NAD+ and mitochondria in granulosa cells: energy for ovulation and oocyte quality

Scientific delta of the report

The English twin freezes, in one factual chain, the same post-adversarial reranking as the Spanish report and corrects the bilingual QA failure: required headings now contain mechanism / mecanismo, and the numeric inventory is locked to the Spanish draft. L9-5’s title does not describe a single mitochondrial ATP limited by granulosa NAD+. Proximal ovulation, if cofactor-limited, splits into two non-interchangeable sugar nucleotides that must be titrated in compact COCs at 4–12 h — UDP-GlcNAc/GFPT2 first (L9-5-ED-H13R), cytosolic NAD+/NADH over subsaturated UGDH second (L9-5-ED-H11R) — while 12 h mural OCR, the NAMPT/FK866 pool, already-expanded IVF cumulus, and the healthspan bridge stay outside the ovulatory estimand. The human contrast “high mural NAD / low OCR” does not exist in the same donor. The only coherent distal bridge remains E2/P4 exposure with age-at-natural-menopause trade-offs, not “more ovulatory NAD”.

1. Executive scientific abstract

NAD+ and mitochondria in granulosa do not form a serial module of “energy to ovulate”. LH/hCG sits upstream of the metabolic program. Beneath it are three non-serial cofactor demands: cytosolic NAD+ at UGDH/GAPDH for UDP-GlcUA and hyaluronan; matrix NADPH via IDH2 (ahead of NNT) toward CYP11A1 and pregnenolone; and NAD+ at HSD3B2 for luteal progesterone. Mural OCR peaks at 12 h post-hCG, after the COC expansion window (~4 h). Treating luteal respiration as an ovulatory gate is an error of time and of cell [1,2].

Human evidence does not support that granulosa NAD+ falls with age. In 23 ICSI donors, cumulus NAD+ did not correlate with age, but the cells spent 1–2 h at 37 °C: the null does not inform the in vivo pool [3]. In mural ART granulosa, relative AMA NAD+ rises (52.48 ± 6.44 vs 87.80 ± 2.64, FC 1.67) as peak area, without Seahorse, collinear with AMH (SMD 2.23) [4]. Mural OCR does associate with age in 85 IVF women, and high OCR associates with failed fertilization in ovarian infertility [1]. Those two readouts are not paired. Shi’s ATP row (means 2410 vs 247, arrow ↑, FC 1.02) remains quarantined [4].

Donor-aware computation in 19 ART women shows that expansion identity is HAS2/GFPT2, not UGDH (Δ=+0.174 log2, 12/19, p=0.29 vs HAS2 +0.916, 17/19). NAMPT is mural-high (Δ=−0.866, 18/19). IDH2/CYP11A1 mark mural. SLC25A51 RNA is detectable (GSE46490, 3/3 pairs), but G4-MATRIX remains closed [5–7]. Human UGDH operates at Km NAD+ 0.7 ± 0.1 mM versus extraovarian free NAD+ ~0.1 mM: the NAD–HA link is kinetic of ratio, not of salvage pool [8,9]. FK866 at 4 h does not titrate that lane; Reverchon used 10 nM × 48 h with NAD data not shown and an IGF-1 P4/E2 endpoint, not HA [10,11].

The HBP does move animal expansion: without glucosamine volumetric expansion is ≤25%; DON abolishes it in vitro and, i.p., suppresses murine ovulation; glucosamine restores matrix and worsens blastocysts [12–14]. DON is not GFPT2-specific. HAS2 shRNA reduces expansion without touching the LH program [15]. Therefore the leading ovulatory-substrate hypothesis is UDP-GlcNAc/GFPT2, not NAD. The NAD that can still limit HA is GAPDH/LDH turnover NAD+/NADH during the glycolytic burst.

The bridge to women’s health and longevity remains H0. Genetically later ANM improves bone/T2D and increases hormone-dependent cancers, with a null for lifespan/CVD/Alzheimer [16]. There is no measured mediation by granulosa NAD or OCR. Pharma has no NAD candidate: it has, at most, a lane-occlusion tree.

Executive conclusion: maturity H1 for UGDH enzymology and animal HBP, H2 for HAS2/GFPT2/IDH2/NAMPT RNA, H0 for human ovulatory causality, NAD transfer, ovarian SLC25A51, granulosa-NAD-mediated oocyte quality, and healthspan. The smallest discriminator is Q0-METROLOGY followed by a 2×2 R_cyto × GFPT2 in compact COCs at 4–12 h, with UDP-GlcUA and UDP-GlcNAc as coprimaries. Already-mucified human cumulus ~36 h post-hCG does not titrate the HA burst.

2. Scientific question and relevance

Question

What mechanisms connect NAD+ and mitochondria in granulosa cells — energy for ovulation and oocyte quality — with women’s health and longevity, and which new hypotheses can be falsified?

The question decomposes into four gates that must not be fused:

1. Autonomy. Does a mural or cumulus granulosa NAD pool or ratio cause an ovulatory function that does not require the oocyte (HA/expansion, FSH-driven E2, or luteal P4)?
2. Transfer. Does granulosa matter as a Cx37/TZP bridge of precursor, adenylates or another metabolite into the oocyte (D0-TRANSFER from L9-1)?
3. State. Are age contrasts composition, luteinization, ART or preanalytics (F_granulosa from L9-4 is inherited as a readout, not as a result)?
4. Distal. Is any organ event mediated by a granulosa NAD trajectory, with competing risks, and not by E2/P4 exposure or a shared systemic cause?

“Energy to ovulate” is not ATP or ΔΨm. “Oocyte quality” is not MII, polar body or fertilization in isolation. Later menopause is not healthspan.

3. Scope, population and life stage

The population of interest is the non-tumoral adult female ovary, classified by STRAW+10 when available, not by age alone. The main window is late reproductive (−3a) and early transition (−2), because functional granulosa and possible ovulation remain. ART material is selected by infertility, stimulation, hCG/agonist, time to processing and follicular survival: it informs ex vivo mechanism, not natural ovulation.

Floating mural from follicular fluid, antral mural from the same follicle, compact cumulus, already-expanded post-hCG cumulus, intact COC, oocyte, theca, and the extrafollicular niche are distinguished. KGN and tumor lines qualify as reagents; they do not decide the normal ovary. PCOS, POI, severe endometriosis, gonadotoxicity and NAD-precursor use are recorded as strata, not mixed with physiological aging.

Prescription, supplement, dose, NMN vs NR comparison (L9-3), CD38/flavonoids (L9-4), oocyte SIRT6 (L9-2), current Lua data, and any fusion of fertility, menopause, bone, CVD and mortality into a longevity composite remain out of scope.

4. Background knowledge and mechanism map

4.1 What granulosa NAD is not

Bulk ovarian NAD(H), optical NAD(P)H, relative peak area, blood NAD, oocyte NAD and a colorimetric total-NAD kit are not the compartmental pool. A stable pool can hide high turnover (F_granulosa). NADH, NAD+/NADH and NADP+/NADPH are distinct objects. OCR, ΔΨm, mtDNA copy and ROS are not mitochondrial quality.

4.2 Original chain to destroy

Age → granulosa NAD+ ↓ → mitochondria ↓ → less ovulatory ATP → worse oocyte → menopause/longevity.

That chain is now inference. LH turns the program on; NAD is a cofactor of the switched-on module, not the timer [1,2,17].

4.3 Revised model: four lanes

LH/hCG → LHCGR → AMPc/PKA → AREG/ERK/C/EBPβ/FOS
  ├─ Lane A (4–8 h, COC): GLUT1, glycolysis, HBP, UGDH (2 NAD+/UDP-GlcUA), HAS2 → HA
  ├─ Lane B (mural): StAR → CYP11A1 + FDXR/FDX1 (matrix NADPH: IDH2 > NNT/ME3) → pregnenolone
  ├─ Lane C (~12 h mural): HSD3B2 (NAD+) → P4; mural OXPHOS/FAO
  └─ Lane D: Cx37/TZP, cGMP, adenylates, pyruvate; oocyte NMN→NAD conversion requires COC; NAD cargo = X

Lane 0: ART/luteinization/composition/preanalytics fabricate age contrasts
Distal: E2/P4 × ANM trade-offs; granulosa NAD/OCR ──X──> healthspan

UGDH oxidizes UDP-Glc in two NAD+-dependent steps; at saturating UDP-Glc (~155 μM) NAD+ recharge is the slow step [8]. HAS2 polymerizes HA on the cytosolic face: extracellular UDP-GlcUA is not a rescue [18]. CYP11A1 requires matrix NADPH; CYP19A1 requires ER NADPH; HSD3B2 requires NAD+ [19]. Flint and Denton (1970) localized NADP-IDH and NADP-ME exclusively to cytosol of superovulated rat ovary and stimulated steroid from acetate without raising OCR: that anchors E2/cytosol and does not absolve CYP11A1 of matrix NADPH [20].

5. Evidence method

Scoping, the map, primary verification, synthesis, public computation, hypothesis generation, adversarial review and experimental design were integrated. Decisive claims were checked against DOI/PMID/PMCID. Shi 2026 and Morimoto 2024 were reread at this stage against the article HTML/PMC. Classes: H human, A animal, V in vitro/ex vivo, C computational, I inferred. Inherited verification verdicts: V+, V±, R, Q.

CD38 and SIRT6 are not reopened. Lua data are not used. RNA ≠ cofactor. p>0.05 is not equivalence. Inferential unit = donor or litter, not well or COC.

6. Evidence map

7. Contradictory evidence and null findings

7.1 High mural NAD / low OCR collage

Shi measured peak areas in frozen Percoll granulosa, n=10 young vs 10 AMA, without Seahorse on patient cells. Morimoto measured mural OCR in 85 women without NAD+. Fusing them fabricates a contrast that does not exist (EV-C11). Shi’s OCR is KGN+H2O2 50 μM/2 h with viability 53.5%: cytotoxicity, not senescence [4].

7.2 Cumulus NAD+ null versus murine decline

Mihalas 2026 (preprint; L.E.W. co-founder of Jumpstart Fertility) saw no human NAD+ fall with age and did see NAM/meNAM fall. The authors attribute flat NAD+ to degradation. The murine arm without PMSG, n=4 pools, does lose cumulus NAD+ at 12–14 months [3]. Species and preanalytics are not averaged.

7.3 Ubiquitous UGDH versus cumulus UGDH

In floating FF granulosa, UGDH does not copartition with HAS2. In mural and cumulus from the same MII follicle, UGDH is CGC-biased. That is preparation evidence, not NAD evidence (AR-K2, COMP-C8).

7.4 Expansion versus oocyte quality

Glucosamine increases HA and worsens blastocysts; O-GlcNAc is in cumulus, not oocyte [14]. The subtopic fused proximal ovulation and competence. H13 dissociates them: more HBP can expand and harm. High mural OCR associates with worse fertilization in ovarian infertility [1]. “More mitochondria = better oocyte” is already inverted.

7.5 Flint versus CYP11A1

Cytosolic NADP-IDH/ME in whole rat ovary do not kill matrix NADPH for P450scc. The 1970 homogenate may have lost matrix. H14-E2 is strengthened; H14-pregnenolone is weakened; H8 (IDH2 first) is the default of lane B [19,20].

7.6 Serial NAD→OCR→ovulation

Incompatible with cofactor topology and with mural≠COC kinetics. It is retained only as a clean depletion that phenocopies a prespecified autonomous function. Today it is the lowest-priority hypothesis.

8. Multiscale mechanism synthesis

8.1 Molecular

Three enzymes, three cofactors. UGDH spends 2 NAD+ per UDP-GlcUA and is subsaturated for NAD+ at extraovarian physiological concentrations; GAPDH (Km NAD+ ~20–50 μM, not measured in granulosa) can still live when UGDH has already dropped. GFPT2 limits the other HAS2 substrate (UDP-GlcNAc) with glutamine, not NAD. CYP11A1 receives electrons from matrix NADPH. HSD3B2 oxidizes pregnenolone with NAD+. Complex I uses matrix NAD+ at 12 h mural, outside the HA window.

NAMPT is in human granulosa [10] and is transcriptomically mural [5]. Titrating salvage in cumulus for 48 h asks the wrong compartment and the wrong window.

8.2 Cellular

Luteinizing ART mural ≠ compact cumulus ≠ proximal antral mural. Salustri: cumulus and proximal antral granulosa synthesize HA; peripheral mural does not [26]. FF Percoll is the wrong material for lane A. Identity (FSHR/LHCGR/CYP19A1/HAS2) and luteinization are gates, not covariates (L7-1, L6-5).

Mitochondria do two jobs in the same organelle: the respiratory chain and CYP11A1 compete for IMM, ΔpH and NADPH/NADH. A high OCR can steal reductant from steroidogenesis. That offers a non-H1 reading of the Shi pattern (high relative NAD+/NADP+, low TCA) if it is ever paired with OCR: underconsumption, not youth.

8.3 Intercellular and oocyte

The murine COC is required to convert NMN into oocyte NAD+; CBX 200 μM or denudation do not identify Cx37 or cargo [3]. Dalton moves ATP with AGA [22]. Richard and Baltz localize cGMP to Cx37 [27]. Simon: Gja4 KO destroys Graaf — developmental architecture, not adult NAD cargo [28]. Local ranking: architecture/adenylates/pyruvate/cGMP > NAD transfer. L9-2 holds a potentially autonomous oocyte DDR module; L9-5 does not attribute MII to UGDH.

8.4 Follicle and ovulation

In vivo ovulation is proteases, prostaglandins, vascularization, muscle and an expanded COC. No cellular NAD substitutes for that. The cleanest proximal construct of “energy to ovulate” is HA/expansion in the 4–12 h window. LH ROS is physiological and necessary [29]. FOS moves glycolysis/cholesterol without moving P4 [17].

8.5 Systemic and life course

There is no measured chain from granulosa NAD/OCR → organ event (SCOPE-C4, MS-C19). Ruth 2021: late ANM is not net longevity [16]. 78c did not prolong survival in 15 females (L9-4-K9): it is not imported. A systemic cause (inflammation, metabolism, circadian) can correlate granulosa NAD with healthspan without ovarian mediation. If granulosa sustains E2/P4, the distal bridge is endocrine exposure with trade-offs, identical to L9-1-K10 and L9-4-K10.

8.6 Laboratory lineage

9. Computational layer

Decision: use of conventional paired transcriptomes. BioNeMo omitted: embeddings do not measure cofactor. GSE202601 was not reopened.

GSE18559 (Kõks 2010; 19 IVF women; floating mural vs cumulus) [5]:

• informative anchors: HAS2 +0.916 log2 (17/19, p=2.7×10⁻⁵); CYP11A1 −0.655 (19/19);
• UGDH +0.174 (12/19, p=0.29) — ubiquitous, ceiling ~11.1 log2;
• GFPT2 +1.25 (19/19);
• IDH2 −0.330 (17/19); FDXR −0.304 (18/19); NNT −0.098 (p=0.16);
• NAMPT −0.866 (18/19, p=7.6×10⁻⁶);
• HSD3B2 not mural-high (luteal ceiling);
• SLC25A51 transcribed, weak CGC bias.

GSE46490 (n=3 RNA-seq pairs) [7]: SLC25A51 ≥1 count in 3/3; n=3 does not estimate effect. G4 is not closed for non-expression.

E-MEXP-3641 (12 deposited MII follicles; the paper reports 21 patients) [6]: UGDH +0.74 (9/12, p=0.027); GFPT2 +2.54 (12/12); IDH2 −0.573; NNT −0.946. Sign concordance with HAS2/CYP11A1. COMP-H9 weakened, not refuted: UGDH localization is conditional on whether “mural” is floating FF or the same follicle.

Impact: titrate HAS2/HA first, not UGDH; for NADPH titrate IDH2 first, not adrenal NNT; do not use UGDH/NAMPT RNA as expansion engagement.

10. Primary hypothesis

L9-5-ED-H13R v1 — UDP-GlcNAc/GFPT2 limits HA; NAD is the wrong cofactor of the ovulatory title

Falsifiable statement: at equivalent NAD+/NADH and UDP-GlcUA, reducing GFPT2/UDP-GlcNAc flux will decrease HA/expansion in the 4–12 h window; glucosamine will rescue HA under GFPT2 loss without restoring the NAD+/NADH ratio.

Parents: AR-H13R; HG-H13; HG-C2; Chen 1993; Sutton-McDowall 2004/2006; GSE18559 GFPT2 19/19; AR-C3.

Proposed mechanism: HAS2 needs UDP-GlcUA and UDP-GlcNAc. GFPT2 is the computation’s CGC-stable node. If UDP-GlcNAc is scarce, NAD is surplus. DON aborts expansion; i.p. it suppresses murine ovulation, but DON is treated as off-target until GFPT2 KD reproduces it.

Predictions:

1. Loss of GFPT2 (not DON alone) lowers HA and UDP-GlcNAc at intact ratio and UDP-GlcUA.
2. Glucosamine rescues GFPT2 KD and not the NAD+/NADH reductive clamp.
3. OSMI-1 (H13b) counts only if HA falls with fixed UDP-GlcNAc and fallen HAS2 protein.
4. Excess HBP will expand and lower embryonic competence; excess NAD will not reproduce that trade-off.

Evidence for: animal/in vitro HBP H1; human GFPT2 RNA H2; expansion/competence dissociation [12–14]; CGC-stable computation.

Evidence against: DON inhibits other amidotransferases and lowers glucose uptake [30]; in vivo ovulation after systemic DON may be extrafollicular toxicity; HAS2 transcription may dominate [15]; UGDH remains biochemically obligatory at very low NAD.

Kill: NAD+/NADH or UDP-GlcUA move HA with UDP-GlcNAc in equivalence; GFPT2 loss leaves HA intact; glucosamine does not rescue when UDP-GlcNAc was measured.

Status/maturity: strengthened_contrary leading substrate; H1 animal / H0 human causal. Confidence 0.46.

11. Competing hypothesis

L9-5-ED-H11R v1 — the NAD that can limit HA is turnover NAD+/NADH, not the NAMPT pool

Falsifiable statement: at equivalent HAS2, LH-time, viability and total NAD(H) pool, lowering cytosolic NAD+/NADH will decrease UDP-GlcUA and HA before ECAR/lactate and before UDP-GlcNAc; oxidizing the ratio (pyruvate or cytosolic LbNOX) will rescue HA without adding NAD or glucosamine.

Parents: AR-H11R/AR-H16; Egger 2011; Cambronne 2016; Morimoto 4 h burst; mural-high NAMPT; AR-C1/C2.

Proposed mechanism: UGDH is subsaturated for NAD+. GAPDH is not. A small net UGDH sink versus glycolytic flux only limits HA if the ratio falls near the 0.7 mM Km. LDH recycles NAD+; FK866 at 4 h does not deplete the pool (Tan: 20–30 h in lines) [11]. H16: LDH loss phenocopies the reductive clamp.

Predictions:

1. L/P clamp or cytosolic LbNOX moves UDP-GlcUA/HA; mito-LbNOX does not; FK866 4 h does not, or only with fallen viability.
2. ¹³C-Glc → lactate flux remains in equivalence when HA has already fallen.
3. SLC25A51 K91A/R278A does not reproduce the defect if cytosol is clamped.
4. 4-MU does not count if HAS2 mRNA falls [31].

Evidence for: UGDH enzymology; Km inequality; 4 h COC kinetics ≠ 12 h OCR; operational failure of FK866/extracellular UDP-GlcUA.

Evidence against: Km and free NAD+ are not from granulosa; nobody titrated ratio against HA; HBP can move HA at intact NAD; HAS2 transcription may dominate.

Kill: the ratio falls and UDP-GlcUA/HA remain in equivalence; HBP moves HA with fixed ratio and UDP-GlcUA; HAS2 protein explains everything; mitochondrial LbNOX reproduces the phenotype.

Obligatory prefilter, not a mechanistic null — L9-5-ED-H12R v1: within-state (identity, luteinization, trigger, viability, mural≠cumulus, quench <30 s) abolishes NAD/OCR/HA-Δ by age/STRAW. An age null does not stop H11R or H13R in young material (AR-C6). H12b (any null is explained by adding covariates) is rejected.

Status/maturity: H11R proposed_primary_revised; H1 extraovarian kinetics / H0 COC; confidence 0.28. H12R strengthened_filter; confidence 0.50.

Parallel lane B, not ovulation — L9-5-ED-H8 v1: at non-limiting cholesterol, IDH2 loss lowers matrix NADPH and pregnenolone with E2 and HA in equivalence. Flint anchors H14-E2 (IDH1/ME1 → CYP19A1). NNT in parallel, not first. Confidence 0.30.

12. Translational hypothesis

L9-5-ED-HT3 v1 — lane occlusion, not “ovarian NAD”

Falsifiable statement: a salvage precursor, an NAD sink, an OCR index or a transcript (UGDH, NAMPT, IDH2) will not reach ≥80% occlusion of the lane it claims (A=HA, B=pregnenolone, C=clamped P4) if measured in floating ART granulosa without identity or quench; mural OCR will not occlude A.

The 80% is a provisional gate, not a biological constant. HUMAN_QA_REQUIRED. It is not called a biomarker.

Predictions:

1. NMN/NR or FK866 48 h in Percoll will move P4 or total NAD without occluding cumulus HA.
2. UGDH/NAMPT RNA will not outperform HAS2/GFPT2 as HA correlates.
3. OCR will not predict HA or fertilization monotonically.
4. Shi’s AMA peak-area contrast will not replicate in quenched within-state absolute LC–MS.
5. None of the above predicts an organ event (L9-5-HG-H15; G8 closed).

Kill: an input occludes ≥80% of a prespecified lane, the effect disappears on restoring that cofactor/substrate and not the others, in ≥2 donors, with Q0 standing.

Status/maturity: parked_translational_test; integrated H0; H1 for already-audited metrology failures. Confidence 0.45 for “the current panel does not occlude A”.

13. Falsifiable predictions and kill criteria

Provisional quantitative rule: with valid Q0, kill a lane if the coprimary change stays below max(MDC95, 10% of the positive-control range) or if HT3 occlusion is <80%. The 10/20/80% thresholds are gates, not constants.

14. Discriminating experiment

Q0-METROLOGY — hard stop

Qualify isotope-dilution LC–MS/MS of NAD(P)(H), NAM/meNAM; UHPLC–MRM of UDP-GlcUA vs UDP-GlcNAc vs UDP-GalNAc [32]; HA (HABP ELISA + hyaluronidase or [³H]glucosamine, not diameter alone); LC–MS steroids; Peredox/SoNar calibrated by L/P [33,34]; viability. Recovery 80–120%, QC CV ≤15%. If GlcNAc/GalNAc do not baseline, H13a degrades to generic HBP and is not killed. Resolution failure stops G5.

Q0-PAIR — H12a filter, not a stop license

Same ART donor, mural≠cumulus, quench <30 s, absolute LC–MS, identity, luteinization, descriptive OCR/mass. Start n=12 (6+6); one blinded recalculation to n=8–16/stratum. Residual HA is descriptive: already-expanded cumulus is not G5-A' (ED-C1). If age dies within-state, H12a wins the age contrast and G5-A' opens in a young model (ED-C2).

G5-A' — 2×2 R_cyto × GFPT2 in compact COC (smallest real experiment)

Model: compact COCs from superovulated prepubertal mice, harvest before in vivo hCG, 4 and 12 h with FSH/AREG/EGF ± serum (Salustri: HA 20–30× at 3–12 h; serum retains matrix) [35]. Conditional human replica: compact IVM, not mucified ICSI cumulus.

Total lactate+pyruvate fixed; glucose conserved because Chen/Sutton need it; Peredox must move. Nested: LDH/oxamate (H16); FK866 4 h (pool negative); OSMI-1/Thiamet-G (H13b); DON ± glucosamine (H13c); UGDH KD as enzyme control; 4-MU only if HAS2 mRNA is reported. Forbidden: extracellular UDP-GlcUA.

Coprimaries: HA at 4 h and 12 h, UDP-GlcUA, UDP-GlcNAc, HAS2 protein, NAD+/NADH, NAD pool, viability. Embryonic competence is not an A coprimary; if measured, it is the Sutton-McDowall trade-off (AR-K6).

Pilot n=6 litters; confirmatory n=8–12/cell, one recalculation for 80% power of a 30% material change or TOST ±20%. Unit = litter. Holm on coprimaries.

Stop of A: if the 2×2 does not move HA outside equivalence, the title “NAD energy to ovulate” dies. HAS2 transcription and state remain. NMN/NR is not opened.

G5-B / G5-C / G4 / G3 — conditionals

G5-B (only if A is interpretable or a parallel steroidogenesis program is declared): post-hCG human mural or non-luteinized murine mural; 22R-OHC + trilostane; IDH2 vs IDH1/ME1; parallel NNT; compartmental iNap/TPNOX [19,36]. G5-C: P4 at clamped pregnenolone. G4-MATRIX: SLC25A51 WT vs K91A/R278A only if B or OCR move. G3-TRANSFER: only if G5-A' is an autonomous null; load, washout, Cx37 (not CBX alone). G8 closed.

The minimum that changes the decision is Q0-METROLOGY + G5-A'. Q0-PAIR orders the age contrast; it is not a mechanistic prerequisite of young lane A.

15. Biomarkers and stratification

There is no validated biomarker. Experimental candidate readouts:

• R_cyto (sensor + LC–MS) and NAD(H) pool, not optical NAD(P)H;
• absolute UDP-GlcUA and UDP-GlcNAc, isomer-resolved;
• matrix+medium HA normalized per COC/protein, plus HAS2 protein;
• pregnenolone at clamped cholesterol and P4 at clamped pregnenolone;
• descriptive OCR/mass, never A primaries;
• FSHR/LHCGR/CYP19A1/HAS2/PTX3 and viability as identity gates.

Strata: mural≠cumulus, time post-hCG, trigger (hCG vs dual), AMH/oocytes as covariates not mediators, STRAW+10 if available, PCOS/endometriosis as registry. AMH, retrieved oocytes, MII, menopause, blood NAD and UGDH RNA do not substitute for engagement (AR-C5, HG-C5).

16. Individual variability

Magnitude and sign can vary by:

• mural vs cumulus vs lutein identity and Percoll fraction;
• reserve, ART protocol and dual trigger;
• preanalytics (37 °C, freeze–thaw, quench);
• HAS2, GFPT2, UGDH, LDH, IDH1/IDH2 expression/activity;
• glucose/glutamine availability and cytosolic redox state;
• CLOCK/circadian (L4.1) and ER-stress (L2) as rivals;
• unknown LATAM ancestry and transportability.

Heterogeneity must not rescue a null primary post hoc. Modifiers are prespecified after Q0.

17. Pharma relevance and maturity

The potential asset is not a salvage precursor for “ovulatory energy” or oocyte quality. The NAMPT/FK866 pool is misspecified; human cumulus NAD does not inform the pool; relative AMA mural NAD is high; high OCR associates with worse fertilization. Nor is it an HBP booster: glucosamine expands and worsens the embryo. DON/azaserine/DRP-104 are oncologic glutamine antagonists, disqualified as ovarian probes.

If G5-A' is positive, the tool opportunity (not a ready drug) is lane occlusion: cytosolic NAD+/NADH ratio in a HAS2 cell, or GFPT2/UDP-GlcNAc flux. If G5-B is positive, matrix NADPH (IDH2) for pregnenolone, not ATP. Lever chemistry exists outside the ovary (LbNOX/TPNOX, OSMI-1, trilostane, 22R-OHC) for target validation, not for human ovarian exposure.

Go to early discovery: Q0 passes; G5-A' occludes A with substrate/ratio rescue; the HA vs competence trade-off is measured and not hidden.

No-go for an “ovarian NAD / oocyte quality / delayed menopause” indication: null A factorial; DON-only positive; Percoll/KGN only; OCR as primary; precursor without occlusion; healthspan bridge.

Current maturity: H1 UGDH enzymology and animal HBP; H2 HAS2/GFPT2/IDH2/NAMPT RNA; H0 human ovulatory causal and H0 Pharma. There is no partnering readiness, candidate or human trial. HUMAN_QA_REQUIRED. H5 is not marked.

18. Limitations

1. Almost all available human granulosa is post-hCG ART; natural STRAW+10 is absent.
2. UGDH Km and free NAD+ are not from granulosa; Cambronne is HEK293T.
3. Primary G5-A' is superovulated animal; compact human IVM may not materialize.
4. Chen, Sutton-McDowall, Sugiura and Flint are not natural periovulatory woman; DON is not GFPT2-specific.
5. Mihalas 2026 is a preprint with NAD-precursor COI; the human null was used cognitively as pool conservation.
6. Shi: peak areas, n=10, dual trigger unstratified, corrupt Table 2 ATP; OCR only in injured KGN.
7. 2010–2012 RNA is not 2026 protein or flux; chip ceiling on UGDH/HSD3B2.
8. Garcia 2013 qualified sugar nucleotides in breast lines, not COCs.
9. LbNOX/iNap/Peredox are not demonstrated in cumulus; if transduction fails, the L/P clamp remains the only lever.
10. 10/20/30/80% margins are inherited gates, not known effects.
11. Cx37 NAD transfer, ovarian SLC25A51 and granulosa NNT remain H0.
12. Genetic ANM does not simulate a localized adult NAD intervention.
13. Nothing is prescribed. Lua data are not used. In vivo ovulation and lifespan are not inferred from one cell.

19. Conclusions

L9-5 converts an ovarian-rejuvenation narrative — NAD falls, mitochondria fail to supply energy, the oocyte ages, the woman loses years — into a falsifiable sequence that can end early with an informative no-go.

What is usefully demonstrated is narrow: mural and COC do not share periovulatory kinetics; human mural OCR associates with age and, in one stratum, associates inversely with fertilization; HAS2/GFPT2 specify cumulus and CYP11A1/IDH2 specify mural; UGDH spends NAD+ and is kinetically subsaturated; the HBP can abort animal expansion; human NAMPT exists and is mural-high; ANM is not net longevity.

What is not demonstrated is the subtopic title. There is no human NAD–OCR pair. There is no quenched cumulus pool. There is no NAD vs HA titration. There is no Cx37 NAD cargo. There is no functional ovarian SLC25A51. There is no mediation to an organ event.

The highest-value hypothesis is no longer “restore ovarian NAD”. It is more original and narrower: the expansion cell is specified by HAS2/GFPT2; the best-supported HA limiter today is UDP-GlcNAc; the only NAD that can still touch that lane is a cytosolic ratio over ubiquitous UGDH during a 4 h burst; and 12 h mural OCR is luteinization, not ovulation. H12 filters age. H8 asks pregnenolone, not ATP. Healthspan remains H0. A clean refutation of the NAD title is scientific success.

20. References

1. Morimoto A, et al. Granulosa cell metabolism at ovulation correlates with oocyte competence and is disrupted by obesity and aging. Hum Reprod. 2024;39:2053–2066. DOI: 10.1093/humrep/deae154. PMID 39013118; PMCID PMC11373349.
2. Pansera M, Neeraj N, Siddappa D, Schuermann Y, Duggavathi R. ERK1/2–C/EBPβ regulation of Slc2a1/GLUT1 after hCG. Theriogenology. 2025;236:13–20. DOI: 10.1016/j.theriogenology.2025.01.028. PMID 39893797.
3. Mihalas B, et al. NAD+ metabolism in cumulus and oocytes with age. Preprint 2026. DOI: 10.64898/2026.07.22.739977. COI: Jumpstart Fertility.
4. Shi M, Jia Z, Yang X, et al. Age-related mitochondrial energy metabolism reprogramming occurs in granulosa cells during ovarian aging. Front Endocrinol. 2026. DOI: 10.3389/fendo.2026.1726339. PMID 41704484; PMCID PMC12907147.
5. Kõks S, Velthut A, et al. Differential microarray of human mural vs cumulus granulosa. Mol Hum Reprod. 2010. DOI: 10.1093/molehr/gap103. GSE18559.
6. Grøndahl ML, et al. The dormant and the fully competent oocyte. Mol Hum Reprod. 2012. DOI: 10.1093/molehr/gas035. E-MEXP-3641.
7. Velthut-Meikas A, et al. RNA-seq of paired mural and cumulus. Mol Endocrinol. 2013. DOI: 10.1210/me.2013-1058. GSE46490.
8. Egger S, Chaikuad A, Kavanagh KL, Oppermann U, Nidetzky B. Structure and mechanism of human UDP-glucose 6-dehydrogenase. J Biol Chem. 2011;286:23877. DOI: 10.1074/jbc.M111.234682. PMCID PMC3129169.
9. Cambronne XA, et al. Biosensor reveals multiple sources for mitochondrial NAD+. Science. 2016;352:1474. DOI: 10.1126/science.aad5168.
10. Reverchon M, et al. Visfatin/NAMPT in human granulosa and steroidogenesis. Mol Hum Reprod. 2013. DOI: 10.1093/molehr/gat002.
11. Tan B, et al. Pharmacological inhibition of nicotinamide phosphoribosyltransferase (NAMPT). J Biol Chem. 2013. DOI: 10.1074/jbc.M112.394510.
12. Chen L, Russell PT, Larsen WJ. Functional significance of cumulus expansion. Mol Reprod Dev. 1993;34:87–93. PMID 8418823.
13. Sutton-McDowall ML, Gilchrist RB, Thompson JG. Cumulus expansion and glucose metabolism. Reproduction. 2004;128:313–319. DOI: 10.1530/rep.1.00225.
14. Sutton-McDowall ML, et al. Glucosamine supplementation during IVM. Biol Reprod. 2006;74:881–888. DOI: 10.1095/biolreprod.105.048553.
15. Sugiura K, Su YQ, Eppig JJ. Targeted suppression of Has2 mRNA in mouse cumulus. Mol Reprod Dev. 2009;76:537–547. DOI: 10.1002/mrd.20971. PMCID PMC2669700.
16. Ruth KS, et al. Genetic insights into biological mechanisms governing human ovarian ageing. Nature. 2021;596:393–397. DOI: 10.1038/s41586-021-03779-7. PMID 34349265; PMCID PMC7611832.
17. Choi H, et al. FOS transcription factor regulates human granulosa cell luteinization. Endocrinology. 2021. PMID 34171102; PMCID PMC8315293.
18. Vigetti D, et al. Cell energy metabolism and hyaluronan synthesis. Int J Mol Sci. 2020. PMCID PMC7780193.
19. Miller WL, Auchus RJ. The molecular biology, biochemistry, and physiology of human steroidogenesis. Endocr Rev. 2011. PMID 21051590.
20. Flint APF, Denton RM. The role of nicotinamide-adenine dinucleotide phosphate-dependent malate dehydrogenase and isocitrate dehydrogenase in the supply of NADPH for steroidogenesis. Biochem J. 1970;117:73–83. DOI: 10.1042/bj1170073. PMID 4393612.
21. Liu Y, et al. Age-related changes in human granulosa cell mitochondria. Hum Reprod. 2017. DOI: 10.1093/humrep/dex309. PMID 29045673.
22. Dalton CM, Szabadkai G, Carroll J. Measurement of ATP in single oocytes. J Cell Physiol. 2014. DOI: 10.1002/jcp.24457.
23. Luongo TS, et al. SLC25A51 is a mammalian mitochondrial NAD+ transporter. Nature. 2020. DOI: 10.1038/s41586-020-2741-7.
24. Wang Y, et al. NAD+ metabolism in porcine follicular atresia. Free Radic Biol Med. 2025. DOI: 10.1016/j.freeradbiomed.2025.07.049. PMID 40752633.
25. Wang Y, et al. H2O2-induced NAD decline and progesterone in mouse granulosa. J Ovarian Res. 2026. DOI: 10.1186/s13048-026-02025-w. PMID 41764539.
26. Salustri A, Yanagishita M, Underhill CB, Laurent TC, Hascall VC. Localization and synthesis of hyaluronic acid in cumulus and mural granulosa. Dev Biol. 1992.
27. Richard S, Baltz JM. Prophase I arrest in mammalian oocytes mediated by natriuretic peptide and cyclic GMP. Biol Reprod. 2014. DOI: 10.1095/biolreprod.114.118505. PMID 24804968.
28. Simon AM, Goodenough DA, Li E, Paul DL. Female infertility in mice lacking connexin 37. Nature. 1997;385:525–529. DOI: 10.1038/385525a0.
29. Shkolnik K, et al. Reactive oxygen species are indispensable for ovulation. Proc Natl Acad Sci USA. 2011. DOI: 10.1073/pnas.1017213108.
30. Gutnisky C, et al. Glucosamine supplementation and DON in bovine COCs. Reproduction. 2007.
31. Kultti A, et al. 4-Methylumbelliferone inhibits hyaluronan synthesis. Exp Cell Res. 2009. PMID 19285976.
32. Garcia AD, et al. UHPLC–MS/MS of nucleotide sugars. Rapid Commun Mass Spectrom. 2013;27:1794–1800. DOI: 10.1002/rcm.6631. PMID 23821573.
33. Hung YP, Albeck JG, Tantama M, Yellen G. Imaging cytosolic NADH–NAD+ redox state with a genetically encoded fluorescent biosensor. Cell Metab. 2011;14:545–554. DOI: 10.1016/j.cmet.2011.08.012.
34. Titov DV, et al. Complementation of mitochondrial electron transport chain by manipulation of the NAD+/NADH ratio. Science. 2016;352:231–235. DOI: 10.1126/science.aad4017.
35. Salustri A, Yanagishita M, Hascall VC. Synthesis and accumulation of hyaluronic acid during expansion of the mouse cumulus. J Biol Chem. 1989;264:13840–13847. DOI: 10.1016/S0021-9258(18)80077-1.
36. Tuckey RC. Cholesterol side-chain cleavage by mitochondria from human placenta and from porcine ovary. Eur J Biochem. 1989;185:131–137. DOI: 10.1111/j.1432-1033.1989.tb15203.x.
37. Li J, et al. Multi-omics of mouse cumulus metabolism during oocyte maturation. J Ovarian Res. 2023. DOI: 10.1186/s13048-023-01237-8.
38. Ortiz-Meoz RF, et al. A small molecule that inhibits OGT. ACS Chem Biol. 2015;10:1392–1397. DOI: 10.1021/acschembio.5b00004.
39. Smits MAJ, et al. Human oocyte NAD+ metabolome with age. Hum Reprod. 2023. DOI: 10.1093/humrep/dead177. PMCID PMC10628503.
40. Anaya-Hernández A, et al. IDH2 in a placental steroidogenic complex. 2019. PMID 31542634.

21. Claims and hypotheses of this report

Claims added: L9-5-EN-C1, L9-5-EN-C2, L9-5-EN-C3.

• L9-5-EN-C1: L9-5’s ovulatory title is frozen as two non-serial sugar nucleotides (UDP-GlcNAc/GFPT2 leading; cytosolic NAD+/NADH over UGDH second) titratable only in compact COCs at 4–12 h.
• L9-5-EN-C2: the human high mural NAD / low OCR collage, the post-ICSI cumulus null and Shi’s ATP row remain rejected or quarantined after primary rereading of Shi 2026 and Morimoto 2024.
• L9-5-EN-C3: Pharma and healthspan do not inherit an NAD candidate; they inherit a lane-occlusion tree and closed G8 (ANM trade-offs).

Hypotheses touched: L9-5-ED-H13R v1, L9-5-ED-H11R v1, L9-5-ED-H16 v1, L9-5-ED-H12R v1, L9-5-ED-H8 v1, L9-5-ED-HT3 v1, L9-5-HG-H15 v1, L9-5-AR-H11R v1, L9-5-AR-H13R v1, L9-5-HG-H11 v1, L9-5-SCOPE-H1 v3, L9-5-SCOPE-H4 v3.

Nothing is written into memoria agentes/Lua Labs/reportes/. Laboratory memory is not consolidated. Nothing is prescribed.