Hydrogen gas (H₂), the lightest gas in the universe, can act as an important antioxidant. Supplementation of exogenous H₂ can improve many diseases. However, the physiological and pathological significance of endogenous H₂ is not clear. In the recent issue of Cell, McCurry et al. reveal that endogenous H₂ drives 21-dehydroxylation for transforming glucocorticoids to progestins, which may further influence female health, such as postpartum depression (1). We are very interested in this milestone study that reveals the physiological effects of endogenous H₂. Based on the history, here, we will discuss the process of this important finding, the uncovered questions, the current status, and the future of endogenous H₂ in female health.

In 1969, Eriksson et al. reported that the feces from germ-free (GF) rats contain corticoids but no pregnanolones, while the feces from conventional rats contain pregnanolones but no corticoids (21-hydroxylated steroids), which suggested that gastrointestinal progestin formation is dependent on gut microbiome (2, 3). Similar to this, it has been reported that fecal slurries from rats and humans can convert glucocorticoids into progestins through reductive removal of a C21 hydroxyl group, a process called 21-dehydroxylation (4, 5). Therefore, this means that the presence of 21-dehydroxylation products in the gut might depend on the gut microbiome. However, it is unclear how gut bacteria chemically modify steroids and how the resulting metabolites affect the host health.

In 2024, Dr. Megan D. McCurry et al. had developed an ultrahigh-performance liquid chromatography (UHPLC)-based method to quantify glucocorticoids and progestins in the biological fluids and tissues (1). Similar to H. Eriksson’s reports (2, 3), total levels of tetrahydroprogesterones (THPs) were significantly reduced in the feces of GF female mice compared with specific-pathogen-free (SPF) female mice (1). Then, they further identify physiologically relevant substrates for 21-dehydroxlation through targeted corticoid profiling on human bile and found that the average concentrations of tetrahydrodeoxycorticosterone (THDOC), e.g., 3α5αTHDOC and 3α5βTHDOC, in the abundant biliary corticoids were higher than other components (1). This indicated that the gut bacteria are exposed to physiologically relevant concentrations of 3α5αTHDOC and 3α5βTHDOC. Because 3α5αTHDOC levels in bile are higher in pregnant women, and its 21-dehydroxylated products have potential biological activities in the context of pregnancy and women’s health (6–8), they next focused on 3α5αTHDOC and hypothesized that 3α5αTHDOC can be 21-dehydroxylated into 3α5αTHP.

To verify their hypothesis, Megan D. McCurry cultured the feces of pregnant GF and SPF mice in the presence of 3α5αTHDOC and measured the levels of progestin, and found that a large amount of 3α5αTHP and 3β5αTHP were produced from pregnant SPF mouse fecal slurries, while fecal slurries from pregnant GF mice did not produce THPs (1). The culture pools with the human feces from healthy females and males 21-dehydroxylated 3α5αTHDOC, and moreover, the feces from pregnant human donors contained substantially higher levels of THPs than the feces from males and non-pregnant females. These indicate that both murine and human gut microbiome have the abilities to 21-dehydroxylate 3α5αTHDOC to produce progestins (1). However, why is there more 21-dehydroxylation product of 3α5αTHDOC in the context of pregnancy? One reason is that 3α5αTHDOC levels in bile are higher in pregnant women (7).

Dr. Megan D. McCurry therefore sought to isolate and characterize the 21-dehydroxylating species from human fecal microbial communities. In a culture pool isolated from a female donor, they found that Gordonibacter pamelaeae, a close relative of Eggerthella lenta, was the only bacterial species that function as active 21-dehydroxylators (1). Because the abundance of G. pamelaeae is low, they used an orthogonal assay and acquired the type strain of this bacterium, G. pamelaeae DSM 19378 (1). They found that G. pamelaeae DSM 19378 weakly 21-dehydroxylated 3α5αTHDOC to produce 3α5αTHP and 3β5αTHP, and this activity was significantly increased in co-culture with gut commensal Escherichia coli Nissle 1917 (EcN); however, EcN alone was unable to perform this transformation (1). They further tested the prevalence of 21-dehydroxylation activity in Eggerthellaceae family. In a 26-strain library of E. lenta and Gordonibacter species from the human gut, none of these strains performed 21-dehydroxylation in monoculture, and 12 strains produced varying levels of THP in co-culture with EcN (1). Thus, the members of the Eggerthellaceae family 21-dehydroxylate 3α5αTHDOC and indicate that this activity is induced by EcN (1).

What are the key mechanisms of induction of 21-dehydroxylation by E. coli? It has been reported that 21-dehydroxylation activity is inversely correlated with media redox potential (9). According to this, Dr. Megan D. McCurry first examined the hypothesis that E. coli lowers the redox potential and thus promoting 21-dehydroxylation. Their data showed that lower redox potential does not enable 21-dehydroxylation in E. lenta 14A (1). Second, they tested whether a 21-dehydroxylation-promoting factor was produced by EcN. The physical interactions between microbes can occur during cooperative metabolism; however, they found that contact between the two species is not required to induce 21-dehydroxylation (1). Third, they further tested whether EcN released extracellular inducing factors, and they found that EcN syringe-filtered supernatants but not vacuum-filtered supernatants induced 21-dehydroxylation in E. lenta (1). The basic difference between vacuum filtration and syringe filtration is that the dissolved gases are removed in the former. They therefore hypothesized that EcN may produce a gas that leads to 21-dehydroxylation in E. lenta. In healthy individuals, the colonic gases are composed of H₂, carbon dioxide (CO₂), methane (CH₄), nitrogen (N₂), and oxygen (O₂), as well as several odiferous trace gases, and the former three are produced solely by colonic microbes (10). Thus, which gas or gases could be responsible for 21-dehydroxylation?

It has been reported that E. coli can produce H₂ during its stationary phase of growth (11), which is when Dr. Megan D. McCurry started to observe substantial 21-dehydroxylation. Under anaerobic conditions, H₂ can be used by microbes as an electron donor to drive sulfate reduction and methanogenesis (12). Moreover, bile acid oxidation by E. lenta can be inhibited by H₂, and E. lenta performs reductive metabolic reactions under high H₂ partial pressure (13). Therefore, they hypothesized that H₂ might provide the reducing equivalents required for this highly reductive transformation.

Methylene blue (MB) is well known to react with an equimolar amount of H₂ in the presence of platinum (Pt) or palladium to produce colorless reduced MB (leucomethylene blue, leucoMB), as follows: MB blue + 2H⁺ + 2^e- → leucoMB colorless (14). Thus, Dr. Megan D. McCurry used MB with colloidal Pt as H₂ detection reagent based on the initial work by Tomoki Seo (14) and wildly used by our group (15) and others (16, 17). They found that E. lenta 14A grown under a H₂ condition, rather than a N₂ condition, performed 21-dehydroxylation. The H₂ levels and 21-dehydroxylation activities of E. coli with hydrogenase mutations co-cultured with E. lenta 14A were more than four times lower than those of E. coli BW25113 co-cultured with E. lenta 14A (1). In contrast to this, the H₂ levels and 21-dehydroxylation activities were not significantly influenced by disrupting a reductive pathway unrelated to H₂ production through loss of cysJ, the NADH:flavin oxidoreductase for the sulfite reductase in E. coli, when co-cultured with E. lenta 14A (1). Thus, H₂ produced by E. coli is a major driving force that promotes robust 21-dehydroxylation in E. lenta ( Figure 1 ). Beyond EcN 1917, the Gram-positive strains Clostridium scindens VPI 12780 and C. perfringens ATCC 13124, and the Gram-negative strains Citrobacter rodentium ATCC 8090 and C. freundii ICC 168, which express hydrogenase or hydrogenase homologs, can also induce 21-dehydroxylation in E. lenta 14A (1).

It should be noted that 21-dehydroxylation was not reduced to undetectable levels in co-cultures with H₂-deficient mutants, indicating that other factors produced by EcN may also contribute to this reaction (1). Their data also showed that there are low levels of H₂ produced in the co-culture of E. lenta 14A and E. coli with hydrogenase mutations (1). Thus, where do they come from? Indeed, E. lenta also contains annotated but unconfirmed hydrogenases, and they found that the E. lenta 14A monoculture can produce levels of H₂ comparable to those of the E. coli Nissle monoculture (1). In another independent experiment, Megan D. McCurry found that the E. lenta 14A monoculture and G. pamelaeae DSM 19743 monoculture produce the same levels of H₂ (1). 3α5αTHP and 3β5αTHP were only detected in the G. pamelaeae DSM 19743 monoculture, but not in the E. lenta 14A monoculture (1). This indicated that G. pamelaeae DSM 19743 may contain or produce other factors, rather than H₂, to induce 21-dehydroxylation. The synergistic H₂ production and 21-dehydroxylation of co-culture by E. lenta 14A and EcN 1917, and by G. pamelaeae DSM 19743 and EcN 1917, were higher than the monoculture of E. lenta 14A (ND) or G. pamelaeae DSM 19743 (1). Therefore, they concluded that higher H₂ produced in co-culture promoted 21-dehydroxylation activity.

Thus, which candidate genes could be involved in 21-dehydroxylation in E. lenta? They performed comparative genomics analysis on the genome sequences of producer and non-producer strains from the Actinobacteria strain library, and identified gene cluster Elen_2451–Elen_2454 as the candidate genes for 21-dehydroxylation (1). By co-culturing one of the top producer strains, E. lenta 14A, with EcN with or without tungstate, which is known to inhibit the oxidoreductase activity of Elen_2453, and by homologously and heterologously expressing the Elen_2451–2454 cluster, they confirmed that Elen_2451–2454 is responsible for 21-dehydroxylation in E. lenta (1). The feces from pregnant subjects are enriched in Elen_2451–2454 cluster-containing bacteria, and the concentrations of THPs positively correlated with the abundance of Elen_2451–2454 homologs (1). The feces from co-colonized female GF mice with producer strain E. lenta 14A and EcN contained significantly more THPs than female GF controls (1). Moreover, the co-colonized GF female mice with EcN and the cluster-containing strain E. lenta 14A, rather than the cluster-lacking strain E. lenta A2, produced higher levels of total THPs and 3α5βTHP in the feces (1). Elen_2451 is a formate dehydrogenase family accessory protein FdhD, Elen_2452 is 4Fe-4S ferredoxin iron-sulfur binding domain protein, Elen_2453 is molybdopterin oxidoreductase, and Elen_2454 is an SPFH domain band 7 family protein (1). However, we are still not clear about enzymatic mechanisms by which the above cluster regulate 21-dehydroxylation.

The ovarian hormone (progesterone and estrogen) levels increase over 100-fold during pregnancy, and consequently, neurosteroid allopregnanolone (also as brexanolone, 3α5αTHP) is also elevated in the brain (18, 19). The GABA_AR expression in the brain is reduced by neurosteroids during pregnancy to prevent sedation (18–20). Upon parturition, hormones rapidly return to pre-pregnancy levels, but the expression of GABA_AR may take time to recover (19). Therefore, a long-lasting mismatch between neurosteroid levels and GABA_AR numbers may underlie postpartum depression (19). The serum levels of allopregnanolone were detectable postpartum and were significantly decreased in women with maternity “blues” (8). Brexanolone is a first Food and Drug Administration (FDA)-indicated drug for postpartum depression (21). The gut bacteria both consume and produce H₂, and the remaining net H₂ in a fecal culture from donor F2 was sufficient to complete 21-dehydroxylation in co-culture and produced THPs (1). The THPs, such as 3α5αTHP, 3α5βTHP, 3β5αTHP, and 3β5βTHP, are two orders of magnitude higher in feces from pregnant people in the third trimester compared with the feces of male and nonpregnant female subjects (1). However, we are unclear about the changes in levels of gastrointestinal-derived THPs in pregnant women with postpartum depression before and after delivery. We do not know whether progestins produced by E. lenta remain in the gastrointestinal tract or are absorbed into enterohepatic or systemic circulation (1). Thus, it is not clear whether gut-derived THPs can affect postpartum depression ( Figure 1 ). Supplementation of exogenous H₂ protected against depression in mice (22), and if THPs can be absorbed into enterohepatic or systemic circulation and have effects on depression, it is difficult to distinguish the observed effects of exogenous H₂ on depression that are caused directly by H₂ or indirectly by produced THPs or by the synergistic effects of H₂ and THPs ( Figure 1 ).

Sex entails cutting-edge science but is bathed in mystery, and it is a fundamental pleasure and quality-of-life issue (23). H₂ has emerged as a novel bioactive molecule involved in modulating sexual organ homeostasis (23). Supplementation of exogenous H₂ can improve erectile dysfunction in a model of diabetic rats (24) and alleviate polycystic ovary syndrome (25), chemotherapy-induced ovarian injury (26), uterine inflammation (27), mastitis (28) and breast cancer (29, 30), and postmenopausal osteoporosis (31). It also has the effect of vasodilation, and attenuates chronic intermittent hypoxia-induced hypertension in rats (32, 33). However, the effects of endogenous H₂ on female health, such as post-menopausal endothelial dysfunction, osteoporosis after menopause, normal breast development and mastitis, ovarian function, uterine inflammation, vaginal microbiota, and clitoral erection are unclear ( Figure 1 ). To answer the effects of endogenous H₂ on female health, we should confirm biological distributions of endogenous H₂ (and its related metabolites, such as THPs) produced by microbiota, and analyze the changes of endogenous H₂ (and its related metabolites, such as THPs) and microbiota in blood and feces and organs/tissues before and after disease (or compare between normal individuals and patients). The animal models should be used to answer the causal relationship between endogenous H₂ and female health. Moreover, the strategies to reduce endogenous H₂ and increase endogenous H₂ should be employed. Reducing endogenous H₂ is typically achieved by the systematic use of antibiotics; however, the effects of antibiotics are broad and not specific, and both beneficial and harmful bacteria will be affected; therefore, this indiscriminate treatment cannot accurately answer the effects of endogenous H₂. After killing gut microbial community via antibiotics, it seems that supplementing with indicated strains can provide a relatively accurate answer to the physiological functions of indicated bacteria (34). We should note that it is not possible to have only one type of bacteria in the body, and the indicated bacteria can interact with one or more other bacteria or the host cells in multiple dimensions, thereby producing a range of physiological effects.

Although there are many challenges in H₂ medicine, especially endogenous H₂, Megan D. McCurry’s study provides a new paradigm for future research on H₂ medicine. Based on the expression profile of hydrogenase, we should investigate the physical and chemical essences of production and utilization of H₂ by human microorganisms and the physical and chemical essences of utilization of H₂ by the host cells and, thus, confirm the significance of endogenous H₂ for human homeostasis and pathogenesis of diseases.