Fertility protection during chemotherapy treatment by boosting the NAD(P)⁺ metabolome

The improved survival of cancer patients has led to an increasing need for strategies that protect ovarian function and the finite ovarian reserve from commonly used cytotoxic chemotherapy (Bertoldo et al, 2020b; Woodruff, 2009). While effective in oncology, these agents can lead to long-term adverse health outcomes for female cancer patients, including endocrine disruption, amenorrhea, premature ovarian insufficiency and infertility (Green et al, 2009a; Hudson, 2010; Morgan et al, 2012), with life-long health, psychological and social consequences (Gorman et al, 2015; Hudson et al, 2013). Aside from the devastating social and health consequences of infertility and its impacts on family planning (Bertoldo et al, 2020b), this can inflict damage to health through accelerated ovarian ageing and premature ovarian insufficiency, with early menopause symptoms due to estrogen insufficiency. In women of reproductive age or younger, options for preserving future fertility focus on the cryopreservation of reproductive material including oocytes, embryos, and in the case of child and adolescent cancer patients, ovarian tissue biopsies (Bertoldo et al, 2020b). These strategies must be undertaken prior to initiating chemotherapy, which may not be an option depending on the urgency of cancer treatment.

The best chance of maintaining future fertility options prior to chemotherapy treatment is the cryopreservation of oocytes or embryos following superovulation and IVF, however, this protocol can be too lengthy to delay chemotherapy and has a limited success rate. Moreover, superovulation followed by oocyte freezing is not suitable for childhood cancer patients (Deli et al, 2019). While these cryopreservation strategies provide options for the future ability to conceive genetic offspring, they do not prevent premature ovarian and endocrine failure which can lead to other long-term adverse outcomes including endocrine failure, osteoporosis, metabolic disease and neurocognitive disorders (Bruning et al, 1990; Jeanes et al, 2007). In addition, many cancer patients do not have timely access to the advanced medical care needed for such fertility preservation procedures (Letourneau et al, 2012b). Concurrent treatment of the GnRH agonist goserelin with chemotherapy provides some benefits to fertility preservation in breast cancer patients (Moore et al, 2015), however, these effects are moderate and the interpretation of these data is complicated by differences in the proportion of women seeking pregnancy (Oktay et al, 2015). What is needed is a non-invasive pharmacological strategy to protect ovarian function and fertility from the damaging effects of chemotherapy.

In addition to infertility, cancer survivors face an increased risk of chronic disorders including metabolic disease, cardiovascular disease, neurological disorders, bone disorders and cancers unrelated to the initial diagnosis (Hudson et al, 2013). Together, this diverse constellation of disorders resembles a form of biological ageing, presenting a challenge to maintaining the long-term health of this patient cohort. Several interventions previously shown to slow biological ageing have been shown to prevent the acceleration in ovarian decline during chemotherapy treatment. For example, the mTOR inhibitor rapamycin, which is a robust pharmacological intervention for extending lifespan in model organisms (Lamming et al, 2012), also prevents chemotherapy-induced infertility (Goldman et al, 2017). Similarly, inhibition of PI3K/Akt signalling, which extends lifespan in model organisms, also prevents chemotherapy-induced ovarian depletion (Kalich-Philosoph et al, 2013). These results indicate that other longevity interventions might also delay chemotherapy induced infertility, particularly if they have previously been shown to slow or reverse ovarian ageing. Recently, we (Aflatounian et al, 2022; Bertoldo et al, 2020a; Campbell et al, 2022; Habibalahi et al, 2022) and others (Huang et al, 2022; Miao et al, 2020; Yang et al, 2020) showed that declining female fertility during biological aging was in part due to a decline in levels of the prominent redox cofactor nicotinamide adenine dinucleotide (NAD⁺). Reversal of this age-associated decline using the NAD⁺ precursor nicotinamide mononucleotide (NMN) restored functional fertility, even when treatment was started at a post-reproductive age. More recently, others have found elevated expression of the NAD⁺ glycohydrolase enzyme CD38 in the ovary during reproductive ageing, and showed that blocking the degradation of NAD⁺ through small molecule inhibition of CD38 can prolong female reproductive lifespan (Perrone et al, 2023; Yang et al, 2024).

NAD⁺ plays another key role in slowing ovarian function by serving as a co-substrate of poly-ADP-ribose polymerase (PARP) enzymes in the DNA damage response (Durkacz et al, 1980), which is important to chemotherapy induced infertility (Nguyen et al, 2019). These enzymes can be among the greatest consumers of NAD⁺ in the cell, depleting cellular NAD⁺ reserves to the point of impacting cell survival (Fang et al, 2016). Chemotherapy can also induce infiltration into the ovaries of immune cells (Du et al, 2022), which can express the NAD⁺ consuming enzyme CD38, another determinant of ovarian ageing (Perrone et al, 2023; Yang et al, 2024). Further, many chemotherapy agents undergo drug metabolism by reductase enzymes that are fuelled by an NADPH cofactor, with NADP⁺ and NADPH levels dependent on the availability of NAD⁺ as a substrate for NAD⁺ kinase. Considering the role of NADPH-fuelled drug metabolism, NAD⁺-consuming PARP enzymes in DNA repair following chemotherapy, and recent findings around the role of NAD⁺ in reproductive ageing, we sought to test whether supporting NAD⁺ homeostasis could impact female infertility caused by chemotherapy treatment. To address this, we used a series of orthogonal approaches to test the role of NAD(P)⁺ homeostasis in chemotherapy induced infertility, including treatment with the NAD⁺ precursor NMN, along with transgenic overexpression of the NAD⁺ biosynthetic enzymes nicotinamide mononucleotide adenyltransferase 1 and 3 (NMNAT1 and NMNAT3). We find that these interventions can partially prevent a loss of ovarian function and functional fertility caused by the anthracycline chemotherapy drug doxorubicin (Dox), and importantly, that this protection against infertility does not come at the cost of reduced efficacy of chemotherapy against cancer cell lines in vitro or tumour models in vivo.

To test whether NAD(P)⁺ homeostasis is involved in chemotherapy-induced ovarian toxicity, we first used treatment with nicotinamide mononucleotide (NMN), an orally bioavailable metabolic precursor to NAD⁺ and NADP⁺ (Yoshino et al, 2011). In our experimental model, 8-week-old female C57BL6 mice were treated with or without chemotherapy, with or without the co-administration and ongoing treatment of NMN, for a 2 × 2 study design (Fig. 1A). Animals treated with NMN received a single injection at the time of chemotherapy, in addition to ongoing treatment with NMN in drinking water at 2 g/L, which started 72 h prior to chemotherapy and continued until the end of all experiments. These included measures of oocyte yield, ovarian reserve, follicle health and fertility. Functional readouts of ovarian function and fertility were from 8 weeks (56 days) following chemotherapy, by comparison, the minimum period needed for primordial follicles to develop to the antral stage or ovulation in mice has been estimated from between 20 days (Pedersen, 1970), 42 days (Oakberg, 1979) to 47 days (Zheng et al, 2014), with a recent estimate of 14–16 days for the primary follicle transition (Richard et al, 2023).

We first tested this in animals treated with the anthracycline chemotherapy drug, doxorubicin (Dox), which is associated with an increased risk of ovarian failure in patients (Green et al, 2009a; Green et al, 2009b; Letourneau et al, 2012a; Lipshultz et al, 2017) and which causes direct oocyte toxicity, ovarian toxicity and infertility in mice (Ben-Aharon et al, 2010; Lopes et al, 2020; Perez et al, 1997; Wang et al, 2019). Dox was used at 10 mg/kg, which was chosen based on previous work demonstrating lasting changes to the ovarian reserve (Ben-Aharon et al, 2010). Animals were treated with Dox and/or NMN, and 8 weeks later were stimulated with pregnant mare’s serum gonadotrophin (PMSG), with the number of germinal vesicle (GV) stage cumulus oocyte complexes (COCs) recovered from ovarian antral follicles used as a readout for oocyte yield (Fig. 1B). As expected, Dox caused a decrease in COC yield, however, this was rescued by NMN co-treatment (Fig. 1B). To test whether this was unique to this single chemotherapy agent, we also repeated the same experimental design in animals that were instead treated with cisplatin (CDDP, 5 mg/kg), which is also associated with ovarian toxicity (Bildik et al, 2015; Lopes et al, 2020; Nguyen et al, 2019). This dose was chosen based on previous work, which established a reduction in follicle counts and demonstrated DNA damage in adult mice (Nguyen et al, 2019). In addition to its ovarian toxicity, cisplatin was chosen due to its widespread use in cancer treatment, and the previously described impacts of platinum drugs on the NAD metabolome (Guan et al, 2017; Kim et al, 2014). As with the previous result, NMN co-treatment also protected against a loss in oocyte yield caused by cisplatin (Fig. 1C). To further validate these findings using an orthogonal approach, rather than NMN treatment, we next used transgenic mice that globally over-expressed the NAD⁺ biosynthetic enzymes NMNAT3 (Fig. 1D,E), which localises to mitochondria, and NMNAT1 (Fig. 1F), which localises to the nucleus (Yahata et al, 2009). These enzymes catalyse the final step of NAD⁺ biosynthesis, converting NMN into NAD⁺ (Fig. 1G). Nmnat3^Tg/+ animals were partially protected against a loss of oocytes caused by cisplatin, but not DOX (Fig. 1D,E), while oocyte numbers in Nmnat1^Tg/+ animals were protected against Dox (Fig. 1F). To assess the meiotic competence of these oocytes, we measured their meiotic progression through germinal vesicle breakdown (GVBD) and polar body extrusion (PBE) (Fig. EV1). While meiotic progression was always lowest in animals that received chemotherapy alone, this was not statistically significant, and we cannot conclude that chemotherapy impacts oocyte maturation. Regardless, chemotherapy, and its rescue by NMN, impacts follicle survival, as evidenced by COC yield (Fig. 1). Together, these experiments support the concept that NAD⁺ levels could be a target to maintain oocyte numbers in the context of chemotherapy treatment.

To test whether this increased COC yield translates to changes in fertility, two separate breeding trials were undertaken, where animals were treated with or without Dox and/or NMN (Fig. 2; (n = 10/group), or with or without cisplatin and/or NMN (Fig. 3; (n = 10/group), with the number of live offspring born used as readouts for functional fertility (Figs. 2 and 3). The Dox breeding trial (Fig. 2A) started at 8 weeks (56 days) and continued for over 250 days after chemotherapy, which is in excess of the ~47–50 days required for a full cycle of follicle development in sexually mature adult mice (Zheng et al, 2014), the uppermost limit from estimates that range down to 20 days (Oakberg, 1979; Pedersen, 1970). As expected, Dox caused a notable reduction in the overall number of pups born during the breeding trial (Fig. 2B–G), with NMN co-treatment with Dox partially protecting against this decrease in reproductive output. Figure 2B presents these data as the cumulative number of pups per group as a function of time, which is relevant to interpreting potential impacts on the primordial reserve versus the impacts on growing follicles, with similar data showing reproductive output from individual animals in each group shown in Fig. EV2A. Notably, using a cut-off for mating that took place more than 150 days following treatment (i.e. ~3 waves of complete folliculogenesis), four of ten Dox+NMN treated animals had pregnancies that yielded 26 pups, compared to two of ten from the Dox alone group, which yielded only 4 pups (Figs. 2C and EV2A). Further, these data are also shown as a function of cumulative breeding output over successive mating cycles (Fig. 2C), which were confirmed by the presence of a vaginal plug following timed mating which allowed correlation of successful copulation with live birth outcomes—individual data shown in Fig. EV2B. In line with the overall difference in the number of pups born throughout the entire breeding trial (Fig. 2B–D), the number of pups born per round of copulation per female (Fig. 2F) and the number of live pups per litter (Fig. 2G) were also partially rescued in animals that received both Dox and NMN, compared to those receiving Dox alone. Interestingly, the number of pups in each litter was increased by NMN treatment even in the absence of Dox treatment (Fig. 2G), when compared to the untreated controls, which is in line with our previous work (Bertoldo et al, 2020a). While this study used n = 10 per group, some panels in Fig. 2 and Fig. EV2 show less than 10 datapoints in each group, due to a failure to achieve mating or pregnancy—notably in the Dox alone group.

Given these results, we next extended these breeding trial studies to a previously described model of cisplatin induced infertility (Gonfloni et al, 2009; Kerr et al, 2012), where the cisplatin (CDDP) insult is delivered to prepubertal mice which have a high primordial follicle reserve. In this model, 1-week-old pups were treated with or without CDDP (2 mg/kg) alone, and 2 weeks later, weaned onto drinking water supplemented with or without NMN (2 g/L) (Fig. 3A). As expected, CDDP adversely affected fertility (Figs. 3B–G and EV2C,D). Overall, NMN similarly ameliorated fertility loss in animals treated with CDDP (Fig. 3) as observed in the doxorubicin trial (Fig. 2). Animals receiving CDDP followed by NMN had more total pups/dam (Fig. 3B,E), more pups/mating event (Fig. 3C,F), with larger litters (Fig. 3D,G), than animals receiving CDDP without NMN. As with the Dox breeding trial, animals receiving CDDP and NMN had improved fertility potential for an extended period. At >100 days after CDDP treatment, 100% (9/9) of animals that received CDDP followed by NMN had litters, yielding 107 pups. By comparison, only 22% (2/9) of animals receiving CDDP alone had litters beyond this timepoint, yielding only 6 pups (Figs. 3B and EV2C). At 150 days post-chemo, 5/9 from the CDP + NMN group had litters, yielding 32 pups, compared to zero litters from the CDDP alone group (Fig. EV2C). Notably, follicle maturation in mice occurs in two distinct waves, with the first wave, which lasts until ~3 months of age, maturing from primordial to antral stage in only 23 days, with the subsequent wave maturing in 47 days (Zheng et al, 2014). Unlike the previous breeding trial (Fig. 2) where Dox treatment was in adults, cisplatin was delivered at the prepubertal age of 7 days, during mobilisation of the first wave of follicles. These differences in breeding outcomes at 100 and 150 days represent around 3 and 4 cycles of follicle maturation, respectively. Further, cisplatin treatment was in prepubertal mice, when follicles are not yet growing. Together, this suggests that the benefits of NMN treatment are persistent, and are not restricted to the pool of follicles that were actively growing at the time of chemotherapy. Together, these data demonstrate protection of ovarian function by NMN in the face of chemotherapy, as evidenced by oocyte yield (Fig. 1) and improved fertility (Figs. 2 and 3).

The changes in COC yield (Fig. 1) and fertility (Figs. 2, 3 and EV2) are likely to be mediated at the level of the ovary, via treatment effects on ovarian reserve and/or follicle development. Chemotherapy has been described to deplete the ovarian reserve through both direct toxicity to oocytes (Nguyen et al, 2019), granulosa cell apoptosis (Lopes et al, 2020; Morgan et al, 2013; Yuksel et al, 2015) and follicle atresia leading to premature activation of the finite primordial follicle pool, leading to premature depletion of the ovarian reserve (Goldman et al, 2017; Kalich-Philosoph et al, 2013). To investigate effects on the ovarian reserve and follicle development, we treated animals with or without Dox +/− NMN (Fig. 4). Eight weeks later, animals were subjected to vaginal smearing for at least 10 days to track their cyclicity (Appendix Fig. S1). At diestrus, animals were euthanised and ovaries collected for detailed assessment of ovarian reserve and follicle health (Fig. 4). These ovaries were subjected to resin embedding, thick (20 µm) sectioning and stereology, with counting by a blinded investigator of all stages of follicle development in every third section across each entire ovary, including follicles that were assessed as healthy or unhealthy (Fig. 4). Morphological assessment included the classification of several unhealthy follicle types, described further in Methods, with examples shown in Fig. EV3, which are summed into a single “unhealthy” category. Briefly, unhealthy follicles included those which were atretic, with missing oocytes, zona pellucida remnants, enlarged oocytes with undifferentiated granulosa cells, multinuclear, vacuoles present, enlarged oocyte with undifferentiated GCs, biovular follicles, >10% pycnotic granulosa cells, damaged oocytes and others (Fig. EV3). As expected, Dox treatment caused a reduction in the primordial follicle reserve, which was not changed by NMN treatment (Fig. 4A). Of the remaining follicles that were assessed, there was, however, a stark difference in their health, with a four-fold increase in the number of unhealthy primordial follicles with Dox treatment (Fig. 4A). Notably, this increase in unhealthy primordial follicle numbers was restored by NMN co-treatment (Fig. 4A), with a similar, non-significant trend also observed in transitory (Fig. 4B) and primary (Fig. 4C) follicles, with a trend (p = 0.091) for an overall reduction in the number of unhealthy follicles of all stages across the entire ovary (Fig. 4D).

In addition to these differences in the health of the ovarian reserve, NMN co-treatment with Dox also led to an increase in the number of large antral follicles (Fig. 4E,F), with a similar trend in the number of corpora lutea (Fig. 4G). These findings would be in line with the increased oocyte yield observed in Fig. 1, as well as the increase in litter size observed in Figs. 2 and 3, and could suggest that NMN can improve the number of follicles that reach the antral stage and subsequent ovulation. The corpus luteum normally secretes progesterone, and while there was a trend towards reduced serum progesterone levels in animals treated with Dox only, this was not significant (Fig. 4H). Representative images of entire ovarian sections from each group are shown (Fig. 4I–L), and broadly reflect the blinded counts of antral follicle numbers. These representative images (Fig. 4I–L) suggest a reduction in overall size of the ovary following Dox treatment, which was also observed macroscopically (Fig. EV4A) and although the overall weights of ovaries showed a similar trend, this was not significant (Fig. EV4B). Given that NMN improved follicle health without impacting the ovarian reserve (Fig. 4), it is possible that the improved COC yield (Fig. 1) and fertility (Figs. 2 and 3) are due to its protection of granulosa cells and other somatic cell types in the ovary, which would be in line with previous findings on the toxicity of Dox towards these cells (Morgan et al, 2013).

For exogenous NMN to protect the ovary, one assumption is that it must be bioavailable and able to elevate NMN in this tissue. To assess this, we repeated dosing with NMN and Dox and collected ovaries at an acute timepoint of 6 h following treatment, which were subjected to targeted mass spectrometry analysis of the NAD⁺ metabolome (Fig. 5), including the metabolites nicotinamide (Fig. 5A), NMN (Fig. 5B), NAD⁺ (Fig. 5C), NADP⁺ (Fig. 5D), 1-methyl-nicotinamide (Fig. 5E), nicotinic acid adenine dinucleotide (Fig. 5F), NADH (Fig. 5G), and NADPH (Fig. 5H). The role of each of these metabolites in NAD⁺ biosynthesis is summarised in Fig. 5I. These data confirmed that exogenous NMN treatment elevated ovarian NMN, replenishing its levels following its depletion by Dox treatment (Fig. 5B). As expected, exogenous NMN treatment increased levels of the excretion product 1-methyl-nicotinamide (Fig. 5E), likely reflecting the excretion of excess NAD⁺ precursors. One rationale for this investigation was that cytotoxic chemotherapy drugs would deplete the NAD⁺ metabolome due to their activation of NAD⁺ consuming PARP enzymes. While there was a trend towards declining NAD⁺ levels with Dox treatment (Fig. 5C), this effect did not reach significance (p = 0.069), and nor was there an impact of either NMN or Dox treatment on the reduced equivalent NADH (Fig. 5G). Given the critical role of this cofactor in cell survival, it is likely that NAD⁺ levels are tightly defended by upregulating NAD⁺ biosynthesis, depleting levels of precursors such as NMN (Fig. 5B). There was a strong interaction (p = 0.01) between Dox and NMN treatment on NADPH (Fig. 5H), where a reduction in NADPH with Dox treatment was rescued by NMN co-treatment. This is consistent with well-characterised pathways for Dox metabolism, all of which are powered by an NADPH cofactor (Fig. 5J). This includes the one-electron reduction of Dox into its semiquinone form by P450 reductase enzymes, two-electron reduction into the alcohol metabolite doxorubicinol by carbonyl reductase enzymes, reduction into doxorubicin 7-deoxyaglycone by the action of cytosolic NADPH-dependent glycosidases, and hydrolysis by NADPH-dependent microsomal oxidoreductases into doxorubicin hydroxyaglycone (Fig. 5J). Every single one of these drug metabolites is dependent upon NADPH consumption, explaining its decline with Dox treatment. In addition to this, the reactive doxorubicin semiquinone radical can be recycled back to doxorubicin, generating an ^•O₂^- superoxide that leads to reactive H₂O₂ formation, which is neutralised into H₂O by the conversion of reduced glutathione (GSH) into its oxidised form (GSSG). The glutathione system is also likely to be active in this scenario to neutralise other sources of cellular ROS that are generated by Dox treatment. The regeneration of oxidised GSSG into reduced GSH by glutathione reductase is again powered by NADPH, further depleting its levels. Together, these known pathways for Dox detoxification would explain the reduction in NADPH levels (Fig. 5H). Given that NADP⁺ is generated by an NAD kinase that uses NAD⁺ as a substrate, providing exogenous NMN could allow for increased flux from tightly defended NAD⁺ levels into NADP(H) production (Fig. 5I). In line with this, Dox treatment and NMN treatment had opposing effects on ovarian NMN (Fig. 5B), with a reduction in NMN during Dox treatment and an increase in ovarian NMN with exogenous NMN treatment. This reduction in NMN with Dox (Fig. 5B) was most likely due to its flux into NADP(H) production, where NADPH is required by xenobiotic detoxification enzymes for the metabolism of Dox.

Although NADPH is required for the metabolism and eventual detoxification and excretion of Dox, the immediate products of NADPH-fuelled Dox metabolism include doxorubicinol (Fig. 5J), which is responsible for the cardiotoxicity of Dox (Olson et al, 1988). One conceptual risk of restoring NADPH (Fig. 5H) through providing NMN could be an elevated conversion of Dox into doxorubicinol and enhanced cardiotoxicity. To address this, we tested NMN in a model of Dox-induced cardiotoxicity in mice, where animals were co-treated with or without NMN in the same 2 × 2 design, and cardiac function was assessed by µ-ultrasound imaging (Fig. EV5). We observed no exacerbation of Dox-induced cardiotoxicity, with a trend towards reduced changes in cardiac function during NMN treatment (Fig. EV5), which is consistent with recent findings using other strategies to elevate NAD⁺ or prevent its breakdown by CD38 (Margier et al, 2022; Peclat et al, 2024; Zheng et al, 2019). Together, these data and other findings suggest that if NMN treatment leads to increased carbonyl reductase activity, this does not translate into an increased risk of cardiac dysfunction.

To obtain a broader view of molecular changes in ovarian function that could mediate these changes, we conducted label-free quantitative proteomics analysis (Fig. 6) of whole ovaries that were collected at six hours following doxorubicin, using the contralateral ovaries from animals used for metabolomics in Fig. 5. As expected, Dox treatment led to substantial changes in the proteome (Fig. 6A,B). Within Dox treatment, a subset of proteins were altered by NMN co-administration (Fig. 6C), normalising protein levels back to those of non-Dox treated animals—highlighted by cluster 4 of the heatmap (Fig. 6A), and select proteins shown in Fig. 6E–P. Importantly, acute NMN treatment on its own did not significantly alter any proteins when compared to untreated controls (Fig. 6D). This could suggest that any benefit provided by NMN treatment would occur in the context of rescuing a challenged state, such as from chemotherapy treatment or biological ageing (Bertoldo et al, 2020a), rather than altering the underlying baseline physiology of the ovary. Doxorubicin treatment led to striking reductions in proteins involved in ceramide biosynthesis, including ceramide synthase 5 (Cers5) and ceramide transporter 1 (Cert1), which were rescued by NMN co-treatment (Fig. 6G,H). Ceramide biosynthesis feeds into the production of sphingosine-1-phosphate (S1P), with the balance between ceramide and S1P levels acting as a determinant of apoptosis in oocytes (Morita et al, 2000). Notably, treatment with S1P can prevent the loss of primordial follicles caused by chemotherapy (Li et al, 2014) and irradiation (Paris et al, 2002). Ceramide biosynthesis involves the conversion of serine and palmitoyl-CoA to 3-dehydrosphinganine, which is a substrate for the enzyme 3-dehydrosphinganine reductase. This enzyme requires NADPH as a cofactor to produce sphinganine, which is the direct substrate of ceramide synthase (CERS) enzymes including ceramide synthase 5. Given that NADPH is reduced by Dox treatment (Fig. 5H) due to its likely consumption in Dox metabolism (Fig. 5J), one possibility is that decreased activity of the NADPH-dependent enzyme 3-dehydrosphinganine reductase leads to decreased levels of the CERS5 substrate sphinganine, with decreased substrate availability (Fig. 5C) resulting in downregulation of this enzyme (Fig. 6G). Similarly, decreased ceramide production could also lead to downregulation of its transporter, CERT1 (Fig. 6H).

Other proteins that changed with Dox alone included Cdc42 effector protein 5 (Fig. 6E), which may be important given the role of Cdc42 in meiotic resumption and cell polarity (Cui et al, 2007). The most profound change observed was in diphthamide biosynthesis 2 (DPH2), which was reduced by Dox but restored by NMN co-treatment (Fig. 6K). Diphthamide is an evolutionarily conserved post-translational modification that occurs at a single histidine residue of translation elongation factor 2 (EF-2), which is needed for accurate protein translation by preventing -1 frameshift “slippage” during translation (Liu et al, 2012). Diphthamide biosynthesis genes are essential to normal development, and deletion of these genes in mice leads to developmental defects including embryonic lethality, while mutations found in humans are associated with developmental delays, intellectual disability, craniofacial and genital abnormalities (Nakajima et al, 2018). It is unclear why Dox treatment reduced ovarian DPH2 levels, nor why its levels were altered by NMN co-treatment. Further work should aim to investigate changes in EF-2 activity and its diphthamide modification in the context of chemotherapy induced infertility.

Importantly, any strategy to counteract the toxicity of chemotherapy must avoid compromising the ability of chemotherapy to reduce tumour growth. Given the ability of NMN to protect against chemotherapy induced toxicity in the ovary, one concern could be that systemic NMN treatment would have the effect of promoting tumour growth and/or reducing the oncological anti-tumour efficacy of chemotherapy, limiting its clinical utility. This question is especially pertinent given the role of NAD⁺ in maintaining cancer cell metabolism (Navas et al, 2023; Tateishi et al, 2017), and the development of NAD⁺ biosynthetic inhibitors targeting nicotinamide phosphoribosyltransferase (NAMPT) to treat cancer (Tateishi et al, 2017; von Heideman et al, 2010). We therefore sought to test whether NMN co-administration with the chemotherapy agents doxorubicin and cisplatin would impair the ability of these compounds to slow tumour growth in mice bearing an orthotopic xenograft of the highly aggressive, invasive and poorly differentiated triple-negative breast cancer cell line MDA-MB-231 (Fig. 7A–H).

This tumour experiment was conducted to test for the non-inferiority of NMN in combination with chemotherapy, versus chemotherapy alone, and to ensure a robust finding was repeated in two independent labs, with data from these separated into Fig. 7A–D and Fig. 7E–H. Surprisingly, treatment with NMN alone reduced tumour growth (Fig. 7A,E), while co-administration of NMN with both doxorubicin (Fig. 7B,F) and cisplatin (Fig. 7C,G) did not reduce their efficacy. To further test for non-inferiority of NMN co-treatment with chemotherapy against chemotherapy alone in a wider range of chemotherapy agents across entire dose-response curves in other cancer models, we next turned to in vitro studies in a further six human cancer cell lines, which were derived from endometrial (HEC-1A, RL95-2), mammary (MCF-7, MDA-MB-231) and ovarian (A2780 and SK-OV-3) cancers (Fig. 7I). These studies showed that NMN did not interfere with the dose–response curves of the chemotherapy agents bleomycin, cisplatin, doxorubicin, etoposide, gemcitabine, methotrexate, paclitaxel, or vincristine, using in vitro cell viability (crystal violet staining) as a readout for their efficacy (Fig. 7I). The lack of effect of NMN on dose-response curves for chemotherapy further support the idea that NMN will not reduce the oncological efficacy of these drugs. Together, these in vivo and in vitro data suggest that an NAD⁺ precursor such as NMN may be safe to use in the context of cancer, as a strategy to reduce the impact of cytotoxic chemotherapy on female fertility, without impairing cancer treatment.

Together, these data use orthogonal pharmacologic and genetic approaches to show that strategies to elevate NAD⁺ can sustain ovarian function and fertility during chemotherapy. This has implications for the clinical management of female cancer patients of reproductive age or younger, including paediatric and adolescent cancer patients for whom ovarian stimulation and cryopreservation of oocytes or embryos prior to chemotherapy is not possible. These data are in line with previous findings around the role of NAD⁺ precursors in female fertility during biological ageing (Bertoldo et al, 2020a; Habibalahi et al, 2022; Huang et al, 2022; Miao et al, 2020; Yang et al, 2020), as well as the impact of NAD⁺ precursors in ameliorating the toxicity of chemotherapy towards other organ systems and in age-related disease (Li and Wu, 2021; Wu and Sinclair, 2016). These include impact of NAD⁺ precursors on chemotherapy induced peripheral neuropathy (Hamity et al, 2020), hearing loss (Kim et al, 2014), and renal toxicity (Guan et al, 2017). This work adds another adverse impact of chemotherapy treatment that could potentially be ameliorated using NAD⁺ precursors.

Interestingly, other work has investigated the impact of NMN treatment on infertility caused by chemotherapy and radiation treatment, concluding there was no protection from NMN (Stringer et al, 2020). One explanation for this discrepancy is that NAD⁺ precursors are likely to be most beneficial when NAD⁺ homeostasis is challenged by insults including a need for drug detoxification, activation of the DNA damage response and altered metabolic flux, as occurs with chemotherapy treatment and/or ageing, but not in unchallenged scenarios where there is no excess demand on the NAD⁺ metabolome. In the current study, NMN was provided concurrently with and following chemotherapy treatment, whereas the study by Stringer et al provided NMN to young, untreated animals for 7 days prior to irradiation or chemotherapy treatment, but not after (Stringer et al, 2020). When the NAD⁺ metabolome is replete and undisturbed, as in young healthy animals that have not received chemotherapy, exogenous NAD⁺ precursors are likely to be excreted as excess surplus, rather than accumulated. NAD⁺ metabolites are excreted through the urine following their conversion to 1-methyl-nicotinamide, and the concept that excess material would be rapidly excreted matches our own data, where NMN treatment drastically increased the production of 1-methyl-nicotinamide (Fig. 5E). Other differences between these studies were the choice of ovarian insult, with our study using doxorubicin and cisplatin, rather than cyclophosphamide and γ-irradiation, and the choice of timepoint, with our studies conducted at 2 months post-chemotherapy, rather than 5 days. Interestingly, the trend towards increased numbers of corpora lutea during co-treatment with NMN and doxorubicin (Fig. 4G) was also mirrored during NMN treatment and cyclophosphamide and γ-irradiation (Stringer et al, 2020), though neither result was statistically significantly different.

This retention of ovulatory function (Fig. 1) and fertility (Figs. 2 and 3) after chemotherapy was not caused by NMN preventing a loss of primordial follicles (Fig. 4A), but rather by reducing atresia in primordial and preantral follicles. We hypothesize that Dox+NMN treatment had both short- and long-term effects on folliculogenesis and fertility. In the wave of folliculogenesis occurring at the time of chemotherapy, co-treatment with NMN attenuated follicle atresia caused by Dox, leading to the survival of growing antral follicles as evidenced by the increased numbers of large antral follicles (Fig. 4F) containing viable COCs (Fig. 1B,E) which had the capacity to support pregnancies and development to term (Figs. 2 and 3). However, the Dox+NMN treatment also led to more pups than Dox alone in the subsequent waves of folliculogenesis (i.e. from 50 to 250 days post chemotherapy), which cannot be attributed to changes observed in the growing follicle population 8 weeks after treatment (Fig. 4). This prolonged benefit of NMN on fertility may have been caused by the protection of toxic effects of Dox on ovarian reserve health (Fig. 4A), or by the ongoing administration of NMN alone during this period affecting folliculogenesis and ovulation rate independent of Dox treatment (as per Fig. 2G), or alternatively NMN detoxifying effects of Dox on non-ovarian tissues or organs such as the uterus (Griffiths et al, 2020).

Interestingly, the impacts of NMN treatment to fertility were not restricted to delivery at the time of chemotherapy. We used a previously described model of cisplatin induced infertility (Gonfloni et al, 2009; Kerr et al, 2012), whereby cisplatin is delivered to prepubertal mice (Fig. 3A), when the primordial reserve is highest. As NMN is delivered in the drinking water, treatment was not possible in lactating animals, and these animals only started NMN treatment at weaning, 2 weeks after the cisplatin insult. Despite NMN treatment not starting until 2 weeks after cisplatin, we still observed improved fertility (Fig. 3), suggesting that the benefits of NMN may be related to ongoing repair processes in the ovarian reserve. The mechanism for this finding should be subject to follow-up study.

The difference in fertility (Figs. 2 and 3) and follicle health with Dox+NMN treatment may be related to supporting levels of NADPH (Fig. 5H), which is required to regenerate the cellular antioxidant glutathione in the face of ROS generation from Dox treatment. While these measurements were taken in the context of Dox treatment, it is worth noting that the same strategies also protected against decreased oocyte yields (Fig. 1E,F) and fertility (Fig. 3) caused by treatment with cisplatin. As with doxorubicin, cisplatin treatment also results in ROS generation, which is neutralised by the NADPH-powered glutathione system, warranting investigation of this strategy for other forms of chemotherapy which are detoxified via this system. Together, this rescue in follicle health may provide a physiological mechanism by which NAD(P)⁺ repletion protects against the fertility destroying effects of chemotherapy. In this scenario, the flux of metabolic precursors through to NADPH formation could be rate limiting to follicle health during chemotherapy. One question is whether the availability of NAD⁺ precursors or the expression and activity of NAD⁺ biosynthetic enzymes is more important to this flux. Overexpression of the NAD biosynthetic enzymes NMNAT1 and NMNAT3 provided some protection against a loss in COC yield (Fig. 1C,D,F), however, the effect size was less robust than for NMN treatment (Fig. 1B,E). The direct substrate for NMNAT1 and NMNAT3 is NMN, which was decreased by Dox treatment (Fig. 1B), and it may be that declining substrate availability is rate-limiting to NAD(P)⁺ flux, providing a mechanism for the impact of NMN treatment on ovarian function and fertility. One other possibility for altered NAD⁺ flux during chemotherapy treatment could be altered expression of the NAD⁺ consuming enzyme CD38 (Perrone et al, 2023; Yang et al, 2024), the increased expression of which is a cause of doxorubicin induced cardiotoxicity (Peclat et al, 2024), however, our proteomics data (Fig. 6) did not show any change in the levels of this enzyme with Dox treatment. Another potential mechanism could be the role of NAD⁺ consuming PARP enzymes which are involved in repairing DNA damage, which could be induced by chemotherapy. Our in vitro experiments with PARP inhibitors were inconclusive, and future work should aim to investigate this in more detail.

Although our observations would be consistent with a model whereby repletion of the NAD(P)⁺ metabolome could provide a buffer to replenish NADPH levels in the ovary, further work will be needed to determine the extent to which these systems are relevant to the protection against ovarian dysfunction and infertility. The metabolism of xenobiotics such as doxorubicin primarily occurs in the liver, however we readily detected the expression of NADPH-dependent enzymes that carry out doxorubicin metabolism in the proteomes of these ovary samples (Fig. 6). Although we cannot use raw peptide intensity values to compare the abundance of different protein species, it is notable that we could readily detect high numbers of peptides for NADPH-dependent carbonyl reductase 1 (CBR1) (supplementary files). This is consistent with observations that CBR1 is highly abundant in the ovary, composing 1–4% of all cytosolic protein (Iwata et al, 1990b), where its activity fluctuates during the ovarian cycle (Espey et al, 2000; Iwata et al, 1990a). The activity of this enzyme in the ovary could be a double-edged sword: on the one hand, the NADPH-dependent metabolism of DOX into doxorubicinol by CBR1 allows for the subsequent conjugation and excretion of this drug, and in doing so could deplete NADPH levels needed for the regeneration of GSH needed to neutralise ROS. On the other hand, doxorubicinol is the metabolite of Dox responsible for cardiomyopathy, though this occurs through its interaction with Ca²⁺ channels (Olson et al, 1988) which may be less relevant to ovarian function. Further, we found that NMN co-treatment with Dox did not increase cardiac dysfunction, which is in line with other recent work testing the impact of administering NAD⁺ precursors, or inhibiting the NAD⁺ consuming enzyme CD38 (Margier et al, 2022; Peclat et al, 2024; Zheng et al, 2019). Future work should aim to test this concept by comparing the levels of chemotherapy metabolites in ovaries from animals treated with or without NMN. Finally, it will be important to expand this investigation to a broader list of chemotherapy agents and to avoid extrapolating these findings beyond doxorubicin, the agent which was primarily used in these studies. While cytotoxic chemotherapy agents remain a mainstay of oncology, it is likely that profoundly different mechanisms of action will limit the utility of NAD⁺ boosting strategies to protect against infertility.

The observation of reduced tumour growth with NMN alone (Fig. 7) was surprising, as this experiment was originally intended to measure non-inferiority of NMN co-treatment compared to chemotherapy alone. Further work is required to understand how this occurred. It is likely that the tumour response to elevated NAD⁺ is likely to vary by mutation, with caution warranted. As these xenograft studies were in immunocompromised animals, enhanced immune surveillance is an unlikely mechanism. One hypothesis is that increasing NAD⁺ availability impacts the metabolism of cancer cells, resulting in a shift away from Warburg-type glycolytic metabolism towards oxidative phosphorylation, as we proposed previously (Wu et al, 2014). Although in vivo treatment suggested that NMN could have an impact on tumour growth, in vitro experiments with a matrix of chemotherapy and cancer cell lines showed no impact of NMN on cell survival—suggesting that any impact may be due to altered growth kinetics, rather than interfering with the efficacy of chemotherapy. Regardless, these non-inferiority data provide encouraging early evidence that NAD⁺ raising therapies may be safe to use in cancer patients, though the heterogeneity of tumour types means that these studies should be interpreted with caution, and additional studies are needed in different cancer models. Our observation of enhanced chemotherapy action towards tumours due to NMN co-treatment is in line with previous findings for nicotinamide riboside (NR) and paclitaxel treatment (Hamity et al, 2020). More recently, Jiang et al also showed that NMN treatment could slow tumour growth and metastases in a similar triple-negative breast cancer orthotopic xenograft tumour model (Jiang et al, 2023), again supporting our findings (Fig. 7).

There were several limitations of our study. As a model for chemotherapy induced infertility we used two chemotherapy agents (doxorubicin and cisplatin), testing three interventions (NMN treatment, NMNAT1-Tg and NMNAT3-Tg overexpression) and using three outcomes for fertility (COC yield, breeding, ovarian histology). It would have been ideal to test every permutation of these insults, interventions and outcomes, however, we were unfortunately limited by animal availability and cost. These experiments could be conducted in future work, along with testing a wider range of chemotherapy interventions, including clinically relevant chemotherapy combinations. Future work should also aim to test a wider range of chemotherapy doses: it is likely that this intervention will shift the dose–response curve of ovarian toxicity, and while our work showed a promising reduction in infertility against previously described gonadotoxic doses (Ben-Aharon et al, 2010; Gonfloni et al, 2009; Kerr et al, 2012; Nguyen et al, 2019), it is unclear whether this protection will translate to higher doses. Another aspect of these experiments that requires further work is in the timing of treatment: we exposed animals to NMN before, during, and long after the chemotherapy insult. While the metabolomics (Fig. 5) and proteomics (Fig. 6) data may provide interesting hints as to potential mechanism, these samples were taken at an early timepoint only 5 h after Dox treatment. This period of treatment might not be relevant, and it may be the case that the protection offered by NMN is a long-term, chronic effect that takes place over weeks of treatment. If so, those molecular changes (Figs. 5 and 6) may be irrelevant to the mechanism of action for NMN treatment, and future work should aim to narrow the window of when NMN treatment offers its protection, to help identify when these molecular changes might be relevant to this phenotype. Metabolomics data can have a wide data distribution, and the experiments in Fig. 5 are likely under-powered: once the timeframe of efficacy for NMN treatment can be established, these experiments could be re-examined using higher experimental power. An important aspect for future work that we highlighted was that NAD⁺ consuming PARP enzymes could be triggered by chemotherapy in the ovary, leading to an NAD⁺ decline. While there was a trend towards declining NAD⁺ levels with chemotherapy (Fig. 5C), future experiments should aim to clearly test this possibility, ideally through in vitro treatment with small molecule PARP inhibitors. To better elucidate a potential mechanism that would explain our findings, single-cell RNA sequencing should be considered for future work. To complement this, in situ immunohistochemistry staining for DNA damage and apoptosis markers in ovaries could, at a minimum, provide hints around which cell types are likely to contribute to the observed phenotype. We attempted these studies, but observed considerable variability in staining patterns and intensity within treatment groups, and no treatment effects—as described above, this may relate to an inappropriate selection of timepoints, or other important parameters. Our inability to precisely identify both the molecular mechanism and the relevant cell type(s) that mediate this phenotype are a weakness of this study, and future work should aim to address this in greater detail using in situ immunohistochemistry and single-cell sequencing.

These studies reveal that NAD⁺-raising compounds represent an effective pharmacological approach to reduce female infertility caused by cancer treatment. This has immediate implications for the clinical management of cancer patients by alleviating the risk of ovarian failure from chemotherapy as a consideration in oncology, providing a potential alternative to cryopreservation protocols that require invasive superovulation and oocyte pick-up procedures, which can delay cancer treatment, and may have limited success rates (Bertoldo et al, 2020b; Woodruff, 2009). Here, we demonstrated that co-treatment with NMN during either doxorubicin or cisplatin treatment did not compromise the efficacy of these agents, and in the case of the MDA-MB-231 orthotopic xenograft, surprisingly, reduced tumour growth. This work represents a clinically tractable intervention to non-invasively preserve ovarian function and female fertility during chemotherapy without diminishing the efficacy of cancer treatment, potentially acting as a fertoprotective neoadjuvant agent (Woodruff, 2017) to improve the long-term health and quality of life of cancer survivors.

All raw data have been uploaded to our Mendeley data site (https://doi.org/10.17632/fs5y6ggskv.1) as .csv and .txt files, with annotated R scripts for processing data, statistical analysis and generating figures. For histological assessment of ovarian reserve, all sections were digitally scanned for follicle quantitation, however, each of these scans resulted in large files which have not been uploaded due to file size limits on open data servers. These have been maintained for long-term storage on UNSW servers and are available upon request.

The source data of this paper are collected in the following database record: biostudies:S-SCDT-10_1038-S44321-024-00119-w.

Wing-Hong Jonathan Ho: Conceptualization; Investigation. Maria B Marinova: Validation; Investigation. Dave R Listijono: Investigation. Michael J Bertoldo: Supervision; Investigation; Methodology. Dulama Richani: Data curation; Investigation; Methodology. Lynn-Jee Kim: Investigation. Amelia Brown: Investigation. Angelique H Riepsamen: Investigation. Safaa Cabot: Investigation. Emily R Frost: Writing—review and editing. Sonia Bustamante: Investigation; Methodology. Ling Zhong: Investigation; Methodology. Kaisa Selesniemi: Investigation. Derek Wong: Investigation. Romanthi Madawala: Investigation. Maria Marchante: Investigation. Dale M Goss: Investigation. Catherine Li: Investigation. Toshiyuki Araki: Resources. David J Livingston: Validation. Nigel Turner: Resources. David A Sinclair: Supervision; Writing—review and editing. Kirsty A Walters: Supervision; Validation; Methodology. Hayden A Homer: Conceptualization; Supervision; Funding acquisition; Methodology; Project administration; Writing—review and editing. Robert B Gilchrist: Conceptualization; Supervision; Funding acquisition; Methodology; Project administration; Writing—review and editing. Lindsay E Wu: Conceptualization; Data curation; Formal analysis; Supervision; Funding acquisition; Visualization; Writing—original draft; Project administration; Writing—review and editing.

Source data underlying figure panels in this paper may have individual authorship assigned. Where available, figure panel/source data authorship is listed in the following database record: biostudies:S-SCDT-10_1038-S44321-024-00119-w.

RBG is a consultant to City Fertility CHA Global and is a Scientific Advisory Board member to Cooper Surgical. MJB is currently an employee of Ferring Pharmaceuticals, which produce drugs used in reproductive medicine. LEW, HAH and DAS are co-founders, shareholders, directors and advisors of Jumpstart Fertility Inc, which was founded to develop NAD⁺ precursors for the treatment of age-associated female infertility. The salaries of MJB and DG were paid in part by contract research from Jumpstart Fertility to UNSW. KS was an employee of Jumpstart Fertility. DJL is an employee and shareholder in Metro International Biotech, which is developing NAD⁺ precursors as therapeutics for a range of conditions. LEW and DAS are advisors and shareholders in EdenRoc Sciences, the parent company of Metro Biotech NSW and Metro Biotech, and in Life Biosciences LLC and its daughter companies. DAS is also a founder, equity owner, advisor, director, consultant, investor and/or inventor on patents licensed Cohbar, Galilei Biosciences, Tally Health, EdenRoc Sciences, MetroBiotech, and Life Biosciences. He is an inventor on a patent application filed by Mayo Clinic and Harvard Medical School that has been licensed to Elysium Health. For a full list and details see https://genetics.med.harvard.edu/sinclair/. MMC is currently an employee of Gameto Inc., which is developing cell-based therapies for reproductive medicine.

The salary of LEW is supported from a Hevolution/American Federation for Aging Research (AFAR) New Investigator award. This work was supported by the National Health and Medical Research Council (NHMRC) of Australia, through grants APP1103689 and APP1122484 to LEW, DAS and HAH, APP1139763 to RBG, LEW and KAW, a Career Development Fellowship to LEW (APP1127821) and Senior Research and Investigator Fellowships to RBG (APP1117538 and APP2009940). The salary and experimental costs of MB and DMG working in the lab of RBG and LEW were partly supported by Jumpstart Fertility. KS was an employee of Jumpstart Fertility. Orthotopic xenograft tumour models were performed at external contract research organisations under funding from Jumpstart Fertility and Metro Biotech. We gratefully acknowledge assistance from the UNSW Biological Resource Centre and Tzong-Tyng Hung from the Biological Resource Imaging Laboratory (BRIL) of the Mark Wainwright Analytical Centre, UNSW, which is a member of the National Imaging Facility, and the Bioanalytical Mass Spectrometry Facility of the Mark Wainwright Analytical Centre at UNSW. We also wish to thank the Solina Chau Foundation and Mr Hejun (Steven) Zhang for their philanthropic support.