CD38 as an NAD+ siphon: natural inhibitors and reproductive longevity

Scientific delta of the report

The final synthesis corrects the narrative that CD38 directly “drains” NAD+ from the oocyte and that apigenin or quercetin can reverse that process. The evidence supports a narrower claim: CD38 controls a material fraction of the bulk NAD+ pool in the whole murine ovary, with a predominantly extrafollicular signal, but transmission of that perturbation into newly synthesized NAD+ in granulosa cells or the oocyte, mediation of a reproductive function, and human causality have not been demonstrated. The pseudobulk analysis adds a new hypothesis but does not confirm it: at 3–9 months the aggregate increase in Cd38 is mostly compositional and the positive within-lineage signal is endothelial, whereas at 14–17 months a myeloid signal appears; after adversarial review, the leading explanation is atresia/remodeling first and CD38 as an amplifier of the niche, not an established endothelial→myeloid relay. Apigenin and quercetin are reclassified as pleiotropic probes conditional on chemical species, free exposure, and genetic occlusion. The decisive experiment must measure lineage-specific activity, absolute cell number, and isotopic transfer into granulosa cells before testing compounds or longevity.

1. Executive scientific abstract

CD38 is an ectoenzyme and NADase capable of hydrolyzing NAD+ and producing NAM, ADPR, and cADPR. In the aging murine ovary, its expression and activity increase in predominantly extrafollicular compartments—stroma, vasculature, corpora lutea, atretic areas, and immune cells. Two genetic/pharmacological studies show that global Cd38 loss or inhibition with 78c increases bulk ovarian NAD+; the absence of any additional increase from 78c in Cd38−/− supports target dependence for that pool [1,2]. This is the strongest causal link.

The downstream links remain open. No tracing experiment shows that inhibiting CD38 in endothelial, myeloid, or stromal cells increases newly synthesized NAD+ in granulosa cells or the oocyte. Nor has it been determined whether a function changes through NAD+ conservation, ADPR/cADPR–Ca²+ signaling, altered inflammation, or systemic effects. Constitutive knockout changes neonatal follicle formation and the whole organism; therefore, more follicles early in life do not identify a slower rate of adult depletion. Murine reproductive results are discordant: Yang et al. report higher AMH, more follicles and oocytes, improved spindle integrity, and larger litters; Perrone et al. find no sustained late benefit, observe convergence of WT/KO NAD+ at 20 months, and document a neonatal phenotype [1,2].

Public-data computation changes the validation order, not the causal conclusion. In GSE232309, with 4 mice aged 3 months and 4 aged 9 months, Cd38 did not increase within the myeloid lineage (−0.318 log2CPM), increased in endothelium (+0.519 log2CPM), and was unstable/sparse in stroma. The captured myeloid fraction increased 2.13-fold, and a decomposition attributed 58.6% of the aggregate Cd38 change to composition. In GSE236712, with 3 young and 3 old pools, myeloid Cd38 increased +1.018 log2CPM at 14–17 months. Because age, enrichment, QC, and experimental unit differ, these findings are neither a longitudinal trajectory nor a replication; they motivate an age×lineage interaction test with activity measured in one cohort [16,17].

Apigenin and quercetin inhibit purified CD38 or lysate with approximate IC50 values of 10–16 µM and increase NAD+ in WT, but not Cd38−/−, cells at 25–50 µM [6]. This is biochemical evidence, not ovarian or clinical evidence. Available human exposure studies measured total plasma concentrations or concentrations after hydrolysis of conjugates: approximately 0.127 µM apigenin after parsley and up to 2.48 µM total quercetin derived from conjugates in a small crossover study; neither free ovarian aglycone nor target engagement was measured [7,8]. Oral quercetin was null in aged mice and null or unfavorable in young mice, whereas 10–20 µM ex vivo produced favorable signals in selected murine/human oocytes [9–12]. Apigenin also inhibits NLRP3/IL-1β and Ca²+ flux in macrophages even without CD38, demonstrating functional pleiotropy [20].

The connection to women’s health and longevity is not monotonic either. Genetics of age at natural menopause supports benefits for bone and perhaps type 2 diabetes, but higher risk of some hormone-dependent cancers and no robust effect on lifespan, cardiovascular disease, or Alzheimer disease [13,14]. In the only longevity study of 78c, benefit was significant in males, not females; the treated female group was small, and the null result does not establish equivalence [5]. Ovarian preservation, prolonged hormonal exposure, and systemic CD38 inhibition are distinct interventions.

Executive conclusion: current maturity is H1 for control of the bulk murine ovarian NAD+ pool, H2 only for murine transcriptomic localization, and H0 for follicular transmission, functional mediation, flavonoid specificity, human causality, and longevity. The leading hypothesis is that remodeling/atresia and composition increase the total contribution of CD38-high cells and that CD38 amplifies the niche without demonstrated follicular transmission. The smallest discriminator is a single 3/9/15-month cohort with absolute cell counts, surface CD38, and lineage-specific NADase activity, followed by adult endothelial versus myeloid deletion and isotopic tracing into granulosa cells. If the change remains bulk or produces no functional effect, flavonoids and translation stop.

2. Scientific question and relevance

Question

Does age- or stage-dependent CD38 activity, localized to an adult ovarian compartment, consume a material fraction of NAD+ that reaches the follicle and cause loss of reproductive function; and can apigenin or quercetin reproduce a benefit through CD38 inhibition at a plausible tissue exposure?

The question contains four gates that must be passed independently:

1. Target: material catalytic CD38 activity in an adult ovarian lineage.
2. Transmission: altered local flow/pool into newly synthesized NAD+ in granulosa cells or the oocyte.
3. Function: change in an independent outcome—progression/atresia, competence, live birth—not merely pool size or AMH.
4. Chemical translation: free species, target engagement, occlusion by Cd38 loss, and catalytic rescue.

Relevance to women’s health depends on a fifth gate: showing that maintained ovarian function mediates an organ-specific outcome without being explained by direct somatic action, uterine effects, baseline state, or competing risks.

3. Scope, population, and life stage

The human population of interest is the nonmalignant adult female ovary from early reproductive life through the early menopausal transition, ideally classified by STRAW+10 rather than age alone. Assisted-reproduction tissue is selected by infertility, stimulation, maturity, and survival; it can inform ex vivo mechanism but does not automatically represent the natural ovary.

Available causal evidence comes from female mice. Ages 3, 8–9, 12, 14–17, and 20 months do not map one-to-one to human stages. Estrous cycle, corpora lutea, colony, microbiota, background, and survival alter comparisons. A germline knockout only informs possibility and cumulative effects; it does not identify an intervention initiated after maturity.

The report distinguishes:

• intact ovary, extrafollicular niche, and follicle;
• granulosa/cumulus and oocyte;
• the postovulatory oocyte, which represents a distinct acute clock;
• endometrium and uterine receptivity;
• endocrine function, reserve, throughput, competence, and live birth;
• organ-specific somatic outcomes and survival.

Consumption recommendations, supplements, or dosing; extrapolation to human fertility; and any use of current Lua data are excluded.

4. Background knowledge and mechanistic map

4.1 What “siphon” means

The term is valid only if CD38 explains a quantifiable contribution to NAD+ consumption flux in a defined compartment. RNA expression, protein abundance, percentage of CD38+ cells, activity with a fluorogenic substrate, and a bulk pool are not equivalent to in vivo flux.

The NAD+ pool reflects the balance of synthesis/input and consumption. NAMPT and NMNAT sustain salvage; transporters and cumulus–oocyte communication determine which precursor reaches each cell; PARP, sirtuins, CD38, BST1, and other enzymes consume pools that may be partly separated. Absolute NAD+, NADH, the NAD+/NADH ratio, and optical NAD(P)H are not interchangeable.

4.2 Originally favored chain

Age/SASP/inflammation → extrafollicular CD38+ cells → consumption of niche NAD+/NMN → less follicular NAD+ → poorer bioenergetics, repair, and resilience → more atresia/lower competence → shorter reproductive longevity.

This chain is plausible, but only the segment perturbed CD38 → bulk ovarian NAD+ pool in mice has been demonstrated.

4.3 Revised model

The final model contains four tracks:

Track A — compositional amplification, current leader. Mitochondrial/DDR injury, atresia, fibrosis, or lower salvage precede niche change. Follicular loss and SASP expand or activate CD38-high cells. CD38 further lowers bulk NAD+ but may not transmit a material perturbation to the follicle.

Track B — causal niche siphon. Endothelial or myeloid CD38 consumes extracellular NAD+/NMN or alters purinergic products; the change precedes injury and reaches granulosa cells/the oocyte. This version requires tracing and adult lineage-specific perturbation.

Track C — cADPR–Ca²+ messaging. CD38 changes substrate and messengers simultaneously. Function may follow a Ca²+ waveform even when NAD+ is not limiting. Because Ca²+ pulses are physiological for maturation/fertilization and overload is harmful, the sign may depend on the window.

Track D — pleiotropic flavonoid. Micromolar apigenin/quercetin aglycone interacts with CD38 and also with inflammation, redox, kinases, transport, and other enzymes. Without occlusion, the effect is provisionally attributed to pleiotropy.

4.4 Two clocks that must not be merged

Ovarian-niche aging unfolds over months/years and integrates endowment, recruitment, atresia, stroma, immunity, and the HPO axis. Postovulatory aging unfolds over hours/days in an isolated oocyte. Singh et al. observed lower NAD+, higher Ca²+, ROS, hyperacetylation, retrotransposons, and damage in postovulatory-aged murine oocytes; quercetin 20 µM or NMN 100 µM partially delayed death [15]. CD38 and NAD+ after quercetin were not measured, there was no genetic occlusion, and the animal unit was uncertain. The finding demonstrates an NAD-sensitive phenotype, not a CD38 pathway or reproductive longevity.

5. Evidence method

Scoping, evidence mapping, verification, synthesis, computation, hypothesis generation, adversarial review, and experimental design were integrated. Decisive claims were checked against primary sources through DOI/PMID/PMCID. The 2025–2026 update was verified against primary papers on ovarian inflammaging, apigenin/NLRP3, and myeloid CD38 in endometrium [18–20,23].

Evidence was classified as:

• Human: observational, genetic, pharmacokinetic, or ex vivo; not causal in vivo by default.
• Animal: in vivo perturbation or measurement; not automatically transportable to women.
• In vitro/ex vivo: biochemical, cellular, oocyte, or explant evidence; dependent on exposure and model.
• Computational: prioritizes lineage and generates predictions; does not measure activity or flux.
• Inferred: connects links not measured together and is labeled explicitly.

For decisive sources, design, population/sex/life stage, n, tissue, intervention/comparator, outcome, uncertainty, and identifier were extracted. Null findings, contradictions, potential pseudoreplication, and source misattribution were preserved. Lorenz 2014 studied EGCG, not quercetin, and is not used as evidence about quercetin/COMT.

6. Evidence map

7. Contradictory evidence and null findings

7.1 Perrone versus Yang

The studies converge on control of the murine NAD+ pool but not reproductive longevity. Perrone localizes CD38 outside the follicle, observes higher NAD+ and follicle counts mainly at early ages, convergence of WT/KO at 20 months, no improvement in latency or number of litters, and delayed neonatal follicle formation [1]. Yang reports higher NAD+, AMH, follicle and oocyte counts, improved spindle integrity, and larger litters with KO/78c [2].

These are not direct replications: allele, littermate/colony structure, windows, attrition, and metrics differ. Competitors are neonatal endowment, systemic effects, colony heterogeneity, power/state, or a genuine context-dependent adult benefit. The smallest discriminator is postmaturity inducible deletion with baseline reserve and composition matched.

7.2 Bulk pool versus follicle

The lack of an additional increase from 78c in Cd38−/− supports target dependence for bulk NAD+. It does not show that NAD+ mediates AMH, spindle, oocyte, or litter outcomes. The pool may rise in immune/stromal cells without changing granulosa cells, or function may follow cADPR/Ca²+, inflammation, uterus, or systemic metabolism.

7.3 Composition versus per-cell induction

In GSE232309, the captured myeloid fraction rises 2.13× while myeloid Cd38 pseudobulk falls. The 58.6% decomposition describes captured cells, not mass or activity. Fibrosis, fragility, cell size, and dissociation alter recovery. Composition gains priority but cannot be declared causal without absolute cell counts and per-cell NADase activity.

7.4 Oral quercetin versus ex vivo

Oral quercetin 30 mg/kg for 21 days was null for oocytes, fertilization, and blastocysts in aged mice [9]; in young mice it did not improve most outcomes and yielded unfavorable signals for hatching/trophectoderm, with clustering limitations [10]. By contrast, 10 µM ex vivo improved maturation, fertilization, and blastocyst formation in discarded human oocytes, but the denominator per donor was unclear and SIRT3 dependence was incomplete [11]. The pattern favors exposure/localization or pleiotropy, not CD38.

7.5 Apigenin as a biochemical inhibitor and functional off-target

Apigenin inhibits purified CD38 [6] but reduces NLRP3 activation and Ca²+ even in Cd38−/− macrophages [20]. Therefore, lower inflammation or Ca²+ does not constitute CD38 target engagement. Only catalytic product, occlusion, and rescue can attribute mechanism.

7.6 Reproduction and longevity

Premature menopause is associated with higher mortality, but the gradient does not show that delaying a normal menopause increases survival [22]. MR of age at menopause shows trade-offs and does not support global lifespan [13,14]. The female 78c null is not equivalence, but it blocks extrapolation of the male benefit [5].

8. Multiscale mechanistic synthesis

8.1 Molecular

CD38 transforms NAD+ and simultaneously modifies NAM, ADPR/cADPR and potentially Ca²+ signals. Conserving NAD+ could support redox, ATP, sirtuins, and PARP; reducing cADPR could lessen Ca²+ overload or interfere with physiological pulses. Current tools—KO, 78c, and flavonoids—change both branches. An NAD+–function correlation does not choose the mediator.

8.2 Cellular

The most consistent signal is extrafollicular. CD38+ immune cells may respond to SASP; endothelium may alter precursor availability, perfusion, or purinergic signaling; stroma may amplify fibrosis. Granulosa and cumulus are plausible receivers because they sustain metabolism and transfer to the oocyte, but no niche→granulosa mass balance exists.

The strictest prediction is temporal: lineage activity must change first, then the extracellular precursor/product, then newly synthesized granulosa NAD+, and only finally injury or function. If the signal appears after atresia or only through increased cell number, CD38 is an amplifier or marker, not an initiator.

8.3 Follicular and reproductive

More follicles may reflect:

1. greater initial endowment;
2. lower adult atresia;
3. lower activation/recruitment and lower throughput.

Only the second, without worsening competence and output, supports reproductive preservation. AMH, MII, spindle, fertilization, blastocyst, live birth, and reproductive span are distinct endpoints. Conditioning only on MII or surviving embryos introduces selection.

8.4 Uterine

Hua et al. showed that myeloid CD38 contributes to endometrial NAD+ depletion, senescence, and lower receptivity in mice; its ablation improves the phenotype [23]. Therefore, a larger litter after systemic myeloid loss does not localize a benefit to the ovary. Reciprocal embryo transfers are required: experimental embryos into WT recipients for oocyte competence and WT embryos into perturbed recipients for receptivity.

8.5 Systemic and life course

CD38 acts in immunity, bone, muscle, liver, adipose, brain, and vasculature. Systemic inhibition may change healthspan without passing through the ovary, or preserve ovarian function while creating immune/endocrine costs. Mediation requires comparing localized versus systemic perturbation, measuring the hormonal trajectory, and choosing an organ-specific outcome. There is no obligatory global sign.

9. Computational layer

Decision and resources

Conventional bioinformatics was used because it could change the first experimental lineage. BioNeMo was omitted: embeddings or sequence models do not determine activity, topology, or contextual flux.

GSE232309: 8 animals, 14,349 QC cells; animal-level pseudobulk in the whole ovary [16].

• myeloid: −0.318 log2CPM; descriptive IC95% −0.829 to 0.193; exact p 0.171;
• endothelium: +0.519; IC95% 0.102 to 0.937; exact p 0.057; leave-one-out +0.415 to +0.644;
• stroma: −0.104; wide CI and unstable sign;
• myeloid proportion 2.87%→6.10%, ratio 2.13×;
• aggregate decomposition: 58.6% composition, 41.4% within lineages.

GSE236712: 3 pools per group, FACS CD45+CD11b+, 3 versus 14–17 months [17].

• Cd38 +1.018 log2CPM with the principal HTO gate;
• positive sign with gates 2/3/5 and in all leave-outs;
• exact p 0.10, the minimum two-sided value possible with 3+3;
• the deposit did not allow localization of the result to Mac_1–Mac_5 subtypes without inventing annotation.

Interpretation

The operational priority at mid-age is endothelium first, with a parallel myeloid screen and explicit composition control. The “early endothelium→late myeloid” pattern is exploratory: it combines different resources and does not identify a relay. Inherited human atlases did not show a positive, reproducible CD38−BST1 immune induction, which forbids promoting the murine signal to a human mechanism.

10. Primary hypothesis

L9-4-ED-H13R v2 — bulk amplifier without demonstrated follicular transmission

Falsifiable statement: during murine ovarian aging, injury/atresia and compositional remodeling increase the total contribution of CD38-high cells; CD38 materially lowers niche NAD+ as a secondary amplifier, but an adult lineage-specific loss will leave newly synthesized NAD+ in granulosa/oocyte and an independent follicular function within equivalence when basal injury and initial reserve are matched.

Parents: L9-4-AR-H13R v1; L9-4-COMP-C2; L9-4-AR-C2–C4; L9-4-ED-C1/C2.

Predictions:

1. Atresia, fibrosis, or follicular loss will precede the increase in total CD38 activity.
2. Absolute number of CD38-high cells will explain more of the total change than per-cell activity.
3. Adult deletion will raise bulk/local NAD+, but not F_granulosa nor function beyond the margin.
4. Germline KO will retain an early advantage not reproduced by adult intervention.
5. Perturbing salvage/transport or another consumer will move follicular flux more than CD38 at late ages.

Evidence for: compositional component, atretic signal, P2 phenotype, late NAD+ convergence, lack of tracing, and lack of mediation.

Evidence against: KO/78c control the pool and Yang reports favorable outcomes.

Kill criteria: two orthogonal adult perturbations plus WT/E226Q rescue show that CD38 precedes injury, changes newly synthesized follicular NAD+ above MDC/margin, and improves a function in two cohorts.

Status/maturity: strengthened_primary; descriptive H2, H1 for separate animal links, integrated H0. Priority confidence 0.62.

11. Competing hypothesis

L9-4-ED-H12R v2 — causal niche siphon with lineage direction

Falsifiable statement: catalytically active CD38 in an adult extrafollicular lineage—endothelial first at mid-age and/or myeloid later—alters the NAD+/NMN–ADPR/cADPR metabolome before atresia, reduces newly synthesized NAD+ in granulosa, and causally contributes to a functional loss; if an E→M relay exists, removing endothelial CD38 before 9 months will specifically reduce later myeloid activity.

Parents: L9-4-AR-H12R v1; L9-4-COMP-H11 v1; L9-4-ED-C3/C4.

Predictions:

1. In a single 3/9/15-month cohort, endothelial surface and NADase will increase before myeloid.
2. Endothelial loss will raise F_granulosa and reduce myeloid activity before changing atresia/composition.
3. WT rescue, not E226Q, will restore catalysis, metabolites, and transmission.
4. Restoring follicular NAD+ will recapitulate function if substrate is the mediator.
5. If cADPR–Ca²+ dominates, product/waveform will change first and Ca²+ rescue will outperform NAD+ rescue.

Evidence for: extrafollicular anatomy, CD38 dependence of the pool, endothelial 3→9 signal, and late myeloid signal.

Evidence against: nonlongitudinal datasets, modest endothelial effect, composition, RNA–activity decoupling, late convergence, and absence of human replication.

Kill criteria: endothelial activity does not precede; lineage-specific loss does not change flux into granulosa despite ≥80% inhibition; the change is explained by injury/composition; or WT/E226Q does not discriminate.

Status/maturity: weakened_reframed; integrated H0. Confidence 0.15 for an E→M relay; higher for a non-directional local amplifier.

Mandatory internal rival: L9-4-ED-H16 v2, common cause U→{endothelium,myeloid}. Fibrosis, hypoxia, SASP, or endocrine state can activate both in parallel. It is favored if each loss changes only its own pool and a U perturbation normalizes both.

Mediator rival: L9-4-ED-H14R v2, cADPR–Ca²+ conditioned on proximal evidence. It is not opened if the perturbation does not change products and waveform before function.

12. Translational hypothesis

Family L9-4-ED-HT2A v2 / L9-4-ED-HT2Q v2 — chemical species and occlusion will decide mechanism

Falsifiable statement: at free species and exposures compatible with human biology, apigenin and quercetin will retain a material fraction of their effects after CD38 loss and will not qualify as ovary-specific probes; the only plausible exception is quercetin if an inflammatory niche deconjugates its glucuronides, generates sufficient aglycone, and lowers CD38 product before function.

Shared predictions:

1. Majority conjugates will have little target engagement without deconjugation.
2. Total or nominal concentration will not predict activity; the free tissue fraction will.
3. A residual effect ≥20% without CD38 will follow NLRP3, PI3K–AKT–FOXO3A, redox, Ca²+, or another path.
4. Only proximal TE, ≥80% occlusion, and WT/not-E226Q rescue will attribute CD38.

Apigenin arm: residual anti-inflammatory/Ca²+ effect without CD38 is expected, already shown in macrophages [20]. Anti-IL-1β is not engagement.

Quercetin arm: a gap is expected between conjugated oral exposure and micromolar ex vivo effect. The exception requires ovarian β-glucuronidase/sulfatase, pH, temporal aglycone, a drop in ADPR/cADPR, and loss of the effect when deconjugation is blocked.

Kill of the non-specificity prediction: a plausible free species reduces CD38 activity/product, ≥80% of the function is occluded, it reappears with WT but not E226Q, and the non-CD38 residue stays within equivalence.

Kill of the quercetin exception: local aglycone does not appear, or blocking deconjugation does not change TE and function.

Status/maturity: parked, ovarian H0. Confidence of lack of specificity 0.72 for apigenin and 0.60 for quercetin; deconjugation exception 0.10. HUMAN_QA_REQUIRED before any study in people.

13. Falsifiable predictions and kill criteria

Provisional quantitative rule: with valid Q0 and ≥80% catalytic reduction, kill the adult follicular siphon if the F_granulosa change is smaller than max(MDC95, 10% of the positive-control range) and the CI lies within equivalence. The 10/20/80% thresholds are decision gates that must be anchored to metrology and a positive control; they are not biological constants.

14. Discriminating experiment

Q0 — metrological gate

Qualify LC–MS/MS isotope-dilution for NAD+, NADH, NMN, NAM, ADPR, and cADPR; kinetics with native substrate; surface/topology CD38; cytosolic/mitochondrial Ca²+; viability; and free fraction of compounds. Require recovery 80–120%, CV ≤15%, carryover <20% of LLOQ, stability ±15%, and WT–Cd38−/−–E226Q separation. If cADPR remains below LLOQ, H14R is not decided.

D0A — SAME-COHORT-ACTIVITY

Littermate females at 3, 9, and 15 months; confirmed diestrus for cycling animals and acyclicity as a stratum. One ovary for stereology, atresia, fibrosis, hypoxia, and SASP; the contralateral for absolute counts by beads, endothelial/myeloid/stromal sorting, RNA, surface, NADase/cyclase, and metabolites. Primary: age×lineage interaction on activity per viable cell. Start n=10/age, single blinded adaptation to 8–16 according to variance and margin.

This phase decides whether the transcriptomic pattern exists at the catalytic level and whether the total change comes from cell number or intrinsic activity.

D0B — ADULT-DIRECTIONAL-LOSS

Compare postmaturity inducible endothelial deletion Cdh5(PAC)-CreERT2;Cd38fl/fl and myeloid Cx3cr1-CreERT2;Cd38fl/fl, with littermate controls, tamoxifen in all groups, and a prior reporter. Require ≥80% on-target recombination and ≤10% in other ovarian lineages.

Coprimaries:

1. CD38 activity of the opposite lineage;
2. F_granulosa from [U-13C]NMN and [amide-15N]NAM pulses at 0.5, 2, and 6 h.

The animal is the unit. Pilot n=6/group; confirmatory n=10–16/group for 90% power of a material change or equivalence, adjusted once by blinded variance.

D0C — CATALYSIS-TRANSMISSION-MEDIATION

Only with positive D0B: control, adult loss, CD38 WT rescue, and E226Q, each with selective restoration of NAD+ or Ca²+ waveform. Match WT/E226Q surface/topology. Measure activity/products 0–2 h, F_granulosa and Ca²+ 0.5–6 h, mechanisms 6–24 h, and one prespecified function 24–72 h. Hierarchical model animal→slice→follicle.

D1 — functional localization

Only if D0C demonstrates transmission, mediation, and function:

• embryos derived from experimental oocytes into WT recipients;
• WT embryos into control/perturbed recipients;
• trajectory cohort with baseline, cycles, ovulation, pregnancies, litters, and zeros included.

Reciprocal transfer separates ovary from endometrium. More reserve counts only if it does not reduce throughput and improves live birth per transferred embryo.

T0 — chemical species and occlusion

Only after D0C/D1. Test apigenin and quercetin separately, authentic aglycones and conjugates, 0.03–10 µM measured free concentration, WT/loss/rescue, with a deconjugation module for quercetin. Hierarchical endpoint: free species → CD38 activity/product → occlusion/rescue → function.

Smallest real experiment

The minimum that changes the decision is D0A followed by proximal D0B. If there is no material lineage-specific activity or if deletion does not increase F_granulosa, the adult follicular siphon is killed without running D0C, reproduction, or compounds.

15. Biomarkers and stratification

There is no validated biomarker. The following are experimental candidate readouts:

• NADase/cyclase activity attributable to CD38 per cell and tissue;
• surface/topology CD38, not RNA alone;
• newly synthesized and absolute NAD+ in granulosa, with isotopic enrichment;
• NMN/NAM/ADPR/cADPR in medium and cells;
• cytosolic/mitochondrial Ca²+ waveform;
• absolute lineage counts, atresia, fibrosis, hypoxia, and SASP;
• follicular function, competence, and live birth with complete denominators.

Minimum strata: age and reproductive stage, phase/acyclicity, atretic burden, corpora lutea, pathology, hormonal exposure, inflammation, batch, and ischemic time. In humans, STRAW+10 and surgical/ART provenance will be modeled separately. AMH, blood NAD+, and menopause do not substitute for mechanism.

16. Individual variability

Magnitude and sign may vary by:

• reserve/endowment and recruitment rate;
• estrous phase or STRAW+10 status;
• immune/endothelial composition and fibrosis;
• inflammation, microbiota, metabolism, and the HPO axis;
• salvage, transport, PARP, and sirtuin activity;
• conjugation/deconjugation and protein binding of flavonoids;
• uterine function independent of the ovary;
• genetic background, colony, and ancestry.

Heterogeneity must not rescue a null primary post hoc. Modifiers are prespecified and evaluated after metrology is validated. Transportability to Mexico/LATAM remains unknown; European, ART, or murine magnitudes are not extrapolated.

17. Pharma relevance and maturity

The potential opportunity is not to develop apigenin or quercetin as “naturals.” If D0C and D1 were positive, the thesis would be reversible, spatially restricted modulation of CD38 NADase in a validated ovarian lineage, separating catalysis, Ca²+ signaling, immunity, and uterine receptivity.

Future comparable modalities only after target validation:

1. reversible small inhibitor with biased ovarian distribution;
2. antibody/biologic against ecto-NADase;
3. NADase/cyclase separation tool;
4. flavonoid scaffolds only if they pass exposure, selectivity, and occlusion.

Risks: CD38 may be a non-transmitting amplifier; the Ca²+ signal may change sign; chronic inhibition may compromise immunity; uterus and other organs may dominate outcomes; human exposure and therapeutic window are unknown; later menopause has trade-offs.

Go to early discovery: catalysis and transmission demonstrated, mediator identified, benefit localized to the ovary, proximal human bridge, and margin over immunity/uterus.

Reproductive no-go: trivial activity, bulk without follicle, follicle without function, dominant uterine/systemic effect, failed occlusion, or human bridge within equivalence.

Current maturity: H1 for the murine bulk pool; H2 for transcriptomic localization; H0 for adult ovarian target, integrated mechanism, translation, and Pharma. Not ready for partnering, a candidate, a human trial, or H5. HUMAN_QA_REQUIRED after D0C/D1/H0.

18. Limitations

1. Causal ovarian evidence is murine, global, or pharmacological; no published adult lineage-specific deletion exists.
2. There is no isotopic CD38→granulosa/oocyte flux nor human compartmental NAD+.
3. Reproductive studies differ in allele, colony, age, attrition, and metrics.
4. Computation uses 4+4 animals and 3+3 pools; the endothelial signal is modest and fails a high-coverage gate.
5. scRNA proportions are capture, not mass or activity.
6. Galligos 2026 has little biological replication and inferred communication; it does not validate CD38.
7. Apigenin/quercetin are pleiotropic and their ex vivo studies use direct micromolar concentration.
8. There is no free human ovarian PK, conjugate activity against CD38, or ovarian occlusion.
9. The cADPR–Ca²+ branch is not separated; a global clamp may not reproduce microdomains.
10. Reserve, competence, endocrine function, receptivity, healthspan, and lifespan are distinct estimands.
11. The female 78c study is small; a null does not demonstrate equivalence or safety.
12. Genetics of age at menopause does not simulate a localized adult intervention.
13. Absence of human evidence is search-sensitive; it does not prove universal nonexistence.

19. Conclusions

CD38 deserves to be called a controller of the murine ovarian NAD+ pool, not a demonstrated “oocyte siphon.” Its extrafollicular localization makes a niche mechanism plausible, but it also makes competitive the explanation that atresia and remodeling first increase the number or state of CD38-high cells. Computation suggests that cellular priority may change with stage, but it does not establish an endothelium→myeloid arrow.

The causal claim requires four separations: activity versus RNA, bulk versus follicle, NAD+ versus cADPR–Ca²+, and ovary versus uterus/system. The D0A→D0B sequence is the minimum test. A bulk NAD+ increase without F_granulosa or function would close the reproductive version even if it confirms local biochemistry.

Apigenin and quercetin are biochemical inhibitors and pleiotropic probes, not validated reproductive interventions. Their evaluation only makes sense after showing that CD38 controls flux and function, and must rest on free species, catalytic product, occlusion, and rescue. “Natural” does not inform selectivity or exposure.

Finally, more reproductive years are not equivalent to more female longevity. Any extension must localize organ-specific mediation, separate direct somatic action, and model competing risks. The concrete scientific contribution of L9-4 is to convert a rejuvenation narrative into a falsifiable sequence that can terminate early with an informative no-go.

20. References

1. Perrone R, et al. CD38 regulates ovarian function and fecundity via NAD+ metabolism. iScience. 2023. DOI: 10.1016/j.isci.2023.107949. PMID 37822499; PMCID PMC10562803.
2. Yang Q, et al. NADase CD38 is a key determinant of ovarian aging. Nature Aging. 2024. DOI: 10.1038/s43587-023-00532-9. PMID 38129670; PMCID PMC10798903.
3. Covarrubias AJ, et al. Senescent cells promote tissue NAD+ decline during ageing via activation of CD38+ macrophages. Nature Metabolism. 2020. DOI: 10.1038/s42255-020-00305-3. PMID 33199924; PMCID PMC7908681.
4. Chini CCS, et al. CD38 ecto-enzyme in immune cells is induced during aging and regulates NAD+ and NMN levels. Nature Metabolism. 2020. DOI: 10.1038/s42255-020-00298-z. PMID 33199925; PMCID PMC8752031.
5. Peclat TR, et al. CD38 inhibitor 78c increases mice lifespan and healthspan in a model of chronological aging. Aging Cell. 2022. DOI: 10.1111/acel.13589. PMID 35263032; PMCID PMC9009115.
6. Escande C, et al. Flavonoid apigenin is an inhibitor of the NAD+ase CD38: implications for cellular NAD+ metabolism, protein acetylation, and treatment of metabolic syndrome. Diabetes. 2013. DOI: 10.2337/db12-1139. PMID 23172919; PMCID PMC3609577.
7. Meyer H, et al. Human bioavailability study of apigenin after parsley. Annals of Nutrition and Metabolism. 2006. DOI: 10.1159/000090736. PMID 16407641.
8. Egert S, et al. Crossover exposure trial of quercetin in six young women. British Journal of Nutrition. 2012. DOI: 10.1017/S0007114511003242. PMID 21774840.
9. Wisetmongkolchai S, et al. Oral quercetin and IVF outcomes in aged female mice. 2026. DOI: 10.5935/1518-0557.20260024. PMID 42330175; PMCID PMC13318276.
10. Chantrasiri S, et al. Oral quercetin and oocyte/embryo outcomes in young female mice. 2024/2025. DOI: 10.5935/1518-0557.20240073. PMID 39540602.
11. Cao Y, et al. Quercetin protects aging oocytes by reducing oxidative stress through the SIRT3–SOD2 pathway. Cell Death & Disease. 2020. DOI: 10.1038/s41419-020-03183-5. PMID 33177495; PMCID PMC7658351.
12. Li J, et al. Quercetin in cyclophosphamide ovarian injury and PI3K–AKT–FOXO3A signaling. Reproductive Biology and Endocrinology. 2021. DOI: 10.1186/s12958-021-00743-y. PMID 33892727; PMCID PMC8063466.
13. Ruth KS, et al. Genetic insights into biological mechanisms governing human ovarian ageing. Nature. 2021. DOI: 10.1038/s41586-021-03779-7. PMID 34349265; PMCID PMC7611832.
14. Lankester J, et al. Age at menopause and health outcomes: Mendelian randomization. International Journal of Epidemiology. 2022. DOI: 10.1093/ije/dyac215. PMID 36409989; PMCID PMC10244052.
15. Singh AK, et al. Diminished NAD+ levels and activation of retrotransposons promote postovulatory aged oocyte death. Cell Death Discovery. 2024. DOI: 10.1038/s41420-024-01876-w.
16. Isola JVV, et al. Single-cell atlas of murine ovarian aging, GSE232309. Nature Aging. 2024. DOI: 10.1038/s43587-023-00552-5. PMID 38200272.
17. Zhang Z, et al. Young/aged murine ovarian myeloid atlas, GSE236712. Biology of Reproduction. 2025. DOI: 10.1093/biolre/ioaf122. PMID 40459236.
18. Galligos A, et al. Multicellular origins of murine ovarian inflammaging. Communications Biology. 2026. DOI: 10.1038/s42003-026-09826-1. PMID 41826404.
19. Yu X, et al. Apigenin in VCD-induced ovarian dysfunction and AKT/FOXO3A signaling. 2024. DOI: 10.1002/cbf.70015. PMID 39511926.
20. Lauritzen KH, et al. Apigenin inhibits NLRP3 inflammasome activation in monocytes and macrophages independently of CD38. Frontiers in Immunology. 2025. DOI: 10.3389/fimmu.2024.1497984. PMID 39840045; PMCID PMC11746122.
21. Kawai Y, et al. Macrophage-mediated deconjugation of quercetin glucuronide in inflammatory conditions. Journal of Biological Chemistry. 2008. DOI: 10.1074/jbc.M706571200.
22. Huang Y, et al. Age at natural menopause and all-cause mortality in the China Kadoorie Biobank. 2023. DOI: 10.1136/bmjph-2023-000332. PMID 40017861; PMCID PMC11816857.
23. Hua L, et al. Myeloid-derived CD38 mediates age-related endometrial aging through NAD+ depletion. Aging Cell. 2026;25:e70356. DOI: 10.1111/acel.70356. PMID 41457457; PMCID PMC12745663.
24. Munshi CB, et al. Mutagenesis of the catalytic Glu226 residue of human CD38. Journal of Biological Chemistry. 2000. DOI: 10.1074/jbc.M909365199.
25. Zhao Y, et al. E226Q as a catalytically inactive CD38 control and intracellular cADPR tool. Journal of Biological Chemistry. 2011. DOI: 10.1074/jbc.M111.228379.