function" would overstate the evidence.

2.1 The one direct molecular iron contact — heme binding

Direct PTGES2 evidence [STRONG, biochemical]. mPGES-2 is a glutathione-dependent heme protein: it binds heme (with its central Fe) only in the presence of glutathione, with a heme dissociation constant of ~2.9 nM, and a substantial fraction of the enzyme is heme-loaded in cells ¹⁵. In the GSH–heme holo structure the heme iron is axially coordinated by the thiolate of the bound glutathione ¹⁷ (detailed in 10_heme_configurations.md). Notably, heme loading of mPGES-2 increased as extracellular Fe³⁺ was raised in the culture medium ¹⁵ — the closest thing to iron-responsiveness in the literature, but it reflects heme availability, not regulation of iron-handling machinery.

2.2 Indirect iron link via ferroptosis (cross-referenced)

Direct PTGES2 evidence, but the canonical mechanism is owned by 50_cell_death.md. Ferroptosis is an iron-dependent form of regulated cell death driven by lipid-peroxide accumulation. In acute kidney injury, genetic or pharmacologic blockade of mPGES-2 inhibited ferroptosis via heme-dependent regulation of the p53/SLC7A11/GPX4 axis ¹⁹. This is the principal (indirect) route by which mPGES-2 touches iron biology: through a heme-dependent, iron-catalysed death program — not through iron transport or storage. Tumor relevance of this axis is Predicted, not demonstrated (see 50_cell_death.md).

2.3 Heme competition / sequestration

Direct PTGES2 evidence [STRONG]. Because mPGES-2 binds heme tightly, the holo enzyme acts as a heme buffer/sink, competing with heme-sensing partners for available heme. This is explicit in two independent disease models: mPGES-2 competes with Rev-Erbα for heme in the diabetic kidney ²⁰, and heme regulates NR1D1 activity liberated by mPGES-2 loss in the liver ¹⁸. Functionally this couples the enzyme to the cellular heme pool (an iron- containing prosthetic group), again indirectly to iron.

2.4 Stated absence — no direct iron-homeostasis role

A targeted PubMed search (2026-07-11) returned no PTGES2/mPGES-2 studies on ferritin, iron homeostasis, transferrin/transferrin receptor, ferroportin, hepcidin, DMT1, or the IRP/IRE iron-regulatory system (0 hits for PTGES2 ferritin, mPGES-2 iron homeostasis). A single hit for PTGES2 heme oxygenase (PMID:31181899) is an unrelated chemoproteomics warhead-target paper, not a heme-oxygenase/iron-recycling link. Conclusion: there is no direct heme-oxygenase (HMOX1), ferritin, or iron-homeostasis role established for mPGES-2 — this is a genuine absence, and the iron relationship should be described strictly as heme-mediated and indirect (heme buffering + ferroptosis).

3. Transcription factors

3.1 Directionality note (read first)

The transcription-factor relationships fall into two distinct classes, which must not be conflated:

• Regulators of the PTGES2 gene (TF → PTGES2 transcription): only SREBP2 is demonstrated (§3.2).
• Factors modulated by mPGES-2 activity (PTGES2 → TF, via heme competition or via the PGE2 product): Rev-Erbα/NR1D1 (heme, §3.3), NR4A1 (via PGE2–EP3, §3.4), p53 (via heme- dependent ferroptosis, §3.5).
• Asserted-but-unsupported links: HIF-1α (§3.6, thin), NRF2/KEAP1 (§3.7, absent), NF-κB (§3.8, belongs to mPGES-1).

3.2 SREBP2 — a demonstrated transcriptional activator of PTGES2 [direct, MODERATE]

Direct PTGES2 evidence. SREBP2 (sterol-regulatory-element-binding protein 2) transcriptionally activates PTGES2: atorvastatin (which activates SREBP2 by lowering cholesterol) upregulated PTGES2 via SREBP2-mediated transcription in HCC, and this drove tumor progression ²¹. This is the only transcription factor shown to control PTGES2 gene expression, and it ties PTGES2 induction directly to the sterol/cholesterol-sensing program (see §1.4).

3.3 Rev-Erbα / NR1D1 — heme-sensing, controlled by heme competition [STRONG]

Direct PTGES2 evidence. NR1D1 (Rev-Erbα) is a heme-sensing nuclear receptor (heme is its physiological ligand). mPGES-2 modulates it not transcriptionally but by competing for heme: in NASH, mPGES-2 loss raises NR1D1 activity through heme, reprogramming hepatic lipid genes (CYP4A14, ACOT4) independently of PGE2 ¹⁸; in DKD, mPGES-2 and Rev-Erbα directly compete for heme binding to set FABP5 expression ²⁰. This is the strongest and most mechanistically specific TF link, and it is a heme-buffering effect (§2.3), i.e. PTGES2 → heme availability → Rev-Erbα activity.

3.4 NR4A1 / Nur77 — via the PGE2–EP3 receptor axis [MODERATE; PGE2-pathway inference]

Direct PTGES2 phenotype, but the TF link is a PGE2-pathway (product) mechanism. In ageing β-cells, mPGES-2 deletion antagonised β-cell senescence and improved glucose homeostasis by interfering with the PGE2–EP3–NR4A1 signalling axis ³⁵. Here the nuclear receptor NR4A1 (Nur77) is engaged downstream of the enzyme's product PGE2 acting through the EP3 receptor — a PGE2-pathway inference, distinct from the direct heme-competition mechanism of Rev-Erbα (§3.3). It should be presented as "PTGES2 → PGE2 → EP3 → NR4A1," not as direct heme sensing.

3.5 p53 — via the heme-dependent ferroptosis axis [MODERATE]

Direct PTGES2 evidence. p53 is engaged as part of the ferroptosis mechanism: mPGES-2 deficiency inhibits ferroptosis through heme-dependent regulation of the p53/SLC7A11/GPX4 axis in acute kidney injury ¹⁹. So p53 is coupled to mPGES-2 through heme/redox chemistry and ferroptosis (cross-ref 50_cell_death.md), not as a classical transcriptional regulator of the PTGES2 gene.

3.6 HIF-1α — thin; product-vs-gene distinction is explicit [THIN]

mPGES-2 / PTGES2 has NOT been shown to be a HIF (hypoxia-inducible factor) target gene, and mPGES-2 has NOT been shown to be hypoxia-regulated. A targeted search returned 0 PTGES2-specific hypoxia/HIF papers (PTGES2 HIF, mPGES-2 hypoxia). Two weak, indirect observations exist and must be kept separate:

1. Co-measurement, not correlation (and non-human). A feline oral-squamous-cell-carcinoma study profiled the three PGE2 synthases PTGES1–3 alongside HIF1A and VEGFA in vitro. PTGES2 was included in the panel but was not among the serum-inducible synthases (PTGES1 and PTGES3 were), and no PTGES2-specific association with HIF1A was reported ²⁷. This is a co-measurement in cat cells, not evidence of HIF regulation of PTGES2.
2. Product effect (downstream of the enzyme), not gene regulation. In the same study, exogenous PGE2 — the enzyme's product — induced HIF1A ²⁷. This is the well-known PGE2→HIF direction and says nothing about whether the PTGES2 gene responds to hypoxia.

Product-vs-gene caveat (load-bearing): "PGE2 stabilises/induces HIF-1α" is a downstream-product effect and is not equivalent to "PTGES2 is a HIF target" or "PTGES2 is hypoxia-regulated." No evidence supports the latter two claims.

3.7 NRF2 / KEAP1 — no direct evidence [ABSENT / Predicted]

Stated absence. A targeted search returned 0 hits for PTGES2 NRF2 and PTGES2 KEAP1. There is no direct study placing PTGES2 as an NRF2 target gene or as a KEAP1/NRF2-pathway component. Predicted (mechanistic, unverified): because mPGES-2 is a GSH-dependent, redox-active heme protein ¹⁵ and NRF2 governs the glutathione/antioxidant response, a functional intersection is plausible — but this is a hypothesis grounded only in shared redox chemistry, not observed data, and must not be stated as fact. Do not claim PTGES2 is an NRF2 target.

3.8 NF-κB — belongs to mPGES-1, not PTGES2 [no direct regulation established]

Identity-guarded absence. The classical NF-κB → inducible-PGES transcriptional axis is a property of mPGES-1 (PTGES / O14684) — a different gene with a different (MAPEG) fold, no heme, and cytokine-inducible expression — e.g. NF-κB driving mPGES-1 expression in fibroblasts and amnion cells ³⁶. These are NOT PTGES2 and are cited here only to mark the confusion. mPGES-2 (PTGES2) is constitutively expressed and is not an established NF-κB target. One neuroinflammation report names a "PAK1–NF-κB–COX-2–PTGES2" cascade in kainate-induced seizure and observes co-regulated PTGES2 "induction" as a pathway readout ³⁷; however, it provides no promoter/ChIP evidence that NF-κB transcriptionally controls PTGES2, so it does not establish NF-κB regulation of the gene. Conclusion: attributing NF-κB-driven transcriptional regulation to PTGES2 is not supported — that mechanism belongs to mPGES-1.

4. Evidence-level summary — transcription factors

Bottom line. mPGES-2's connections to lipid and iron metabolism and to transcription factors are predominantly heme/redox-mediated and PGE2-independent. The strongest, most specific findings are the heme-competition control of Rev-Erbα/NR1D1 in hepatic and renal lipid metabolism, the SREBP2→PTGES2 transcriptional link in HCC, and the heme-dependent p53/ferroptosis axis (the enzyme's only, indirect, route into iron biology). HIF-1α, NRF2/KEAP1, and NF-κB do not have supported direct regulatory relationships with PTGES2, and iron homeostasis (ferritin, heme oxygenase, transport) shows a genuine literature absence.