Loss of Anti-Müllerian Hormone Signaling in Mice Affects Trabecular Bone Mass in a Sex- and Age-Dependent Manner

At 2 months of age, female AMHKO mice were slightly, but significantly, heavier than WT mice ( P < 0.05). However, at subsequent ages, the analyzed body weight of female AMHKO and MRKI mice did not differ from WT mice ( Fig. 1A ). In agreement with our previous studies ( 23 , 24 ), female AMHKO mice had an increased number of growing follicles at 2 and 4 months of age ( P < 0.001 and P < 0.01, respectively) and a reduced number or no growing follicles at 10 and 16 months of age compared with WT mice ( P < 0.001 and P < 0.01, respectively) ( Fig. 1A ). Ovaries of female MRKI mice showed a similar pattern as AMHKO mice, with increased follicle numbers at 2 and 4 months ( P < 0.05) and severely reduced numbers at the older ages ( P < 0.001 and P < 0.05 at 10 and 16 months, respectively) ( Fig. 1A ). In line with this, the ovarian weight of AMHKO and MRKI mice was significantly increased at 2 and 4 months of age and significantly reduced at 10 and 16 months of age compared with WT mice ( Fig. 1A ). Given the increased follicle number in AMHKO and MRKI mice at the reproductive ages, we also determined estradiol and progesterone levels throughout the estrous cycle at 4 months of age. Both estradiol and progesterone levels showed the expected estrous variation but did not differ between genotypes. Also, testosterone levels did not significantly differ between the genotypes during the estrous cycle ( Fig. 1B ). In agreement, and confirming previous results in AMHKO mice ( 23 ), no difference in length or regularity of the cycle was observed between the genotypes. General characteristics and sex steroid hormone levels in WT and AMH signaling deficient mice. (A) Body weight, ovarian weight, and number of growing follicles in 2-, 4-, 10-, and 16-month-old female WT (white bars), AMHKO (gray bars), and MRKI (black bars) mice (n = 8-10). (B) Estradiol, progesterone, and testosterone levels during the estrous cycle in 4-month-old WT (dotted black line), AMHKO (solid gray line), and MRKI (solid black line) mice (n = 5 per cycle day). Abbreviations: P, pro-estrus; E, estrus; M, metestrus; D, di-estrus. $ P < 0.05, vs pro-estrus of same genotype. (C) Body weight and testis weight in 4-, 10-, and 16-month-old male mice (n = 8-10). (D) Testosterone levels in 4-month-old male WT (white bars), AMHKO (gray bars), and MRKI (black bars) mice (n = 6-7). Data are presented as mean ± SEM, * P < 0.05, ** P < 0.01, *** P < 0.001 vs WT mice of the same age. # P < 0.05, ## P < 0.01, ### P < 0.001 vs 2-month-old mice of the same genotype. For male mice, body weight of AMHKO and MRKI mice did not differ from WT mice ( Fig. 1C ). Testes of AMHKO and MRKI mice were significantly heavier compared with WT mice at all ages analyzed. Gross morphological analysis confirmed the previously reported presence of all stages of spermatogenesis and the appearance of empty seminiferous tubules with increasing age due to focal atrophy of the germinal epithelium ( 20 )(results not shown). However, testosterone levels did not significantly differ between genotypes, although large variation in levels was observed within the groups ( Fig. 1D ). Aging effects were examined, and WT mice showed the expected age-dependent changes in bone parameters. Both female and male WT mice showed an age-dependent decline in BV/TV (linear trend: P < 0.001). Further, in both female and male knockout models, trabecular bone was lost with increasing age (linear trend: both P < 0.001 [females: Supplementary Fig. 2A and males: Supplementary Fig. S1] ( 30 )). However, female MRKI mice had a higher BV/TV compared with WT mice at the age of 2 months ( P < 0.001). This increased value could be explained by a significant increase in both trabecular thickness and number (both P < 0.001) ( Fig. 2B and 2C ). The higher BV/TV in MRKI mice compared with WT mice persisted at the age of 4 and 10 months. At the age of 10 months, this increase was explained by a significant higher trabecular thickness in both knockout models ( P < 0.05 and P < 0.001 in AMHKO and MRKI mice, respectively) ( Fig. 2B ). At 16 months of age, BV/TV in female AMHKO and MRKI mice was similar to that of WT mice due to a greater decrease in trabecular bone between 10 and 16 months of age in both AMHKO and MRKI mice compared with WT mice ( Fig. 2A ). The decrease in trabecular bone was more pronounced in AMHKO mice, leading to a significantly different BV/TV between AMHKO and MRKI mice ( P < 0.05). This difference between genotypes was best explained by a significantly lower trabecular thickness in AMHKO mice compared with both WT and MRKI mice at 16 months of age ( P < 0.05 and P < 0.01, respectively) ( Fig. 2B ). Trabecular bone parameters in WT and AMH signaling deficient female mice. Trabecular bone microarchitecture in 2-, 4-, 10- and 16-month-old female WT (white bars), AMHKO (gray bars) and MRKI (black bars) mice analyzed using microCT (n = 8-10). (A) Trabecular bone volume fraction, (B) Trabecular thickness, (C) Trabecular number. Data are presented as mean ± SEM, * P < 0.05, *** P < 0.001 vs WT mice of the same age, $ P < 0.05 vs AMHKO mice, $$ P < 0.01 vs AMHKO mice. Concerning cortical bone, we observed an age-dependent increase in marrow area in both female and male WT mice (linear trend: both P < 0.001). A gradual increase of the periosteal perimeter was observed in female WT mice between 2 and 16 months of age (linear trend: P < 0.001) and also in male WT mice between 4 and 16 months of age (linear trend: P < 0.001). Both knockout models showed a similar pattern of aging as the WT mice (female: Fig. 3 and male: Supplementary Fig. S2 ( 30 )). However, differences between female WT and knockout mice were observed at 10 months of age. At this age, female MRKI mice had a significantly higher cortical area and marrow area compared with WT mice (both P < 0.01). As a result, the periosteal perimeter and moment of inertia of MRKI mice was also higher (both P < 0.001) ( Fig. 3 and Table 1 ). Cortical bone parameters in WT and AMH signaling deficient female mice. Cortical bone microarchitecture in 2-, 4-, 10-, and 16-month-old female WT (white bars), AMHKO (gray bars), and MRKI (black bars) mice analyzed using microCT (n = 8-10). (A) Cortical area, (B) Marrow area, (C) Periosteal perimeter. Data are presented as mean ± SEM, * P < 0.05, ** P < 0.01, *** P < 0.001 vs WT mice of the same age, $$$ P < 0.001 vs AMHKO mice. Trabecular and cortical bone parameters in WT and AMH signaling deficient female mice Trabecular and cortical bone microarchitecture in 2-, 4-, 10-, and 16-month-old female WT, AMHKO and MRKI mice analyzed using micro computed tomography (microCT) (n = 8-10); data are presented as mean ± SEM. Abbreviations: Ct.Th, cortical thickness; MOI, moment of inertia; SMI, structure model index; Tb.Pf, trabecular patterning factor; Tb.Sp, trabecular separation; Tt.Ar, tissue area. * P < 0.05 vs WT mice, ** P < 0.01 vs WT mice, *** P < 0.001 vs WT mice, $ P < 0.05 vs AMHKO mice, $$ P < 0.01 vs AMHKO mice, $$$ P < 0.001 vs AMHKO mice. The bone phenotype of both male knockout models resembled that of WT mice at all ages studied. There were no significant differences between the genotypes for any of the trabecular bone parameters (Supplementary Fig. S1 and Supplementary Table S1 ( 30 )). Regarding cortical bone parameters, we only observed a significantly smaller cortical area in AMHKO mice at the age of 10 months compared with WT mice ( P < 0.05) (Supplementary Fig. S2 and Supplementary Table S1 ( 30 )). The increased trabecular bone at reproductive age combined with enhanced loss of trabecular bone at older ages in AMHKO and MRKI mice correlates with the change in growing follicle number at these ages. Therefore, we measured serum levels of inhibin A and B, since these gonadal growth factors have been shown to have regenerative effects on bone. Throughout the estrous cycle at 4 months of age, inhibin A levels showed the expected estrous variation but did not differ between genotypes ( Fig. 4A ). At 2 and 10 months of age in samples randomly collected throughout the cycle, inhibin A levels were similar between genotypes, although large variation within genotypes may have masked potential differences. However, at 16 months of age, inhibin A levels were significantly lower in AMHKO and MRKI mice compared with WT mice ( P < 0.05). Furthermore, in AMHKO mice, inhibin A levels in the aged mice were significantly lower compared to young mice ( P < 0.001) ( Fig. 4B ). Inhibin A and B levels in WT and AMH signaling deficient mice. (A) Inhibin A and (C) Inhibin B levels during the estrous cycle in 4-month-old WT (dotted black line), AMHKO (solid gray line), and MRKI (solid black line) mice (n = 5 per cycle day). Abbreviations: P, pro-estrus; E, estrus; M, metestrus; D, di-estrus. Data are presented as mean ± SEM. * P < 0.05, *** P < 0.001 vs WT mice of same cycle day; $ P < 0.05, $$ P < 0.01 vs pro-estrus day of the same genotype. (B) Inhibin A and (D) inhibin B levels in 2-, 10-, and 16-month-old female WT (white bars), AMHKO (gray bars), and MRKI (black bars) mice (n = 8-10). Data are presented as mean ± SEM. * P < 0.05, *** P < 0.001 vs WT mice of same age. # P < 0.05, ## P < 0.01, ### P < 0.001 vs 2-month-old mice of same genotype. For inhibin B levels, we observed that at 4 months of age, levels were increased at estrus in AMHKO and MRKI ( P < 0.001 and P < 0.05 respectively), while no differences were observed at the other days of the estrous cycle ( Fig. 4C ). In samples collected randomly throughout the cycle, inhibin B levels did not differ between genotypes at 2 months of age. However, at 10 and 16 months of age, inhibin B levels were significantly lower in AMHKO and MRKI mice compared with WT mice ( P < 0.001). In all 3 genotypes, inhibin B levels declined with increasing age. However, in AMHKO and MRKI mice significantly lower levels were already observed at 10 months of age ( P < 0.05 and P < 0.001, respectively), while in WT mice a significant difference was only observed at 16 months of age ( P < 0.01) when compared with 2-month-old mice of the same genotype ( Fig. 4D ). Goldner's staining of femurs from 10-month-old female mice confirmed the observed differences in trabecular bone volume between the genotypes ( Fig. 5A ). Interestingly, clear differences were observed in bone marrow adiposity between genotypes: 4 out of 7 femurs of female AMHKO mice contained very few bone marrow adipocytes in the bone marrow cavity, while none of the WT or MRKI mice displayed this phenotype ( Fig. 5B ). Histomorphometry in WT and AMH signaling deficient female mice. (A) Histomorphometry in 10-month-old female WT and AMH signaling deficient mice performed using Goldner's staining. (B) Categorized bone marrow adiposity measurements. WT (white bars), AMHKO (gray bars), and MRKI (black bars) (n = 4-7). Data are presented as mean ± SEM.

Wild-type (WT), AMHKO, and AMHR2 knockout (MRKI) mice were generated as described previously ( 20 , 21 , 23 ). The MRKI mice (MIS receptor knockin, in which 4.4 kb of the receptor locus that includes the first 6 exons is replaced with a LacZ cassette) were kindly provided by Dr. R. Behringer. AMHKO and MRKI mice and WT littermates on a C57Bl/6 background were housed at standard animal housing conditions at the Animal Facility of the Erasmus MC, which operates in compliance with the “Animal Welfare Act” of the Dutch government. The experiments were performed with permission of the local ethics committee. Mice were sacrificed at 2, 4, 10, and 16 months of age (n = 8-12 mice per genotype and sex) as described previously ( 24 ). For steroid hormone measurements, 4-month-old female mice were sacrificed on each day of the ovarian cycle (n = 5 per cycle day). Daily vaginal smears were taken for a period of at least 2 weeks as described previously to determine the day of the cycle ( 23 ). Cycle length and regularity was determined by analyzing 2 complete cycles. Body weight and weight of testes and ovaries were determined. Femurs were collected and a blood sample was taken. The left testis and left ovary were fixed in Bouin's fixative. The left femurs were fixed in 10% formalin for 48 hours followed by preservation in 70% ethanol. The number of growing follicles was determined as described previously (n = 5 mice per genotype) ( 24 ), using 8-μm sections, mounted on glass slides and stained with hematoxylin and eosin. Nonatretic growing follicles were counted in every fifth section. The number of small preantral (20-170 μm), large preantral (171-220 μm), small antral (221-310 μm), and large antral (>311 μm) was combined to yield the total number of nonatretic growing follicles. Estradiol was measured by immunoassay. After ethyl ether extraction, serum mouse 17β-estradiol (E2) was measured by enzyme-linked immunosorbent assay (ELISA) kit in duplicate according to the manufacturer recommendations (Cat# SE 120084, RRID:AB_2827635, Sigma-Aldrich, Saint-Quentin Fallavier, France). The standard range was as follows: 0, 3, 10, 30, 100, and 300 pg/mL. The antibody is a polyclonal anti-estradiol antibody, coated on microwells ( 25 ). Anti-E2 antibody coated wells were incubated with E2 standards, controls, samples, and E2 enzyme conjugate. During the incubation, a fixed amount of HRP-labeled E2 competes with the endogenous E2 in the standards, samples, or control sera for a fixed number of binding sites of the specific E2 antibody. E2 Peroxidase Conjugate immunologically bound to the well progressively decreases as the concentration of E2 in the specimen increases. Then unbound E2 Peroxidase Conjugate was removed and the wells were washed. After a second incubation with a solution of TMB, the color development was stopped and the absorbance was measured spectrophotometrically at 450 nm. Quality control samples were provided by Probioqual (Lyon, France) and as “blank” control we used serum from prepubertal mice. Accuracy did not differ from 100%. The detection limit was 1 pg/mL. Intra- and inter-assay precision were, respectively, 15% and 20% at the concentration of 1 pg/mL, 8% and 10% at 4 pg/mL, and 5% and 8% at 7 pg/mL. Other steroid hormones were characterized using liquid chromatography-tandem mass spectrometry (LC-MS/MS) as described previously ( 26 ). One hundred μL of serum (sample or quality control) were vortex-mixed with 20 μL of the internal standard solution (a mix of testosterone-1a,2a-D2 and progesterone-1a,2a-D2) and 1.5 mL ethyl ether for 3 minutes. The solutions were then allowed to stand for 20 minutes, when the supernatants were separated and evaporated to dryness at room temperature under a stream of nitrogen. The dried extracts were reconstituted in 100 μL MeOH/water (55/45; V/V). The samples were injected into an Acquity UPLC system (Waters, Manchester, UK), equipped with an Acquity UHPLC HSST3 column (Waters, Guyancourt, France). The mobile phase consisted of a mixture of water and MeOH. It was operated with a flow rate of 0.5 mL/min in gradient mode, at a temperature of 40 °C. Mass spectra were recorded using a Quattro Premier triple quadrupole mass spectrometer (Waters, Manchester, UK). Measurements were performed using positive electrospray ionization (ESI) in Multiple Reaction Monitoring (MRM) data-acquisition mode. The parameters of the electrospray interface were optimized as follows: capillary voltage 4 kV, source temperature 120 °C, desolvation temperature 350 °C, desolvation gas 900 L/hour. Steroid identification was based on an identical retention time and 2 identical mass transitions with authentic reference compounds (testosterone from Sigma-Aldrich, and progesterone from Vetranal). Quantification was performed relative to a calibration series (0, 0.1 to 100 ng/mL of each steroid) with an appropriate internal standard steroid. All measurements were performed in duplicate. Data processing was ensured using MassLynx software (V 4.1, SCN 805, Waters, Manchester, UK). A QuantLynx (V 4.1, SCN 803, Waters) was used to detect and integrate the peaks. Accuracy checked in our unit did not differ from 100%. Detection limits were 0.06 nmol/L for testosterone and 0.1 nmol/L for progesterone. Intra-assay coefficients of variation for these steroids were < 10% at detection level, and < 8% throughout the range of observed values. Inter-assay coefficients of variation were < 12% throughout the range of observed values. Serum levels of inhibin A and B were measured using the multi-species Ansh Lab's ELISAs (Ansh Labs Cat# AL-161, RRID:AB_2783699; Ansh Labs Cat# AL-163, RRID:AB_2783700). Femurs were scanned at a resolution of 9 μm, using a SkyScan 1076 system (Bruker, Kontich, Belgium) in a blinded fashion. According to the published guidelines ( 27 ), the following settings were used: x-ray power and tube current were 40 kV and 250 μA, respectively. Beam hardening was reduced using a 1-mm aluminum filter, exposure time was 2.3 seconds, and an average of 3 pictures was taken at each angle with steps of 0.8° to generate final images. Segmentation of the reconstructed images was done on basis of global thresholding. Using software packages from Bruker (NRecon, CtAn, and Dataviewer), bone microarchitecture parameters were assessed in trabecular and cortical bones of all mice (n = 8-12 mice per genotype and sex). The trabecular bone parameters trabecular bone volume fraction (BV/TV), trabecular thickness (Tb.Th), trabecular number (Tb.N), trabecular separation (Tb.Sp), trabecular patterning factor (Tb.Pf) and structure model index (SMI; 0 = plate-like, 3 = rod-like) were determined in the distal metaphysis of the femur (region of interest [ROI] of 0.9 mm from distal growth plate toward diaphysis). For 16-month-old femurs, the trabecular compartment was manually selected due to the porous nature of the bones. In the mid-diaphysis (ROI of 0.45 mm), tissue area (Tt.Ar), cortical area (Ct.Ar), marrow area (Ma.Ar), cortical thickness (Ct.Th), periosteal perimeter (Ps.Pm) and moment of inertia (MOI) were analyzed. All ROIs were corrected for length of the femurs. Femurs of 10-month-old female mice (n = 4-7 mice per genotype) were routinely embedded in methylmetacrylate as described before ( 28 ). Sections of 6 μm were deacrylated, hydrated, and subjected to Goldner's Masson Trichrome staining. Briefly, sections were subsequently stained in hematoxylin/ferric chloride, ponceau de xylidine/acid fuchsin, orange G, and Light green solutions as described previously in detail ( 29 ). The sections were dehydrated and embedded in Entellan (Electron Microscopy Sciences, Hatfield, PA, USA). Images were taken using a NanoZoomer system and analyzed using NDP viewer (Nanozoomer 2.0 HT, Hamamatsu). Categorized bone marrow adiposity measurements based on the amount of bone marrow fat cells were performed in a blinded fashion. The following score system was used: 0 = near or complete absence of bone marrow fat cells; 1 = presence of bone marrow fat cells occupying less than half of the bone marrow cavity; 2 = presence of bone marrow fat cells occupying more than half of the bone marrow cavity; and 3 = presence of bone marrow fat cells filling nearly the entire bone marrow cavity. Steroid hormone levels during the estrous cycle were analyzed by two-way analysis of variance (ANOVA) followed by Bonferroni post hoc analysis. All other parameters were analyzed by one-way ANOVA followed by Tukey HSD post hoc test. To determine an effect of aging linear regression analysis was performed. Analyses were performed using GraphPad Prism 9 (GraphPad Software, Inc.). A P value <0.05 was considered significant. Data are expressed as mean ± SEM.

In this study, we investigated the effect of AMH signaling deficiency on bone. Female AMHKO and MRKI mice had increased trabecular bone compared with WT mice up to the age of 10 months, but this difference disappeared at 16 months of age. As shown by an increased marrow area and periosteal perimeter, MRKI mice displayed advanced cortical aging at 10 months of age compared with WT mice. Furthermore, we observed that female AMHKO mice of 10 months of age had remarkably few bone marrow adipocytes in their bone marrow cavity compared with the other genotypes. Previously, we demonstrated that AMH deficiency results in a gradual but accelerated ovarian aging ( 23 ). In this study, we observed that MRKI mice, an AMHR2 deficient model, display a similar ovarian aging phenotype. Therefore, we hypothesized that mice lacking AMH signaling could serve as a model for osteoporosis following natural menopause. Postmenopausal osteoporosis is characterized by a reduced trabecular bone mass and cortical thinning ( 8 , 11 ). Despite advanced depletion of the ovarian follicle pool compared to WT mice, at 16 months of age, bone parameters of female AMHKO and MRKI mice did not largely differ from female WT mice. In contrast, young female AMHKO and MRKI mice had more trabecular bone than WT mice. Combined with the relatively normal bone parameters at advanced age, this suggests that deficiency in AMH signaling causes accelerated bone loss after a critical age. Indeed, both female knockout models displayed a greater decrease in trabecular bone between 10 and 16 months of age compared with WT mice. In OVX mice, which constitute the most commonly used model for postmenopausal bone loss, trabecular bone architecture and cortical bone area decline soon after ovariectomy ( 12 ). Chemically induced menopause by 4-vinylcyclohexene diepoxide (VCD) in mice also resulted in a decrease in trabecular bone parameters. However, in contrast to OVX mice, cortical bone parameters were not affected ( 31 ). This is in agreement with our results, in which we only observed effects on trabecular bone but not cortical bone upon AMH signaling deficiency. In their study, Wright et al ( 31 ) suggested that, although both OVX and chemically induced menopause ablate estrogen production, the presence of residual ovarian tissue in the VCD-treated mice attenuates bone loss. Also, in both AMH signaling deficient models ovaries remain present. However, there are also important differences between our mouse models and the chemically induced mouse model of ovarian failure. In female AMHKO and MRKI mice, the increased loss of trabecular bone was only observed after 10 months of age, which correlated with the significant reduced number of follicles at this age. In the study of Wright et al ( 31 ), ovarian failure was induced at 1 month of age, and bone architecture was affected 3 months thereafter. Importantly, VCD treatment causes a rapid depletion of primordial and primary follicles, subsequently resulting in a decline in preantral and small antral follicles ( 32 ). Thus, there is a clear difference in the timing of bone loss. Another important difference is the dynamic of follicle loss. Although AMH signaling deficient mice also have an accelerated depletion of primordial follicles, this process is gradual and furthermore is paralleled with an increase in the number of preantral follicles in mice of 2- to 4-months of age. This increase in number of growing follicles, and thereby altered levels of gonadal growth factors, may likely explain the increase in trabecular bone observed during the early reproductive period of our mice. We previously showed that the increased number of growing follicles does not affect ovulation rate ( 24 ). In agreement, we now show that in AMHKO and MRKI mice estrogen and progesterone levels during the estrous cycle are not affected. Likewise, testosterone levels did not differ between genotypes during the estrous cycle, although levels appeared elevated at pro-estrus in the AMH signaling deficient mice. Although sex steroids are important hormones in bone homeostasis, the ovary also produces other ovarian factors, which potentially may affect bone metabolism. For instance, the gonadal growth factors inhibin and activin have been shown to have direct effects on both osteoblasts and osteoclasts in primary murine bone marrow cultures ( 33 ). Overexpression of inhibin A in mice resulted in an increased bone mineral density and bone volume ( 34 ). In our study, inhibin A levels of mice of reproductive age did not differ between genotypes. Given that inhibin A is mainly produced by pre-ovulatory follicles ( 35 ), this is in line with the normal number of pre-ovulatory follicles in AMHKO and MRKI mice. However, inhibin B levels were increased at estrus in AMHKO and MRKI mice, confirming our previous study in which a 2-fold increase in inhibin levels was observed in 4-month-old AMHKO mice at estrus ( 23 ). Since inhibin B is mainly produced by the earlier stages of follicular growth ( 35 ), this increase at estrus likely reflects the increased and earlier cyclic recruitment of growing follicles, previously described for AMHKO mice ( 24 ). Interestingly, the age-dependent decline in inhibin B levels was accelerated in AMHKO and MRKI mice, correlating with the earlier decline in number of growing follicles. We also noted that inhibin A levels were decreased in aged AMHKO and MRKI mice, suggesting cessation of ovulations at an earlier age. These findings show that changes in follicle dynamics in these mice result in concomitant changes in circulating ovarian growth factors. Whether inhibins account for the increased trabecular bone at reproductive age and accelerated bone loss with increasing age in AMHKO and MRKI mice remains to be determined. However, our results support findings of a cross-sectional age-stratified study of pre- and postmenopausal women, in which it was shown that serum inhibin A and inhibin B levels inversely correlated with bone turnover markers and had better predictive value than follicle-stimulating hormone (FSH) and E2, particularly in pre- and perimenopausal women ( 36 ). Other reproductive hormones may also contribute to the regulation of bone metabolism. As reviewed by Mills et al ( 37 ), hypothalamic and pituitary hormones, such as kisspeptin, gonadotropin-releasing hormone, and FSH, have been implicated. The recent discovery that AMH may regulate GnRH pulsatility ( 38 ) suggests that the effect of AMH signaling deficiency on bone could potentially also be mediated via regulation of the hypothalamic-pituitary-gonadal axis. Unfortunately, we were unable to measure FSH and estradiol levels in the aged mice due to insufficient serum volume. However, changes in FSH and estradiol occur relatively late during reproductive aging ( 39 ), suggesting that changes in circulating levels of gonadal growth factors may contribute to altered bone turnover during the early phase of reproductive aging. The observed effect of AMH signaling deficiency on bone appeared to be sex-dependent, since in male AMHKO and MRKI mice the trabecular and cortical bone parameters were comparable to WT mice. This sex-specific effect supports an indirect effect of AMH on bone homeostasis through other ovarian growth factors or hypothalamic/pituitary hormones. However, in a recent study, AMH was shown to inhibit TRAP-positive multinuclear osteoclast differentiation in osteoclast precursor cells, derived from mouse bone marrow cells ( 40 ). Thus, further research is required to elucidate these differences between in vivo and in vitro effects of AMH on bone cells. Histomorphometric analysis revealed differences in bone marrow adiposity between the genotypes. At 10 months of age, in more than 50% of the female AMHKO mice analyzed, very few bone marrow adipocytes were present in the bone marrow cavity of their femurs. Thus, female AMHKO mice seem to have increased osteogenesis and decreased adipogenesis. This phenotype was only observed in AMHKO mice and not in the MRKI mice. This phenotype is also remarkable, since during and after menopause normally a transition from osteogenesis to adipogenesis occurs ( 41-43 ), and thus further studies are required. The finding that AMH and AMHR2 deficiency differentially affect mesenchymal stem cell differentiation may imply that AMHR2 can signal independent from AMH. However, this is highly speculative as AMHKO and MRKI mice are phenocopies concerning Müllerian duct regression and ovarian phenotype. In conclusion, loss of AMH signaling resulted in an advanced loss of trabecular bone. However, due to an increased bone mass at early reproductive life, which coincided with an increased number of growing follicles, bone parameters of aged female AMHKO and MRKI mice did not largely differ from aged WT mice despite advanced ovarian aging. Although the bone phenotype in mice lacking AMH signaling does not directly resemble menopause-related osteoporosis in humans, our results suggest that mice deficient in AMH signaling may be an interesting model to identify reproductive hormones and their mode of action beyond estrogen that contribute to bone homeostasis.