Muscle-derived myostatin is a major endocrine driver of follicle-stimulating hormone synthesis

Abstract

Myostatin is a paracrine myokine that regulates muscle mass in a variety of species, including humans. Here, we report a functional role for myostatin as an endocrine hormone directly promoting pituitary follicle-stimulating hormone (FSH) synthesis and thereby ovarian function. Previously, this FSH-stimulating role was attributed to other members of the transforming growth factor β family, the activins. The results both challenge activin’s eponymous role in FSH synthesis and establish an unexpected endocrine axis between skeletal muscle and the pituitary gland. The data also suggest that efforts to antagonize myostatin to increase muscle mass may have unintended consequences on fertility.

One-Sentence Summary:

Hormone synthesis and reproduction depend on crosstalk between skeletal muscle and the pituitary gland.

Follicle-stimulating hormone (FSH) is a glycoprotein produced by gonadotrope cells of the anterior pituitary gland. Though recently implicated in post-menopausal bone loss, weight gain, and cognitive decline (1–3), FSH’s best-known actions are in the gonads (4), where it regulates ovarian follicle growth and spermatogenesis (5–6). Synthesis and secretion of FSH and the related luteinizing hormone (LH) are stimulated by gonadotropin-releasing hormone (GnRH) from the brain. FSH and LH, in turn, stimulate gonadal sex steroid production. In response to FSH, the gonads also produce inhibins, TGFβ family ligands that feedback on pituitary gonadotropes to selectively suppress FSH (7).

Inhibins were initially purified from ovarian follicular fluid (8). They share an α subunit and have distinct β subunits, βA for inhibin A and βB for inhibin B. During the purification of inhibins, three additional proteins were serendipitously discovered. Two of these proteins, activin A and activin AB (9), are dimers of inhibin β subunits (βA-βA for activin A and βA-βB for activin AB) and stimulate FSH secretion. Activin B, a homodimer of inhibin βB subunits (encoded by the Inhbb gene), was not purified from follicular fluid. Activins stimulate FSH by binding and signaling through complexes of activin type II (ACVR2A and ACVR2B) and type I receptors (ACVR1B and ACVR1C) to drive transcription of the FSHβ subunit gene (Fshb) ((7), (10), and Fig. 1A). Inhibins are endogenous antagonists, competing for binding to activin type II receptors on pituitary gonadotropes (11, 12).

Fig. 1: FSH synthesis is activin-independent in mice.

(A) Schematic representation of the proposed mode of activin B signaling in gonadotropes. The Xs represent the activin B knockout (KO) and type I receptor, ACVR1B, knockout models examined in Figure 1. Figure created in BioRender.com (BioRender.com/j95f951). (B) Schematic of mouse models used to assess activin signaling in gonadotropes, including wild-type (WT), control (floxed alleles only), Inhbb global KO, gonadotrope-specific Inhbb cKO, and gonadotrope-specific Acvr1b cKO mice. Serum (C) FSH and (D) LH levels (measured by multiplex ELISA) in wild-type or Inhbb KO mice. Pituitary gene expression assessed by RT-qPCR in (E) male and (F) female WT or Inhbb KO mice. Serum (G) FSH and (H) LH levels (measured by ELISA) in control or gonadotrope-specific Inhbb cKO mice. Serum (I) FSH and (J) LH levels (multiplex ELISA) in control or gonadotrope-specific Acvr1b cKO mice. Pituitary gene expression in (K) male and (L) female control or Acvr1b cKO mice. Animals were euthanized at 8 to 10 weeks old. Females were sampled at 7 am on estrus morning. Bar heights are group means. Each circle represents an individual mouse. Rpl19 was used as a housekeeping gene in E, F, K, and L. * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001. ns, non-significant.

The third protein purified from follicular fluid was named follistatin, given its ability to inhibit FSH secretion (13). Follistatins sequester and neutralize activins and the related myostatin (also known as growth differentiation factor 8, GDF8) and GDF11 (14–15). Notably, myostatin and GDF11 bind activin type II receptors. They can also signal via activin type I receptors, but preferentially use the TGFβ type I receptor, TGFBR1 (16).

Activins are largely bound and inactivated by follistatins in blood (17). Therefore, it is currently thought that activins produced locally in the pituitary, and particularly activin B by gonadotropes, are the main drivers of FSH synthesis in vivo (18, 19). However, this model is based on few observations. First, the inhibin βB, but not βA subunit is expressed in rat gonadotrope cells (20). Second, activin B, but not activin A, neutralizing antibodies selectively suppress FSH synthesis and secretion by cultured rat pituitary cells (21). Other data challenge the necessity for activin B in FSH synthesis. The same activin A and B neutralizing antibodies used in rats did not affect FSH production in mouse pituitary cultures (22). Moreover, inhibin βB (Inhbb) knockout mice, which cannot make activin B, have elevated rather than the predicted reduction in FSH levels (24). Importantly, however, mice lacking activin type II receptors in gonadotropes are FSH-deficient (25).

Considering these observations, we asked whether another TGFβ ligand, in addition to or instead of autocrine/paracrine activin B, drives FSH production in vivo.

FSH production is activin B-independent in mice.

Consistent with an earlier report (24), we observed elevated FSH in male and female Inhbb knockout relative to wild-type mice (Fig. 1A–C). LH was also increased in Inhbb knockout males, but not females (Fig. 1D). These changes were associated with increased expression of pituitary Fshb mRNA in both sexes and LHβ (Lhb) in males (Fig. 1E–F). Inhbb mRNA levels were reduced in knockouts; pituitary inhibin βA (Inhba) expression was unchanged (Fig. 1E–F). Fshb mRNA levels were equivalent in pituitary cultures from Inhbb KOs and wild-type controls and were similarly suppressed by inhibin A or follistatin (Fig. S1A–B).

Gonadotrope-derived activin B was thought to be the principal driver of FSH synthesis (18, 20, 26). However, FSH and LH did not differ between control and gonadotrope-specific Inhbb knockouts (Inhbb cKO) (Fig. 1G–H). Litter sizes were normal in Inhbb cKO females (Fig. S1C). Inhbb, but not Fshb, Lhb, or Inhba mRNA levels were reduced in pituitaries of cKOs relative to controls (Fig. S1D). In contrast, global recombination of the floxed Inhbb allele generated mice with phenotypes observed in the original Inhbb knockout strain, including elevated FSH and eyelid malformations (Fig. S1E–F). Thus, recombination of floxed Inhbb generated a null allele. Collectively, these data indicate that gonadotrope-derived activin B is not required for FSH synthesis in mice. Elevated FSH in global Inhbb knockouts likely results from reduced gonadal inhibin B negative feedback.

FSH secretion is reduced in mice lacking the type I receptor, TGFBR1, in gonadotropes.

As mice lacking activin type II receptors in gonadotropes are FSH deficient (25), one or more TGFβ ligands that use these receptors are required for FSH synthesis in vivo. The Inhba subunit is mainly expressed in stem cells in the murine pituitary (27). It was, therefore, possible that activin A from these cells maintained FSH production in Inhbb knockouts. To address this possibility, we conditionally ablated the canonical activin type I receptor, ACVR1B (28), in gonadotropes (Fig. 1A–B, I–L). This would block signaling in these cells by activin A from any source. Serum FSH and LH, and pituitary Fshb and Lhb mRNA levels were unaltered in Acvr1b cKOs relative to controls (Fig. 1I–L). cKOs exhibited increased GnRH receptor (Gnrhr) expression (Fig. 1K–L), as similarly observed in mice lacking Acvr2b, Tgfbr3, or Smad4 in gonadotropes (25, 29, 30). In conditional Tgfbr3 knockouts, this increase did not alter GnRH sensitivity (29).

Paralleling our data in Acvr1b cKOs, activin A or B-stimulated FSH secretion was unaltered in pituitaries from mice lacking the other activin type I receptor, Acvr1c (31). In cultured wild-type pituitaries, however, a small molecule inhibitor of ACVR1B, ACVR1C, and TGFBR1 (SB431542) blocks Fshb expression (23). We therefore asked whether TGFBR1 plays a role in FSH synthesis, even though activins do not signal via this receptor (32, 33) and murine gonadotropes are unresponsive to TGFβ isoforms due to absence of the TGFβ type II receptor (TGFBR2) (Fig. S2A and (34)). Gonadotrope-specific ablation of Tgfbr1 led to decreases in FSH secretion in both males and females (Fig. 2A–B). LH was unaltered (Fig. 2C). Despite the reduced serum FSH, pituitary Fshb expression was not reduced in cKO mice (Fig. 2D–E). Pituitary Gnrhr expression was increased in Tgfbr1 cKOs (Fig. 2D–E). Consistent with their lower serum FSH, Tgfbr1 cKO females had fewer ovarian antral follicles and corpora lutea (Fig. 2F), and produced smaller litters than controls (Fig. 2G). Testis weight was decreased in Tgfbr1 cKO males (Fig. 2H). As revealed by snRNA-seq, Acvr1b expression was not significantly altered in gonadotropes of Tgfbr1 cKOs and vice versa (Fig. S2B–C).

Fig. 2: FSH production depends on the type I receptor, TGFBR1, in gonadotropes.

(A) Schematic representation of the gonadotrope-specific TGFBR1 knockout model used in Figure 2. Figure created in BioRender.com (BioRender.com/x21t595). (B) Serum FSH (Luminex assay), (C) LH (ELISA), and pituitary gene expression (RT-qPCR) in (D) male and (E) female control (gray) and gonadotrope-specific Tgfbr1 cKO (purple) mice. Rpl19 was used as a housekeeping gene in D and E. (F) Antral follicle (AF) and corpora lutea (CL) counts from ovarian sections of 9- to 10-week-old control and Tgfbr1 cKO females (one ovary per mouse). (G) Number of pups per litter in 6-month breeding trials of control and Tgfbr1 cKO females. (H) Testicular weights (normalized to body weight) of control and Tgfbr1 cKO males. Bar heights are group means. Each circle represents an individual mouse. Females were sampled at 7 am on estrus morning. * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001. ns, non-significant.

Mice lacking both ACVR1B and TGFBR1 in gonadotropes are FSH deficient.

ACVR2A and ACVR2B show some functional redundancy in murine gonadotropes (25). Therefore, the combined ablation of ACVR1B and TGFBR1 might similarly be needed to cause complete FSH deficiency (Fig. 3A). Indeed, FSH was virtually undetectable in gonadotrope-specific Acvr1b/Tgfbr1 knockout mice (double conditional knockouts or dcKOs) (Fig. 3B). As reported in other FSH-deficient models (25, 35), LH (Fig. 3C) was elevated in female but reduced in male dcKOs. Patterns of pituitary Fshb and Lhb expression mirrored the serum hormone levels (Fig. 3D–E). dcKO females were infertile (Fig. 3F), with reduced ovarian and uterine weights (Fig. 3G–H, S3A). Ovarian folliculogenesis was arrested at the early antral stage in dcKO females (Fig. 3I). Testicular, but not seminal vesicle weights were reduced in male dcKOs (Fig. S3B–C). Testis histology revealed mature spermatozoa in the seminiferous tubules of both genotypes (Fig. 3J). The efficacy and specificity of the gene knockouts was confirmed in purified gonadotropes (Fig. S3D–G). Collectively, these data indicate that FSH is regulated by a TGFβ ligand (or ligands) that signals preferentially via the type I receptor, TGFBR1, with some compensation by ACVR1B.

Fig. 3: Acvr1b/Tgfbr1 double cKO animals are FSH-deficient and hypogonadal.

(A) Schematic representation of the gonadotrope-specific Acvr1b/Tgfbr1 knockout model used in Figure 3. Figure created in BioRender (BioRender.com/s57z035). (B) Serum FSH and (C) LH levels (measured by ELISA), and pituitary gene expression (RT-qPCR) in (D) male and (E) female 9- to 10-week-old control (gray) and gonadotrope-specific Acvr1b/Tgfbr1 dcKO (maroon) mice. Control females were randomly cycling. Rpl19 was used as a housekeeping gene in D and E. (F) Number of pups per litter in 6-month breeding trials of control and Acvr1b/Tgfbr1 dcKO females. (G) Representative images of female reproductive tracts (top) and testes (bottom). Scale bar: 1 cm. (H) Ovarian weights (normalized to body weight). Bar heights are group means. Each circle represents an individual mouse. * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001. ns, non-significant. (I) Ovarian and (J) testicular histology sections stained with H&E. Boxed areas in the images at the left are expanded at the right. Scale bars: 200 μm.

Myostatin and GDF11 stimulate FSH synthesis in vitro.

The TGFβ family member(s) that regulates FSH must bind to ACVR2A and ACVR2B (25), signal via TGFBR1 and ACVR1B (Fig. 3B), and be antagonized by inhibins and follistatin (Fig. S1A–B). Two non-activin ligands that satisfy these criteria are myostatin and GDF11; however, neither was previously implicated in FSH regulation. Both myostatin and GDF11 stimulated murine Fshb promoter-luciferase (luc) activity in the murine immortalized gonadotrope-like cell line, LβT2 (Fig. 4A), as well as Fshb mRNA expression in murine pituitary cultures (Fig. S4A). The latter effects were antagonized by inhibin A (Fig. S4B). Myostatin and GDF11 actions in LβT2 cells depended on ACVR2A, as knocking down the receptor with a previously validated siRNA (23) impaired ligand-stimulated Fshb promoter activity (Fig. 4B–C). ACVR2A knockdown similarly blocked activin B action in these cells (Fig. S4C). We next validated siRNAs against Acvr1b and Tgfbr1 (Fig. S4D). Knockdown of ACVR1B attenuated activin B (Fig. S4C), but not myostatin or GDF11, induction of Fshb-luc in LβT2 cells (Fig. 4B–C). In contrast, TGFBR1 knockdown inhibited myostatin and GDF11, but not activin B, activity (Fig. 4B–C, S4C).

Fig. 4: Myostatin (MSTN) and GDF11 stimulate FSH synthesis.

(A) LβT2 cells were transfected with a murine Fshb promoter-luciferase (luc) reporter plasmid. Cells were then treated with medium or 0.5–2 nM GDF11 (orange) or MSTN (teal). Data represent mean ± SEM from 3 independent experiments (N=3). LβT2 cells were transfected with the Fshb-luc reporter and 5 nM of control, Acvr2a, Acvr1b, or Tgfbr1 siRNA. Cells were treated with medium or (B) GDF11 (1 nM) or (C) MSTN (2 nM). Bar heights are group means. Each circle represents an independent experiment. (D) Dot plots of Gdf11, Inhba, Inhbb, and Mstn expression in different cell lineages from snRNAseq of adult female and male murine pituitaries. (E) Schematic representation of the gonadotrope-specific Gdf11 and gonadotrope-specific Inhbb/Gdf11 double knockout mice used in Figure 4F and G. Figure created in BioRender.com (BioRender.com/o34c553). Serum FSH (measured by ELISA) in 9- to 10-week-old (F) control and Gdf11 cKO or (G) control and Inhbb/Gdf11 dcKO mice. Females were sampled at 7 am on estrus morning. Bar heights are group means. Each circle represents an individual mouse. (H) Schematic of intravenous (tail vein) injections of the indicated neutralizing antibodies in adult wild-type mice. (I) Percentage change in serum FSH levels 1, 2, 3, and 4 weeks following a single i.v. injection of mouse IgG or 5–20 mg/kg anti-MSTN/GDF11 antibody in adult male wild-type mice. The total amount of injected antibody was balanced to 20 mg/kg using mouse IgG. Data were normalized to FSH levels before injection for each animal. Data represent mean ± SEM (n=3/group). (J) Percentage change in serum FSH levels 2 or 7 days following a single i.v. injection of mouse IgG, anti-activin A, anti-MSTN, or anti-MSTN/GDF11 antibodies (10 mg/kg) in adult male wild-type mice. Data were normalized to FSH levels before injection. Data represent mean ± SEM (n=3/group). (K) Serum FSH levels (ELISA) and (L) number of eggs ovulated by female wild-type mice at natural ovulation following a single i.v. injection with the indicated neutralizing antibodies. Injected females were paired with wild-type males 7 days post-injection and samples collected on the morning of vaginal plugging. Bar heights are group means. Each circle represents an individual mouse. * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001. ns, non-significant.

As autocrine activin B was thought to regulate FSH, we asked whether there might be a role for gonadotrope-derived myostatin and/or GDF11. Single cell and single nucleus RNA-seq data sets revealed Gdf11 expression in the murine pituitary gland, including in gonadotrope cells (Fig. 4D and (27)). In contrast, myostatin (Mstn) was not detected in any pituitary cell type (Fig. 4D and (27)). We therefore conditionally ablated Gdf11 alone and in combination with Inhbb in gonadotropes in vivo (Fig. 4E). FSH was unaltered in Gdf11 cKO or Inhbb/Gdf11 dcKO mice (Fig. 4F–G). We confirmed the efficacy and specificity of the gene knockouts in purified gonadotropes from dcKO animals (Fig. S5A–D). Collectively, these data suggest that activin B, GDF11, or myostatin of gonadotrope origin are not required for FSH production in vivo.

Myostatin regulates FSH synthesis in vivo.

Though gonadotrope-derived GDF11 was ruled out in FSH regulation in vivo, this did not preclude a role for the ligand or for myostatin (MSTN) from extra-pituitary sources. To begin to address this possibility, we treated male wild-type mice with a MSTN/GDF11 neutralizing antibody (RK-35) (36) at 5, 10, or 20 mg/kg (Fig. 4H) and measured FSH for 4 weeks. After 1 week, FSH was reduced by ~75% at all three doses (Fig. 4I). Levels began to increase in a dose-dependent manner thereafter, with FSH returning to baseline at all three doses by 4 weeks (Fig. 4I). Next, we treated wild-type males with 10 mg/kg of the same MSTN/GDF11 antibody, a MSTN-specific antibody (RK-22), or an activin A neutralizing antibody (SW101) (36, 37) (Fig. 4H, 4J). We confirmed the efficacy and specificity of all three antibodies in vitro (Fig. S6A–C). Both the MSTN/GDF11 and MSTN antibodies reduced FSH in vivo by ~75% within 2 days and the hormone remained at this level at day 7 (Fig. 4J). Body composition was unaltered by the MSTN antibody (Fig. S7A–C). The activin A antibody produced a modest and short-lived suppression of FSH (Fig. 4J). Seven days after injections, FSH and pituitary Fshb mRNA levels were reduced in male mice treated with the MSTN/GDF11 or MSTN antibodies, whereas there was no effect of the activin A antibody (Fig. S8A–B). LH did not differ between groups (Fig. S8C).

Next, we treated female wild-type mice with 10 mg/kg of the same antibodies (36, 37) (Fig. 4H). After 7 days, females were paired with untreated wild-type males. On the morning of vaginal plugging (estrus morning; typically within 2–3 days of pairing), FSH and pituitary Fshb mRNA levels were reduced in females treated with the MSTN/GDF11 or MSTN antibody (Fig. 4K, S8D). Reduced FSH was associated with fewer eggs being ovulated (Fig. 4L). The activin A antibody did not affect FSH or pituitary Fshb but led to a greater number of eggs ovulated (Fig. 4K–L, S8D). The treatments did not affect LH or pituitary Lhb expression (Fig. S8E–F).

Muscle-derived myostatin stimulates FSH production.

Circulating myostatin is produced predominantly in skeletal muscle (38, 39). We therefore examined reproductive phenotypes in germline (global; Mstn KO) and muscle-specific Mstn knockout (cKO) mice (Fig. 5A). Both KO and cKO females exhibited delayed puberty onset (Fig. S9A–D) and lengthened estrous cycles (Fig. S9E–F). FSH and pituitary Fshb expression were reduced by about 50% in both male and female (estrus morning) (29) KOs and cKOs relative to controls (Fig. 5B–C, S10A–B). Pituitary Lhb, Prl, and Tshb levels were unchanged in Mstn KOs (Fig. S10C–E). Female KOs and cKOs ovulated approximately half the number of eggs in natural cycles and produced fewer pups per litter than wild-type/control mice (Fig. 5D–G). These knockout mice also had reduced ovarian weights (Fig. S10F). Male KOs exhibited a ~40% decrease in testis size, even when correcting for their increased body weight (38), and reduced intra-testicular testosterone (Fig. 5H and S10G–H). Mstn cKO animals were similarly hypogonadal (Fig. S10I–K). Serum myostatin was decreased by ~90% in Mstn cKOs (Fig. S11A–B). Though not apparent in control mice (Fig. S11C–D), myostatin and FSH levels were positively correlated in Mstn cKOs (Fig. S11E–F).

Fig. 5: Myostatin is a major driver of FSH synthesis and secretion in mice.

(A) Schematic representation of Mstn knockout or muscle-specific conditional KO (cKO) mice used in Figure 5B–K. Figure created in BioRender.com (BioRender.com/g88k582). (B) Serum FSH (ELISA) in 9- to 10-week-old WT/control (gray) and Mstn KO (teal) mice or (C) Mstn cKO (light blue) mice. Females were sampled at 7 am on estrus morning. (D) Number of eggs ovulated on estrus morning in 8- to 9-week-old WT/control and Mstn KO females or (E) Mstn cKO females. (F) Number of pups per litter in 3-month breeding trials of WT/control and Mstn KO or (G) Mstn cKO females. In B-G, bar heights are group means. Each circle represents an individual mouse. (H) Representative image of testes from a wild-type and a global Mstn KO mouse. Scale bar: 1 cm. (I) Schematic of intramuscular injections of control (AAV-control) or myostatin expressing adeno-associated viruses (AAV-MSTN) in Mstn KO mice. (J) Serum myostatin (MSTN) levels and (K) serum FSH levels (ELISA) 4 weeks after intramuscular injections of AAV-control or AAV-MSTN particles in 8-week-old male Mstn KO mice. Bar heights are group means. Each circle represents an individual mouse. (L) Schematic of i.v. injection of neutralizing antibodies in adult wild-type mice. (M) Percentage change in serum FSH levels 2 or 7 days following a single i.v. injection of human and mouse IgG, anti-activin A/B (15 mg/kg), and/or anti-MSTN/GDF11 antibodies (10 mg/kg) in adult male wild-type mice. The total amount of injected antibody was balanced across conditions using mouse IgG or human IgG. Data were normalized to FSH levels before injection for each mouse. Data represent mean ± SEM (n=3/group). (N) Pituitary Fshb expression (by RT-qPCR) in adult male mice one week after IgG or antibody injections. Bar heights are group means. Each circle represents an individual mouse. Rpl19 was used as a housekeeping gene. (O) Schematic of i.v. injection of antibodies in adult male Long-Evans rats (200–250 grams). (P) Percentage change in serum FSH levels 2 or 7 days following a single i.v. injection of mouse IgG or an anti-MSTN/GDF11 antibody (10 mg/kg) in male rats. Data were normalized to FSH levels before injection for each rat. Data represent mean ± SEM (n=3/group). * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001. ns, non-significant.

The data demonstrate that muscle-derived myostatin is necessary for quantitatively normal FSH production. We next tested its sufficiency by expressing myostatin from skeletal muscle of Mstn KO males using adeno-associated viral vectors (AAV-MSTN, Fig. 5I). We observed an increase in serum myostatin after 4 weeks, which was associated with a 65% increase in serum FSH compared to KO mice treated with a control AAV (Fig. 5J–K).

Myostatin is the most critical of four TGFβ ligands driving FSH synthesis in mice.

In contrast to Acvr2a/Acvr2b dcKO (25) or Acvr1b/Tgfbr1 dcKO mice (Fig. 3B), global and muscle-specific Mstn knockout mice produce some FSH (Fig. 5B–C). This suggested that one or more TGFβ ligands may act in myostatin’s absence. To assess GDF11’s role, we treated global Mstn KOs with the MSTN/GDF11 neutralizing antibody. This further reduced (by an additional ~50%) FSH and pituitary Fshb mRNA levels (Fig. S12A–C), as well as the number of eggs ovulated in females (Fig. S12D). As some FSH remained, we examined potential roles for the activins. The addition of an activin A neutralizing antibody to the MSTN/GDF11 antibody did not further reduce FSH or Fshb expression in wild-type male mice (Fig. S13A–B). An antibody with neutralizing activity against both activin A and B (REGN16430, validated in Fig. S14), however, nearly eliminated (~90% reduction) serum FSH and pituitary Fshb mRNA levels when used in combination with the MSTN/GDF11 antibody in wild-type males (Fig. 5L–N). Thus, in the absence of myostatin and GDF11, activin A and/or B is/are sufficient to maintain a small level of FSH production in mice.

Myostatin and GDF11 regulate FSH in rats.

The current concept that autocrine/paracrine activin B is the major driver of FSH synthesis derives principally from studies in rats (18, 19). It was therefore possible that myostatin’s role in mice might be species-specific. We therefore treated adult male rats with 10 mg/kg of the MSTN/GDF11 antibody. As in mice, we observed a decrease in FSH both 2 and 7 days following injection (Fig. 5O–P).

Discussion

Here, we report that skeletal muscle-derived myostatin is the primary TGFβ family ligand driving FSH synthesis in mice. As myostatin is not produced in the pituitary gland, our results challenge two elements of current dogma. First, an endocrine rather than autocrine TGFβ ligand stimulates FSH. Second, this ligand is not activin B or, for that matter, any activin subtype. Interestingly, inhbb knockout fish also exhibit normal fshb expression (42). It is notable that activins were so named because of their presumed roles as important inducers of FSH synthesis and secretion (9, 43). Our data demonstrate that activin A or B, regardless of cellular/tissue origin, only minimally contribute to FSH production or secretion in vivo. Rather, FSH is greatly reduced in myostatin knockout mice (both global and muscle-specific). Myostatin’s dominant role in stimulating FSH is perhaps explained by its high levels in circulation, dwarfing those of GDF11 and activins in mice (40, 41).

In this regard, it is critical to ask whether myostatin is similarly important in other species. Myostatin’s best known function is in the regulation of muscle mass. Cattle breeds with double-muscling phenotypes have inactivating mutations in myostatin (44) and exhibit fertility problems (45). It is interesting to speculate that these may stem from deficiencies in FSH. Though, it should be noted that myostatin may also have direct actions in uterus, placenta, and ovary (46). Whereas myostatin plays a primary role in regulating skeletal muscle mass in mice and cattle, both myostatin and activin A are critical in primates (41). Therefore, in other species, it may be the combinatorial actions of activin-class ligands (myostatin, GDF11, activin A, activin B, and activin AB) that promote FSH synthesis. Though we are keen to assess roles for these proteins in humans, we have not yet been able to acquire samples from individuals treated with neutralizing antibodies comparable to those used here in mice.

As circulating myostatin is predominantly produced in skeletal muscle, our data suggest an additional endocrine axis in the control of reproduction. That the muscle communicates with pituitary gonadotropes is most clearly demonstrated by the reduced FSH in muscle-specific myostatin knockouts and the increase in FSH in myostatin KO mice expressing myostatin from skeletal muscle. However, it is important to acknowledge that there was some residual myostatin (~10%) in muscle-specific cKOs, suggesting a potential role for myostatin from additional sources. Myostatin may be expressed in the brain (47) and we cannot yet rule in or out a neuroendocrine mode of regulation. Indeed, an FSH releasing factor from the hypothalamus has long been sought but not identified (48).

Though the data clearly establish muscle-derived myostatin as the primary driver of FSH synthesis in mice, why this is the case is not yet clear. However, as sufficient musculature is required for both reproductive behavior and maintenance of pregnancy, one could speculate that myostatin provides a peripheral signal to the pituitary indicating a physiological level of preparedness. Consistent with this idea, we observed delayed puberty onset in myostatin deficient females. This muscle-pituitary axis is analogous to adipose tissue communicating via leptin to the central nervous system to regulate reproductive physiology (49). Going forward, it will be interesting to examine the relationship between muscle mass and reproductive health in humans and other animals.

It should be noted that myostatin levels per se may not predict FSH levels. In adult mice, we did not observe a correlation between the two hormones. In contrast, there was a positive correlation between myostatin and FSH in muscle-specific Mstn cKOs. Thus, above a certain threshold (~20 ng/mL), myostatin may be saturating in terms of stimulating FSH. We speculate that changes in FSH, for example across the menstrual or estrous cycle, are most likely explained by variation in inhibin A and/or B levels, rather than by changes in myostatin. Indeed, inhibin A antagonizes myostatin and GDF11 in gonadotropes as effectively as it blocks activin B.

Finally, our findings may have clinical relevance. First, measurements of myostatin, GDF11, and activins, in addition to the inhibins, in circulation should be considered in assessment of unexplained FSH dysregulation. Second, there are several drug development efforts aimed at inhibiting myostatin activity to treat muscle wasting disorders (49) or to prevent lean mass loss in individuals treated with GLP-1 receptor agonists (50). These inhibitors include myostatin neutralizing antibodies, type II receptor blocking antibodies, and ACVR2A/ACVR2B ligand traps (50). The latter inhibit FSH in post-menopausal women (50, 51). Though this has been attributed to activin antagonism, inhibition of myostatin is an alternative explanation. Moving forward, inhibitory effects of any myostatin antagonist on FSH and fertility should be considered.

Supplementary Material

Materials and Methods

Figs. S1 to S14

Tables S1 and S2

References (52–68)

Data and materials availability:

All data are available in the main text or the supplementary materials.