Both enantiomers of β-aminoisobutyric acid BAIBA regulate Fgf23 via MRGPRD receptor by activating distinct signaling pathways in osteocytes

SUMMARY

With exercise, muscle and bone produce factors with beneficial effects on brain, fat, and other organs. Exercise in mice increased fibroblast growth factor 23 (FGF23), urine phosphate, and the muscle metabolite L-β-aminoisobutyric acid (L-BAIBA), suggesting that L-BAIBA may play a role in phosphate metabolism. Here, we show that L-BAIBA increases in serum with exercise and elevates Fgf23 in osteocytes. The D enantiomer, described to be elevated with exercise in humans, can also induce Fgf23 but through a delayed, indirect process via sclerostin. The two enantiomers both signal through the same receptor, Mas-related G-protein-coupled receptor type D, but activate distinct signaling pathways; L-BAIBA increases Fgf23 through Gαs/cAMP/PKA/CBP/β-catenin and Gαq/PKC/CREB, whereas D-BAIBA increases Fgf23 indirectly through sclerostin via Gαi/NF-κB. In vivo, both enantiomers increased Fgf23 in bone in parallel with elevated urinary phosphate excretion. Thus, exercise-induced increases in BAIBA and FGF23 work together to maintain phosphate homeostasis.

In brief

Kitase, Bonewald et al. demonstrate that the exercise metabolite BAIBA induces Fgf23 expression in osteocytes through the same receptor but through distinct, enantiomer-specific signaling pathways and molecular mechanisms.

Graphical Abstract:

INTRODUCTION

Exercise is crucial for the development and maintenance of the musculoskeletal system to prevent musculoskeletal disorders¹ and for whole-body health to prevent or delay chronic diseases.² It has been proposed that the two major tissues responsible for movement, bone and muscle, not only interact mechanically during exercise but also biochemically³ to maintain a healthy musculoskeletal system.^4,5 With exercise, soluble molecules are elevated in the circulation,⁶ contracting muscle releasing irisin, a secreted protein,⁷ and a metabolite, L-β-aminoisobutyric acid (L-BAIBA),⁸ whereas mechanically loaded osteocytes release prostaglandin E₂ and Wnts, known to have positive effects on muscle.⁹ A hormone mainly produced by osteocytes,¹⁰ fibroblast growth factor 23 (FGF23), is also elevated in the circulation with exercise.¹¹

Osteocytes are terminally differentiated osteoblasts, composing over 90%–95% of all bone cells in adult bone, living for decades within the bone matrix sensing and translating mechanical loading into mechanical signals.^12,13 Loading of osteocytes increases expression of numerous factors, including podoplanin/E11/gp38 that elongates dendrites,¹⁴ and osteoprotegerin and matrix extracellular phosphoglycoprotein that both inhibit osteoclastogenesis¹⁵ while decreasing the expression of factors that inhibit osteoblastic bone formation such as sclerostin.^16,17 Sclerostin is decreased in osteocytes in vivo with loading, but with hindlimb unloading, both sclerostin and receptor activator of nuclear factor kappa-B ligand (RANKL, encoded by Tnfsf11), a potent activator of osteoclasts, are elevated. Sclerostin (encoded by Sost) and RANKL not only negatively affect bone but also have negative effects on muscle.¹⁸

Osteocytes also function as endocrine cells to target other organs, such as kidney, to regulate phosphate metabolism by secreting the phosphaturic hormone, FGF23. FGF23 decreases circulating levels of phosphate by binding to the FGF receptor-1 and its co-receptor αKlotho to inhibit renal reabsorption of inorganic phosphate.^19,20 FGF23 is elevated in osteocytes in hypophosphatemic rickets,¹⁰ and transcription is positively regulated by 1,25(OH)₂D, phosphate, parathyroid hormone, and sclerostin.^21–23 Two critical negative regulators are dentin matrix protein 1 and phosphate-regulating gene, with homologies to endopeptidases on the X chromosome (PHEX).^24,25

The metabolite BAIBA has been shown to increase the browning of white adipose tissue and to increase hepatic free fatty acid β-oxidation, thereby preventing obesity and cardiovascular disease,^8,26,27 to reduce fasting blood glucose levels in type 2 diabetes, and to improve insulin resistance.^28,29 These reports suggest that BAIBA, produced by muscle with exercise, can improve lipid and glucose metabolism. However, BAIBA consists of two enantiomers, L(S)-BAIBA and D(R)-BAIBA, produced from valine in the mitochondria of skeletal muscle and thymine in the cytosol of liver and kidney cells, respectively.^26,30 Most studies have used racemic mixtures of BAIBA, so the function of each enantiomer is not clear. Only a few studies have investigated the biological functions of BAIBA enantiomers separately.^30–35 We reported that L-BAIBA, not D-BAIBA, is produced by murine contracted muscle and that L-BAIBA is more potent than D-BAIBA in maintaining osteocyte viability.³² L-BAIBA reduced both bone and muscle loss with hindlimb unloading³² and synergized with suboptimal loading to increase bone formation.³³ In humans, L-BAIBA is a potential marker of bone mineral density (BMD) and body mass index in females, whereas D-BAIBA correlates with age and with physical performance in males,³⁶ with physical activity in lean young women and with hip BMD in non-osteoporotic older women.³¹

Both L- and D-BAIBA are elevated in humans by exercise,³⁰ as is FGF23.^37,38 Here, we report that exercise significantly induced L-BAIBA and Fgf23 in mice. Both BAIBA enantiomers regulate Fgf23 mRNA expression in osteocytes through the same receptor, Mas-related G-protein-coupled receptor type D (MRGPRD), but via distinct downstream signaling pathways and in a time-dependent manner. In vivo studies using L-BAIBA showed early induction of Fgf23 in bone within 3 days, which was absent in mice lacking the MRGPRD receptor. Consistent with delayed gene expression of Fgf23 in bone, D-BAIBA increased FGF23 and urine phosphate at 5 days. These findings have implications for muscle-bone-kidney interactions, where both BAIBA enantiomers may contribute to the homeostasis of phosphate metabolism via timed regulation of Fgf23 production by osteocytes to maintain muscle function during and following exercise.

RESULTS

Treadmill exercise elevates FGF23, urine phosphate, and L-BAIBA

Exercise increases serum phosphate levels, which return to basal levels within 24 h after exercise.³⁹ FGF23, produced by osteocytes, plays a critical role in maintaining phosphate homeostasis.⁴⁰ We subjected male and female mice to treadmill exercise and found significant increases in L-BAIBA. Significant increases in intact FGF23 and creatinine were observed, whereas urine phosphate was significantly increased in males but did not reach significance in females at time points studied (Figures 1A and 1B). In males, but not in females, positive correlations were observed between L-BAIBA and intact FGF23 along with correlations of intact FGF23 with both urine phosphate and creatinine (Figure 1C). These results suggest that L-BAIBA may play a role in exercise-induced increase in FGF23 to regulate phosphate metabolism.

Figure 1. Treadmill exercise elevates L-BAIBA, intact FGF23, and urine phosphate levels.

(A and B) From left to right, the data display liquid chromatography-tandem mass spectrometry quantitation of plasma L-BAIBA and D-BAIBA, ELISA quantitation of plasma intact FGF23, and colorimetric quantitation of urine phosphate and creatinine obtained from males (A) and females (B). Treadmill exercise significantly increased the levels of L- but not D-BAIBA and elevated intact FGF23, urine phosphate, and creatinine.

(C and D) Pearson correlation analysis of L-BAIBA with plasma intact FGF23, as well as intact FGF23 with urine phosphate and creatinine are shown for males (C) and females (D). Positive correlations were observed between L-BAIBA and intact FGF23, while intact FGF23 correlated with urine phosphate and creatinine in males. A positive correlation was observed between intact FGF23 and creatinine in females.

The experiments were performed twice each for males and females and the resulting data combined (n = 9–12 male or female animals per group). Data are presented as mean ± SD, and p values (*p < 0.05, **p < 0.01, ***p < 0.001) are calculated using unpaired t test (A and B). Correlation coefficient (r) and p values obtained by Pearson correlation analysis are indicated (C and D).

The BAIBA enantiomers differentially regulate Fgf23 and Sost mRNA expression in murine osteocytes in vitro

We have shown different biological potencies between BAIBA enantiomers in protecting osteocytes from cell death³²; therefore, we decided to examine their signaling mechanisms (chemical structures of each enantiomer are shown in Figure S1A). In this study, we focused on the transcriptional regulation of Fgf23 and Sost, because both factors are produced by osteocytes.^41–44 Chromatograms in Figure S1 show the purity of synthesized enantiomers. L-BAIBA was more than 95% pure (Figure S1B) and D-BAIBA was 93.4% pure (Figure S1C). These highly pure preparations were used for all experiments in this study.

To determine the biological effects of L-BAIBA in osteocytes, IDG-SW3 cells,⁴⁵ cultured for 28 days to generate a mature osteocyte phenotype, were treated with a dose response of L-BAIBA (1–20 μM) for 24 h and 72 h. Fgf23 and Sost mRNA were quantitated by RT-qPCR (Figure 2). L-BAIBA at both 10 and 20 μM significantly increased Fgf23 mRNA expression by approximately 5-fold at 24 h, returning to basal levels by 72 h (Figure 2A). L-BAIBA had no effect on Sost mRNA at 24 h, but 20 μM L-BAIBA significantly decreased Sost mRNA at 72 h (Figure 2B). In contrast to L-BAIBA, D-BAIBA had no effect on Fgf23 expression at 24 h, but D-BAIBA at 10 and 20 μM increased Fgf23 mRNA at the late time point of 72 h (Figure 2C). Also, in contrast to L-BAIBA, D-BAIBA at 5 and 10 μM increased Sost mRNA at the early time point before returning to baseline at the later time point (Figure 2D). We also analyzed other osteocyte markers (Figures S2 and S3). A significant increase in Pdpn expression was observed for IDG-SW3 treated with 20 μM L-BAIBA at 24 h, but there were no significant changes in osteocyte-specific genes Dmp1, Phex, Mepe, Tnfsf11, and Tnfrsf11b (Figure S2A). However, D-BAIBA decreased Dmp1 and Phex expression at 24 h, with no effect on the other osteocyte genes (Figure S2B). At 72 h, Pdpn and Phex were dramatically increased by L-BAIBA (10–20 μM), but lower concentrations of L-BAIBA (2–10 μM) decreased Mepe expression (Figure S3A). D-BAIBA did not modify the expression of any genes except Fgf23 at 72 h (Figure S3B). Both L- and D-BAIBA did not increase any pro-apoptotic genes at 72 h (Figures S4A and S4B).

Figure 2. BAIBA enantiomers differentially regulate Fgf23 and Sost mRNA expression in mature osteocyte culture.

(A and B) RT-qPCR analysis of Fgf23 (A) and Sost mRNA (B) in day-28 IDG-SW3 treated with L-BAIBA for 24 h and 72 h. High doses of L-BAIBA increased Fgf23 at 24 h, which was returned to baseline at 72 h. In contrast, 20 μM L-BAIBA decreased Sost mRNA at 72 h but had no effect at 24 h. Data are presented as mean ± SD (n = 3 biological replicates per condition), and p values (**p < 0.01) are calculated using one-way ANOVA with Dunnett’s multiple comparisons.

(C and D) RT-qPCR analysis of Fgf23 (C) and Sost mRNA (D) in day-28 IDG-SW3 treated with D-BAIBA for 24 h and 72 h. High doses of D-BAIBA upregulated Fgf23 mRNA at 72 h but had no effect at 24 h. Contrary to Fgf23, Sost mRNA was upregulated at 24 h by 5 and 10 μM D-BAIBA, which was returned to baseline at 72 h. Data are presented as mean ± SD (n = 3 biological replicates per condition), and p values (*p < 0.05, **p < 0.01, ***p < 0.001) are calculated using one-way ANOVA with Dunnett’s multiple comparisons.

(E and F) RT-qPCR analysis of Fgf23 (E) and Sost mRNA (F) in osteocyte-enriched murine bone fragments treated with L- and D-BAIBA for 24 h (100 μM) and 72 h (200 μM). Effects of BAIBA in IDG-SW3 cells were replicated in primary osteocytes. There were no sex differences. Data are presented as mean ± SD (n = 4–7 biological replicates per condition), and p values (*p < 0.05, **p < 0.01, ***p < 0.001) are calculated using one-way ANOVA with Dunnett’s multiple comparisons.

See also Figures S2–S5.

To validate these in vitro results, ex vivo murine osteocyte-enriched bone chips isolated from 3-month-old male and female mice were treated with L- and D-BAIBA for 24 h and 72 h (Figures S5, 2E, and 2F). Higher concentrations of the enantiomers (100 μM at 24 h and 200 μM at 72 h) were required to observe effects on Fgf23 and Sost mRNA expression (Figures S5A and S5B). Effects were similar to those of the IDG-SW3 cell line (Figures 2E and 2F), as 100 μM L-BAIBA elevated Fgf23 mRNA about 4-fold, and 100 μM D-BAIBA increased Sost mRNA about 3-fold in primary osteocytes at 24 h regardless of sex. L-BAIBA at 200 μM had no effect on either Fgf23 or Sost mRNA at 72 h in primary osteocytes. However, 200 μM D-BAIBA elevated Fgf23 mRNA in both male and female osteocytes at 72 h. Unlike the in vitro cell line results, D-BAIBA maintained elevated Sost mRNA in primary osteocytes from 24 h to 72 h (Figure 2F). Both BAIBA enantiomers have biological effects on primary osteocytes and IDG-SW3 but regulate osteocyte markers uniquely.

Both BAIBA enantiomers utilize the same receptor, Mas-related G-protein-coupled receptor type D

We previously demonstrated that L-BAIBA protected osteocytes from oxidative stress-induced cell death through its receptor, MRGPRD.³² As both BAIBA enantiomers regulated the same osteocyte markers, Fgf23 and Sost, in osteocyte in vitro and ex vivo cultures but with different time dependency, we sought to determine whether the enantiomers used the same or different receptors and whether they had an effect on MRGPRD expression. MRGPRD protein levels were quantitated by western blot in day-28 cultured IDG-SW3 cells treated with 10 μM L- and D-BAIBA for 24 h. Both BAIBA enantiomers significantly increased MRGPRD protein levels 4- to 6-fold in IDG-SW3 cells (Figures 3A and 3B). To determine whether BAIBA enantiomers utilize MRGPRD to regulate Fgf23 and Sost expression, MU6840, an antagonist for this receptor, was used.⁴⁶ Pre-treatment with 50 μM MU6840 blocked L-BAIBA-induced Fgf23 mRNA, while both 25 and 50 μM MU6840 blocked D-BAIBA-induced Sost mRNA (Figure 3C). To validate that BAIBA signals through this receptor in primary cells, we used osteocyte-enriched bone chips isolated from 3-month-old Mrgprd-knockout (KO) mice for comparison with wild-type (WT) controls. Mrgprd-KO mice did not exhibit any significant differences from WT mice in bone parameters including microstructure and mechanical properties (Figure S6). At 24 h, 100 μM L-BAIBA increased Fgf23 mRNA expression in both male and female primary osteocytes in bone fragments from WT controls but not osteocytes from Mrgprd-KO mice. Treatment for 24 h with 100 μM D-BAIBA increased Sost mRNA expression in both male and female primary osteocytes of WT controls but not Mrgprd-KO mice, similar to the in vitro cell line results (Figure 3D). Furthermore, we confirmed that MRGPRD mediates L-BAIBA-induced Fgf23 expression in vivo. L-BAIBA administration, provided at 10 and 50 mg/kg/day for 3 days, increased Fgf23 mRNA in WT control bones, but not in Mrgprd-KO bones, in both males and females (Figure 3E).

Figure 3. Both BAIBA enantiomers utilize the same receptor, MRGPRD.

(A) Immunoblotting analysis of MRGPRD in day-28 IDG-SW3 treated with 0–20 μM L-BAIBA and D-BAIBA for 24 h.

(B) Quantitative data for (A). Both BAIBA enantiomers significantly increased MRGPRD with a 4- to 6-fold greater amount compared to controls. MRGPRD expression levels were normalized by β-tubulin. Data are presented as mean ± SD (n = 3 biological replicates per condition), and p values (*p < 0.05, **p < 0.01, ***p < 0.001) are calculated using one-way ANOVA with Dunnett’s multiple comparisons.

(C) 50 μM MU6840, an antagonist against MRGPRD, blocked Fgf23 mRNA expression induced by 10 μM L-BAIBA (left), and 25 and 50 μM MU6840 blocked Sost mRNA expression induced by D-BAIBA in day-28 IDG-SW3. Data are presented as mean ± SD (n = 3 biological replicates per condition), and p values (*p < 0.05) are calculated using two-way ANOVA with Tukey’s multiple comparisons.

(D) Both BAIBA enantiomers elevated Fgf23 and Sost mRNA in wild-type (WT) murine osteocyte-enriched bone fragments but not in Mrgprd-KO osteocytes ex vivo. There were no sex differences. Data are presented as mean ± SD (n = 3–6 biological replicates per condition), and p values (*p < 0.05, **p < 0.01, ***p < 0.001) are calculated using two-way ANOVA with Tukey’s multiple comparisons.

(E) L-BAIBA elevated Fgf23 mRNA in WT bones but not in Mrgprd-KO bones in vivo. There were no sex differences. Data are presented as mean ± SD (n = 3–8 male or female animals per group), and p values (*p < 0.05, ***p < 0.001) are calculated using two-way ANOVA with Tukey’s multiple comparisons.

(F) RT-qPCR analysis of Fgf23 and Sost mRNA in day-28 IDG-SW3 pre-treated with 100 and 1,000 μM of one enantiomer before treatment with 10 μM of the opposite enantiomer. Neither enantiomer was blocked by the opposite enantiomer, even if 100-fold excess was used. Data are presented as mean ± SD (n =4 biological replicates per condition), and p values are calculated using one-way ANOVA with Tukey’s multiple comparisons.

(G) BAIBA signaling pathway. L- and D-BAIBA utilize MRGPRD to regulate the transcription of Fgf23 and Sost, respectively. MU6840 is an antagonist against MRGPRD. Mrgprd-KO mice were also utilized in this study.

See also Figure S6.

As both the L-BAIBA and D-BAIBA enantiomers utilize the same receptor, MRGPRD, we sought to determine whether one enantiomer could block the effects of the other. Day-28 cultured IDG-SW3 cells were thus pre-treated with either enantiomer at 10, 100, and 1,000 μM for 30 min, followed by 10 μM of the converse enantiomer, and the cells were harvested at 24 h. Interestingly, even if cells were pre-treated with 100-fold more of the opposite enantiomer for 30 min, there was no blocking of their effects on either Fgf23 by L-BAIBA or Sost mRNA by D-BAIBA (Figure 3F). This shows that L-BAIBA and D-BAIBA do not block or interfere with the binding to or signaling of the MRGPRD receptor by the opposite enantiomer (Figure 3G).

The BAIBA enantiomers utilize different Gα proteins

MRGPRD was shown to be a G-protein-coupled receptor, coupling with Gαs, Gαq, and Gαi but not Gα_12/13 subtypes.⁴⁷ Therefore, specific inhibitors against each Gα protein were used to determine which Gα proteins L- and D-BAIBA might utilize to regulate the gene expression of Fgf23 and Sost. Melittin, a Gαs inhibitor, ebselen, a Gαq inhibitor, and pertussis toxin, a Gαi inhibitor,^48–50 were added 1 h before BAIBA. Melittin and ebselen, at 10 and 100 μM, blocked L-BAIBA-induced Fgf23 mRNA expression, but pertussis toxin, at 1 and 10 μM, had no effect (Figure 4A). In contrast, 10 μM pertussis toxin blocked D-BAIBA-induced Sost mRNA, but melittin and ebselen had no effect (Figure 4B). As Gαs proteins have been shown to upregulate intracellular cyclic AMP (cAMP), cAMP levels were measured in L- and D-BAIBA-treated IDG-SW3 as described in STAR Methods. L-BAIBA significantly increased intracellular cAMP level at 3 h before peaking at 6 h, and this level was maintained until 24 h compared to the control. In contrast, D-BAIBA significantly decreased cAMP at 6 h compared to the control (Figure 4C). These results suggest that the BAIBA enantiomers use different G-protein-coupled signaling pathways to regulate gene expression even though they bind to the same receptor (Figure 4D).

Figure 4. L-BAIBA activates Gαs and Gαq proteins, while D-BAIBA activates the Gαi protein.

(A) RT-qPCR analysis of Fgf23 mRNA in day-28 IDG-SW3 treated with 10 μM L-BAIBA in the presence of melittin, a Gαs inhibitor (left), ebselen, a Gαq inhibitor (middle), or pertussis toxin, a Gαi inhibitor (right). 100 nM melittin and ebselen blocked L-BAIBA-induced Fgf23 mRNA expression but not pertussis toxin. Data are presented as mean ± SD (n = 3 biological replicates per condition), and p values (*p < 0.05, **p < 0.01, ***p < 0.001) are calculated using two-way ANOVA with Tukey’s multiple comparisons.

(B) RT-qPCR analysis of Sost mRNA in day-28 IDG-SW3 treated with 10 μM D-BAIBA in the presence of melittin (left), ebselen (middle), or pertussis toxin (right). 10 ng/mL pertussis toxin blocked D-BAIBA-induced Sost mRNA expression, but neither melittin nor ebselen did. Data are presented as mean ± SD (n = 3 biological replicates per condition), and p values (*p < 0.05, **p < 0.01) are calculated using two-way ANOVA with Tukey’s multiple comparisons.

(C) cAMP production in day-28 IDG-SW3 treated with BAIBA enantiomers. Upon 10 μM L-BAIBA treatment cAMP increased after 3 h, peaked at 6 h, and maintained that level for up to 24 h. In contrast, 10 μM D-BAIBA decreased cAMP at 6 h but not at other time points. Data are presented as mean ± SD (n =3 biological replicates per condition), and p values (*p < 0.05, **p < 0.01) are calculated using one-way ANOVA with Dunnett’s multiple comparisons.

(D) BAIBA signaling pathway. L-BAIBA utilizes Gαs and Gαq coupled to MRGPRD while Gαi is utilized by D-BAIBA. cAMP was upregulated by L-BAIBA whereas it was inhibited by D-BAIBA with different response timing. Melittin is an inhibitor for Gαs, ebselen for Gαq, and pertussis toxin for Gαi, used in this study.

L-BAIBA activates Gαs/PKA/CBP/β-catenin and Gαq/PKC/CREB signaling pathways downstream of MRGPRD to regulate Fgf23 gene expression

Elevated cAMP has been known to activate protein kinase A (PKA) signaling.⁵¹ To determine whether L-BAIBA utilizes PKA to regulate Fgf23 mRNA, a PKA inhibitor, PKI5–24, was used. PKI5–24 at both 10 and 50 μM prevented L-BAIBA-induced Fgf23 expression, indicating that L-BAIBA utilizes the cAMP/PKA cascade to control Fgf23 expression (Figure 5A). As we had previously shown that L-BAIBA preserved osteocyte viability through the activation of β-catenin,³² we sought to determine whether L-BAIBA also utilizes β-catenin to regulate Fgf23 mRNA. Ser675 phospho-β-catenin, an activated form of β-catenin regulated by PKA, was quantitated by western blot and found to be significantly increased by 6 h in L-BAIBA-treated cells. Phosphorylation of Ser675 β-catenin was maintained for up to 24 h (Figures 5B and 5C). To determine whether the Gαs/PKA signaling pathway mediates L-BAIBA-induced phosphorylation of Ser675 β-catenin, IDG-SW3 cells were treated with the inhibitors melittin and PKI5–24. Both inhibitors blocked Ser675 β-catenin phosphorylation (Figures 5D and 5E). In addition, a CBP/β-catenin inhibitor, ICG-001, which competes with β-catenin for cAMP response element-binding protein (CREB)-binding protein (CBP),⁵² significantly inhibited L-BAIBA-induced Fgf23 mRNA expression at both 100 and 1,000 nM (Figure 5F).

Figure 5. L-BAIBA induces Fgf23 mRNA via Gαs/cAMP/PKA/CBP/β-catenin and Gαq/PKC/CREB signaling pathways downstream of MRGPRD.

(A) RT-qPCR analysis of Fgf23 mRNA in day-28 IDG-SW3 treated with 10 μM L-BAIBA in the presence of PKI5–24, a PKA inhibitor. PKA inhibitor blocked L-BAIBA-induced Fgf23 mRNA expression. Data are presented as mean ± SD (n = 4 biological replicates per condition), and p values (*p < 0.05, **p < 0.01, ***p < 0.001) are calculated using two-way ANOVA with Tukey’s multiple comparisons.

(B and C) Immunoblotting of phosphorylated β-catenin (Ser675, activated form) and total β-catenin (B) in a whole-cell lysate prepared from IDG-SW3 treated with 10 μM L-BAIBA. (C) Quantification of the immunoblots. Phosphorylation levels of β-catenin at Ser675 were increased 4-fold after 6 h, which was normalized by total β-catenin expression levels. Data are presented as mean ± SD (n = 3 biological replicates per condition), and p values (*p < 0.05, **p < 0.01) are calculated using one-way ANOVA with Dunnett’s multiple comparisons.

(D and E) Immunoblotting of phosphorylated β-catenin (Ser675) and total β-catenin (D) in a whole-cell lysate prepared from IDG-SW3 treated with 10 μM L-BAIBA in the presence of 100 nM melittin and 50 nM PKI5–24, Gαs, and PKA inhibitors, respectively. (E) Quantification of the immunoblots. Both Gαs and PKA inhibitors blocked L-BAIBA-induced phosphorylation of β-catenin at Ser675. Data are presented as mean ± SD (n = 3 biological replicates per condition), and p values (*p < 0.05, ***p < 0.001) are calculated using two-way ANOVA with Tukey’s multiple comparisons.

(F) RT-qPCR analysis of Fgf23 mRNA in day-28 IDG-SW3 treated with 10 μM L-BAIBA in the presence of ICG-001, a β-catenin inhibitor. High doses of ICG-001 blocked L-BAIBA-induced Fgf23 mRNA expression. Data are presented as mean ± SD (n = 3 biological replicates per condition), and p values (*p < 0.05, **p < 0.01, ***p < 0.001) are calculated using two-way ANOVA with Tukey’s multiple comparisons.

(G) RT-qPCR analysis of Fgf23 mRNA in day-28 IDG-SW3 treated with 10 μM L-BAIBA in the presence of Go6983, a PKC inhibitor. Go6983 blocked L-BAIBA-induced Fgf23 mRNA expression. Data are presented as mean ± SD (n = 3 biological replicates per condition), and p values (**p < 0.01, ***p < 0.001) are calculated using two-way ANOVA with Tukey’s multiple comparisons.

(H and I) Immunoblotting of phosphorylated CREB (Ser133) and total CREB (H) in a whole-cell lysate prepared from IDG-SW3 treated with 10 μM L-BAIBA for 3, 6, and 24 h. (I) Quantification of the immunoblots. L-BAIBA enhanced phosphorylation of CREB at Ser133 at 24 h. Data are presented as mean ± SD (n = 3 biological replicates per condition), and p values (*p < 0.05) are calculated using one-way ANOVA with Dunnett’s multiple comparisons.

(J and K) Immunoblotting of phosphorylated CREB (Ser133) and total CREB (J) in a whole-cell lysate prepared from IDG-SW3 treated with 10 μM L-BAIBA in the presence of 100 nM ebselen, a Gαq inhibitor. (K) Quantification of the immunoblots. Ebselen inhibited L-BAIBA-induced phosphorylation of CREB. Data are presented as mean ± SD (n = 3 biological replicates per condition), and p values (*p < 0.05, ***p < 0.001) are calculated using two-way ANOVA with Tukey’s multiple comparisons.

(L and M) Immunoblotting of phosphorylated CREB (Ser133) and total CREB (L) in a whole-cell lysate prepared from IDG-SW3 treated with 10 μM L-BAIBA in the presence of 100 nM Go6983, a PKC inhibitor. (M) Quantification of the immunoblots. Go6983 blocked L-BAIBA-induced phosphorylation of CREB. Data are presented as mean ± SD (n = 3 biological replicates per condition), and p values (*p < 0.05, **p < 0.01) are calculated using two-way ANOVA with Tukey’s multiple comparisons.

(N) BAIBA signaling pathway. L-BAIBA utilizes Gαs/cAMP/PKA/CBP/β-catenin and Gαq/PKC/CREB signaling pathways downstream of MRGPRD to regulate Fgf23 mRNA. Melittin is an inhibitor for Gαs, ebselen for Gαq, PKI5–24 for PKA, Go6983 for PKC, and ICG-001 for CBP/β-catenin, used in this study.

Protein kinase C (PKC) is a primary downstream target of Gαq.⁵³ Therefore, to determine whether L-BAIBA utilizes PKC to regulate Fgf23 mRNA, Go6983, a PKC inhibitor, was tested. Go6983 (100 nM) blocked L-BAIBA-induced Fgf23 mRNA expression, indicating that PKC also mediates Fgf23 transcription in L-BAIBA-treated IDG-SW3 cells (Figure 5G). CREB is a transcription factor induced by numerous mediators, including PKC.⁵⁴ Therefore, we quantified the phosphorylation of Ser133 CREB in L-BAIBA-treated IDG-SW3 cells by western blotting and found that CREB was significantly phosphorylated by L-BAIBA at 24 h (Figures 5H and 5I). To determine whether the Gαq/PKC pathway activates CREB, IDG-SW3 cells were treated with L-BAIBA in the presence of ebselen or Go6983 for 24 h, followed by quantitation of phosphorylated CREB by western blotting. Both inhibitors blocked L-BAIBA-induced phosphorylation of CREB at Ser133 (Figures 5J–5M). Taken together, these data show that L-BAIBA utilizes both the Gαs/PKA/CBP/β-catenin and Gαq/PKC/CREB signaling pathways downstream of MRGPRD to induce Fgf23 gene expression in osteocytes (Figure 5N).

D-BAIBA induces Fgf23 gene expression via sclerostin protein and the NF-κB pathway

We previously reported that sclerostin regulates Fgf23 gene expression via the nuclear factor (NF)-κB signaling pathway⁵⁵; because D-BAIBA induced Sost mRNA, we tested whether sclerostin was responsible for the delayed increase in Fgf23. Figure 6A shows the time course of Sost gene expression induced by D-BAIBA in IDG-SW3 cells. Sost mRNA was significantly increased at 24 h, peaked at 48 h, and returned to baseline by 72 h. A significant increase of secreted sclerostin protein into the cell-culture medium as determined by ELISA was observed at 24 h by D-BAIBA but not L-BAIBA (Figure 6B). To determine whether the secreted sclerostin is responsible for D-BAIBA-induced Fgf23 expression, a sclerostin-neutralizing antibody (AbD09097) was used. AbD09097 at 1,000 pg/mL significantly blocked the effect of D-BAIBA on Fgf23 expression, indicating that sclerostin is responsible for the induction of Fgf23 (Figure 6C). We had also shown previously that the induction of Fgf23 by sclerostin was mediated by NF-κB.⁵⁵ Therefore, it was investigated whether D-BAIBA phosphorylates p65 NF-κB in IDG-SW3 cells. Western blotting showed that NF-κB phosphorylation after D-BAIBA treatment was significantly increased at 24 h and maintained for up to 72 h (Figures 6D and 6E).

Figure 6. D-BAIBA-early-induced sclerostin is responsible for the later induction of Fgf23 mRNA.

(A) RT-qPCR analysis of Sost mRNA in IDG-SW3 treated with 10 μM D-BAIBA over a 72-h time course. Sost mRNA expression was increased at 24 h and 48 h but returned to basal levels by 72 h. Data are presented as mean ± SD (n = 4 biological replicates per condition), and p values (*p < 0.05, **p < 0.01 vs. 0 h) are calculated using one-way ANOVA with Dunnett’s multiple comparisons.

(B) Quantification of sclerostin protein in conditioned media from L- and D-BAIBA-treated IDG-SW3 at 24 h. 10 μM D-BAIBA elevated the secretion of sclerostin protein more than 2-fold compared to controls. Data are presented as mean ± SD (n = 4 biological replicates per condition), and p values (*p < 0.05) are calculated using one-way ANOVA with Dunnett’s multiple comparisons.

(C) RT-qPCR analysis of Fgf23 mRNA in IDG-SW3 treated with 10 μM D-BAIBA in the presence of AbD09097, a sclerostin-neutralizing antibody, for 72 h. 1,000 pg/mL AbD09097 blocked D-BAIBA-induced Fgf23 mRNA expression. Data are presented as mean ± SD (n = 3 biological replicates per condition), and p values (***p < 0.001) are calculated using two-way ANOVA with Tukey’s multiple comparisons.

(D and E) Immunoblotting of phospho NF-κB p65 and total NF-κB p65 (D) in whole-cell lysate from 10 μM D-BAIBA-treated IDG-SW3. (E) Quantification of the immunoblots. D-BAIBA enhanced a 4- to 6-fold of NF-κB p65 phosphorylation after 24–72 h. Data are presented as mean ± SD (n = 3 biological replicates per condition), and p values (*p < 0.05, **p < 0.01) are calculated using one-way ANOVA with Dunnett’s multiple comparisons.

(F–I) RT-qPCR analysis of Sost and Fgf23 mRNA expression in IDG-SW3 treated with 10 μM D-BAIBA in the presence of BMS345541, an NF-κB inhibitor for 0–24 h (F), 24–48 h (G), 24–72 h (H), and 48–72 h (I). 500 nM BMS345541 completely blocked D-BAIBA-induced Sost mRNA expression when treated between 0 and 24 h, but Fgf23 mRNA showed no changes. Both Sost and Fgf23 mRNA expressions were elevated by D-BAIBA at 48 h, which were reduced by the treatment of BMS345541 between 24 and 48 h. The treatment of BMS345541 for 24–72 h blocked a late induction of Fgf23 by D-BAIBA, and 48–72 h of treatment of BMS345541 partially reduced D-BAIBA-elevated Fgf23 mRNA, but Sost mRNA showed no changes except a reduction by D-BAIBA in the presence of 500 nM BMS345541 at 72 h. All data are presented as mean ± SD (n = 3 or 4 biological replicates per condition), and p values (*p < 0.05, **p < 0.01, ***p < 0.001) are calculated using two-way ANOVA with Tukey’s multiple comparisons.

(J) BAIBA signaling pathway. D-BAIBA utilizes Gαi/NF-κB/sclerostin/NF-κB signaling pathway downstream of MRGPRD to regulate Fgf23 mRNA. BMS34551 is an inhibitor of NF-κB, and AbO09097 is an antibody against sclerostin.

To determine whether NF-κB was responsible for D-BAIBA induction of sclerostin, and in turn for sclerostin induction of Fgf23, the NF-κB inhibitor, BMS345541 (100 and 500 nM), was used (Figures 6F–6I). When BMS345541 was added to D-BAIBA-treated cultures from 0 h to 24 h, and the cells were harvested at the first 24 h, Sost gene expression was blocked (Figure 6F). This shows that the D-BAIBA-mediated early induction of Sost mRNA is mediated by NF-κB. As expected, the inhibitor had no effect on basal expression of Fgf23 mRNA at 24 h, as shown in Figure 6F. As we had shown previously that the induction of Fgf23 by sclerostin was mediated by NF-κB,⁵⁵ we tested whether sclerostin utilized the same signaling pathway in these experiments. Timing experiments were performed whereby the inhibitor was only added between 24 and 48 h, 24 and 72 h, and 48 and 72 h. BMS3455341 significantly decreased D-BAIBA-induced Sost and Fgf23 mRNA expression when added from 24 to 48 h (Figure 6G). This result suggests that inhibition by BMS345541 partially reduced sclerostin, which in turn resulted in decreased induction of Fgf23 mRNA. When the inhibitor was added at 24 h and the cultures harvested at 72 h, Fgf23 mRNA induction by D-BAIBA was blocked while Sost mRNA returned to baseline (Figure 6H). D-BAIBA-induced Sost expression peaked at 48 h, as shown in Figure 6A; therefore, to determine whether NF-κB activated by sclerostin was responsible for Fgf23 expression, treatment of BMS345541 from 48 to 72 h was performed. A reduction of Fgf23 mRNA was observed at 72 h, suggesting that sclerostin-activated NF-κB plays a role in the increase in Fgf23 mRNA (Figure 6I). Collectively, our data show that NF-κB mediates both D-BAIBA-induced sclerostin production and sclerostin-induced Fgf23 expression (Figure 6J).

To summarize our in vitro and ex vivo results, both BAIBA enantiomers signal through the MRGPRD but utilize different Gα proteins. L-BAIBA signals through Gαs to increase cAMP to target PKA to phosphorylate β-catenin, and L-BAIBA also signals through Gαq to activate PKC, which phosphorylates CREB, leading to an increase in Fgf23 mRNA. In contrast, D-BAIBA utilizes Gαi/NF-κB signaling to induce sclerostin production, with the secreted sclerostin protein then responsible for increased Fgf23 gene expression, also through NF-κB signaling. In conclusion, L- and D-BAIBA regulate Fgf23 through the same receptor but activate distinct signaling pathways.

L-BAIBA treatment for 3 days increases Fgf23 mRNA in bone in 3-month-old male and female mice

We next sought to determine the effects of BAIBA on phosphate metabolism in vivo by quantitating intact FGF23 and urinary phosphate along with Fgf23 mRNA expression in bone. Normal mice were provided BAIBA in their drinking water as described previously.^32,33 For L-BAIBA, a 3-day treatment was performed based on the in vitro and ex vivo results showing that L-BAIBA increased Fgf23 mRNA in osteocytes at an early time point, by 24 h. L-BAIBA (100 mg/kg/day) had no effect on Fgf23 mRNA expression in 5-month-old male and female bones, unlike D-BAIBA (Table 1). Next, a dose-response analysis (1, 10, 50, and 100 mg/kg/day L-BAIBA) was performed for 3 days using 3-month-old male and female mice. L-BAIBA at 10 mg/kg/day for males and 10 and 50 mg/kg/day for females significantly increased Fgf23 mRNA in bone (2.6-fold in 10-mg/kg/day-treated males, 2.2- and 2.3-fold in 10- and 50-mg/kg/day-treated females, respectively). In males, 10 mg/kg/day L-BAIBA also increased urine phosphate levels when corrected for urine creatinine (3.9 ± 0.4 mg/mg in control, 5.9 ± 2.1 mg/mg in L-BAIBA) (Table 1). Thus, the effects of L-BAIBA may depend on the dose, duration of treatment, age, and sex.

Table 1.

In vivo effects of L-BAIBA (3 days) and D-BAIBA (5 days)

Age	Sex	-	Fgf23 mRNA in bone (fold change)		Intact FGF23 in blood (pg/mL)		Pi in blood (mg/dL)		Urine Pi/creatinine
			Mean ± SD	p value	Mean ± SD	p value	Mean ± SD	p value	Mean ± SD	p value
5 months	male	0	1.0 ± 0.0	0.065	107.1 ± 19.4	0.269	8.2 ± 1.0	0.402	3.4 ± 4.3	0.896
		100	0.9 ± 0.0		134.5 ± 35.2		7.1 ± 2.0		3.1 ± 2.5
	female	0	1.0 ± 0.3	0.492	129.7 ± 7.0	0.356	5.6 ± 1.3	0.074	2.8 ± 1.1	0.702
		100	1.1 ± 0.1		120.3 ± 17.0		3.8 ± 0.5		3.2 ± 1.4
3 months	male	0	1.0 ± 0.7	-	135.5 ± 19.9	-	3.8 ± 0.3	-	3.9 ± 0.4	-
		1	2.1 ± 0.6	0.054	ND	ND	ND	ND	3.6 ± 1.2	0.996
		10	2.6 ± 0.9	0.004**	147.3 ± 7.5	0.296	3.8 ± 0.7	>0.999	5.9 ± 2.1	0.046*
		50	1.8 ± 0.6	0.218	ND	ND	ND	ND	4.0 ± 1.9	>0.999
		100	1.1 ± 0.2	>0.999	ND	ND	ND	ND	4.9 ± 0.9	0.627
	female	0	1.0 ± 0.4	-	175.5 ± 32.1	-	4.0 ± 0.4	-	3.6 ± 0.5	-
		1	1.3 ± 0.3	0.724	ND	ND	ND	ND	3.5 ± 1.0	0.999
		10	2.2 ± 0.4	0.006**	ND	ND	ND	ND	2.9 ± 0.5	0.721
		50	2.3 ± 0.6	0.004**	184.9 ± 47.0	0.779	4.8 ± 0.3	0.013**	3.5 ± 1.3	0.828
		100	1.2 ± 0.3	0.902	ND	ND	ND	ND	3.1 ± 1.0	0.876
Age	Sex	D-BAIBA concentration (mg/kg/day)	Fgf23 mRNA in bone (fold change)		Intact FGF23 in blood (pg/mL)		Pi in blood (mg/dL)		Urine Pi/creatinine
			Mean ± SD	p value	Mean ± SD	p value	Mean ± SD	p value	Mean ± SD	p value
5 months	male	0	1.0 ± 0.3	0.020*	107.1 ± 19.4	0.138	8.2 ± 1.0	0.040**	1.2 ± 0.8	0.007**
		100	2.5 ± 1.9		136.4 ± 23.1		6.3 ± 0.8		6.5 ± 2.1
	female	0	1.0 ± 0.3	0.022*	147.4 ± 14.8	0.695	5.0 ± 0.8	0.821	0.6 ± 0.1	<0.0001***
		100	2.1 ± 0.7		144.1 ± 10.4		4.9 ± 1.3		12.8 ± 2.2
3 months	male	0	1.0 ± 0.4	0.010*	235.7 ± 23.4	0.009**	4.0 ± 0.4	0.132	2.0 ± 1.4	0.404
		100	3.1 ± 1.0		288.6 ± 12.6		4.4 ± 0.7		1.7 ± 0.7
	female	0	1.0 ± 0.2	0.862	351.5 ± 65.7	0.747	4.2 ± 1.2	0.747	1.5 ± 0.7	0.092
		100	1.0 ± 0.4		288.6 ± 25.2		4.5 ± 0.0		2.7 ± 0.8

Pi, inorganic phosphate; ND, not determined. Data are presented as mean ± SD (n = 3–9 animals per group;

^*p < 0.05,

^**p < 0.01,

^***p < 0.001 vs. 0 mg/kg/day BAIBA, unpaired t test or one-way ANOVA/Dunnett).

D-BAIBA treatment for 5 days increases Fgf23 mRNA in bone and urine phosphate excretion in 5-month-old male and female mice

To test D-BAIBA on Fgf23 production, 5- and 3-month-old C57BL/6 mice were treated with 100 mg/kg/day D-BAIBA for 5 days. The treatment period was based on the results of the in vitro and ex vivo experiments showing that D-BAIBA increased Fgf23 mRNA in osteocytes at a later time point, 72 h, than that of L-BAIBA at 24 h. D-BAIBA significantly increased Fgf23 mRNA in bones of 5-month-old mice (2.5-fold in males, 2.1-fold in females) and urinary phosphate excretion when corrected for urine creatinine (1.2 ± 0.8 mg/mg in control males, 6.5 ± 2.1 mg/mg in D-BAIBA-treated males; 0.6 ± 0.1 mg/mg in control females, 12.8 ± 2.2 mg/mg in D-BAIBA-treated females) (Table 1). D-BAIBA reduced circulating levels of phosphate in males (8.2 ± 1.0 mL/dL in control, 6.3 ± 0.8 mg/dL in D-BAIBA) but not in females. When repeated in 3-month-old male mice, D-BAIBA increased Fgf23 mRNA in bone (3-fold) and circulating intact FGF23 (235.7 ± 23.4 pg/mL in control, 288.6 ± 12.6 pg/mL in D-BAIBA), but none of these parameters were changed in females (Table 1). Like L-BAIBA, the effect of D-BAIBA on phosphate metabolism may depend on dose, duration of treatment, age, and sex. This is consistent with the concept that age and sex can affect the enzymatic metabolism of enantiomers.⁵⁶

DISCUSSION

We previously reported that contracted skeletal muscle from mice produce L-BAIBA but not D-BAIBA, and that L-BAIBA protects osteocytes from the cytotoxic effects of reactive oxygen species (ROS) through the MRGPRD and is 100–1,000 times more potent than D-BAIBA.³² A recent report demonstrated the elevation of both L- and D-BAIBA by endurance exercise in humans.³⁰ In the current study, we found a significant correlation of L- but not D-BAIBA with FGF23 and urinary phosphate during treadmill running in males. Therefore, we sought to determine the biological effects of each enantiomer on osteocyte function, to determine whether the receptor, MRGPRD, was shared by both enantiomers, and to identify their downstream signaling mechanisms. Enantiomers have been well studied and can differ in their bioavailability, rate of metabolism, metabolites, excretion, potency, and toxicity, and their selectivity for receptors, transporters, and enzymes.⁵⁷ One enantiomer can be active but the other inactive, and one enantiomer can have a physiological or pharmacological effect whereas the other is toxic. Enantiomers can block the action of the opposite enantiomer, signal through the same or different receptors, or potentiate the activity of the opposite enantiomer.⁵⁸ Examples include ketamine where one enantiomer is less active than the other,⁵⁹ eszopiclone that has one inactive enantiomer,⁶⁰ cholesterol that has a blocking enantiomer,⁶¹ and propoxyphene as an example of enantiomers with very different functions where one enantiomer is an analgesic and the other an antitussive.⁶² A famous example of an enantiomer with adverse effects compared to the desired effect of the other is the sedative thalidomide, which was withdrawn from the market when one enantiomer caused congenital defects.^63,64 Therefore, it is very important to understand enantiomer function.

Here, we show that both L- and D-BAIBA utilize the MRGPRD receptor to regulate Fgf23 mRNA, that L-BAIBA signals through Gαs/cAMP/PKA/CBP/β-catenin and Gαq/PKC/CREB, and that D-BAIBA signals through Gαi/NF-κB to regulate sclerostin, which increases Fgf23 expression also through NF-κB signaling. These mechanisms bear some similarity to that of tramadol, whereby both enantiomers contribute to analgesic activity through the same mu-opioid receptor but via different mechanisms; one inhibits serotonin reuptake while the other inhibits norepinephrine reuptake, enhancing inhibitory effects on pain transmission in the spinal cord.⁶⁵ This difference in signaling pathways may explain time-course and dose differences observed for both enantiomers in vitro and in vivo.

Both BAIBA enantiomers utilize the same receptor to regulate Fgf23 and Sost expression in osteocytes, as the MRGPRD antagonist MU6840 blocked both L- and D-BAIBA-induced gene expression. The observation that neither primary osteocytes from Mrgprd null mice nor the mice themselves responded to either enantiomer further indicates that both enantiomers signal through this receptor. It is important to note that the observed ex vivo and in vivo results are due to the absence of the receptors themselves rather than any associated bone phenotype. MRGPRD is one of the G-protein-coupled receptors that interact with heterotrimeric G proteins, Gα, Gβ, and Gγ. Gα and Gβγ separately mediate downstream signaling activities.⁶⁶ As MRGPRD has been shown to couple to different signaling pathways, which utilize Gαs, Gαq, and Gαi,⁴⁷ we tested the inhibitors for these G proteins and found that L-BAIBA-induced signaling was blocked by the Gαs and Gαq inhibitors, while D-BAIBA was blocked by the Gαi inhibitor. At this time, we can only surmise that the enantiomers may change the confirmation of the receptor to activate distinct signaling pathways.

We also demonstrated that L-BAIBA was a potent inducer of cAMP, unlike D-BAIBA. However, despite the fact that D-BAIBA activated Gαi, which is known to inhibit Gαs-induced adenylyl cyclase activity and cAMP accumulation,⁶⁷ in our experiments D-BAIBA did not block L-BAIBA-induced Fgf23 expression even with pre-incubation using a 100-fold greater amount. This could be because L-BAIBA significantly increased intracellular cAMP levels at 3 h which was maintained for 24 h, whereas D-BAIBA showed a significant decrease in cAMP levels at only 6 h. Different timing of cAMP regulation may play a role in the observed lack of interference between the two enantiomers.

To our knowledge, there are no studies showing that Fgf23 expression is regulated by β-catenin signaling. We demonstrated that L-BAIBA phosphorylates (Ser675) β-catenin, and this effect was blocked by melittin (a Gαs inhibitor) and PKI5–24 (a PKA inhibitor). Inhibition of β-catenin by ICG-001 (a CBP/β-catenin inhibitor) blocked L-BAIBA-induced Fgf23 gene expression at 24 h. Based on our results, L-BAIBA utilizes the Gαs/cAMP/PKA/CBP/β-catenin signaling pathway to regulate Fgf23 expression in osteocytes within the first 24 h.

It is well known that Wnt/β-catenin signaling is crucial in bone development, bone formation, and maintenance of bone mass.⁶⁸ We previously reported that activation of the Wnt/β-catenin signaling pathway prevents or reduces osteocyte apoptosis¹² and that L-BAIBA protects against ROS-induced cell death in osteocytes through β-catenin.³² In this study, we showed that β-catenin was phosphorylated at Ser675, known to be phosphorylated by PKA to induce stabilization, accumulation, and nuclear translocation of β-catenin,⁶⁹ in contrast to Ser33, Ser37, and Thy41, which led to β-catenin degradation by the ubiquitin-proteasome system. However, this activation of β-catenin by L-BAIBA does not result in new bone formation unlike its activation through low-density lipoprotein receptor-related protein 5/6 (LRP5/6),³³ suggesting that activation of β-catenin alone by L-BAIBA is not sufficient for bone formation.

Importantly, L-BAIBA rapidly increased Fgf23 expression at 24 h, which returned to baseline by 72 h; however, Phex expression was significantly elevated by that time. This downregulation of Fgf23 by 72 h may be due to increased Phex,⁷⁰ a negative regulator of Fgf23 expression. Moreover, our results suggest that L-BAIBA also utilizes PKC to regulate Fgf23 expression. Gαq activates PKC, which in turn is known to increase FGF23 production.⁷¹ Although both PKA and PKC are known to activate CREB,⁷² it was the PKC inhibitor that completely blocked L-BAIBA-induced CREB phosphorylation in IDG-SW3, indicating that L-BAIBA mainly activates CREB via the Gαq/PKC signaling pathway to increase Fgf23 mRNA. Currently, the mechanism whereby Phex is elevated at 72 h is unknown.

Sclerostin is a well-known inhibitor of Wnt/β-catenin signaling that acts by binding to LRP5/6 to inhibit osteoblast differentiation and function.⁷³ Sost null mice have low levels of FGF23 and high levels of serum phosphate.⁷⁴ Sclerostin antibody has been shown to decrease FGF23 and increase phosphate in Hyp mice serum.⁷⁵ Sclerostin has also been reported to induce Fgf23 expression through the NF-κB signaling pathway in osteocytes.⁵⁵ Here, we show that D-BAIBA increases Sost mRNA at 24–48 h and increases sclerostin secretion in 24 h and that a sclerostin-neutralizing antibody, AbD09097, completely blocked D-BAIBA-induced Fgf23 expression at 72 h. AbD09097 acts as a neutralizing antibody by interacting with the secondary loop of sclerostin, the binding site for LRP5/6.⁷⁶ We also showed that D-BAIBA induced phosphorylation of p65 NF-κB after 24 h and that an NF-κB inhibitor blocks early (24 h) Sost mRNA expression. In addition, late (72 h) Fgf23 expression was also blocked when this inhibitor was added for 24–72 h. These inhibitory effects were partial when added during 24–48 h and 48–72 h but still reduced Fgf23 expression, suggesting the involvement of sclerostin and NF-κB induced by D-BAIBA. Therefore, the NF-κB signaling pathway is critical in regulating both early Sost mRNA expression and sclerostin-induced late Fgf23 expression in D-BAIBA-treated osteocytes.

Our in vivo results demonstrated that 3-day treatment with L-BAIBA increased Fgf23 mRNA in the bone of 3-month-old male and female mice, but there were no changes in 5-monthold mice. However, 5-day treatment with D-BAIBA increased Fgf23 mRNA in both 5-month-old male and female mice, which was accompanied by increased urine phosphate excretion in males but not females. Males at 3 months showed an increase in Fgf23 in bone and an increase in iFGF23 in serum. These results suggest that the dose response to BAIBA enantiomers and the timing of phosphate regulation may depend on age and sex and that compensatory regulation of bioactive intact FGF23 by phosphate may still be intact.

Circulating phosphate levels are essential for normal muscle function. Muscle cells express the phosphate transporters, Pit1 and Pit2, and deletion of these transporters lowers plasma phosphate levels and causes myopathy and early death in mice, indicating that low phosphate affects muscle function and survival.⁷⁷ On the other hand, patients with chronic kidney disease show high levels of serum phosphate levels and low muscle function, and phosphate overload has been shown to induce myotube atrophy in vitro.^78,79 Similar to FGF23,^37,38 serum phosphate levels are elevated by exercise.³⁹ This suggests that phosphate is secreted by skeletal muscle and that phosphate levels must be tightly regulated to maintain optimal muscle function. These reports, in combination with our findings, suggest that BAIBA enantiomers may induce FGF23 production in osteocytes and, together with parathyroid hormone,^80–82 act to normalize exercise-elevated serum phosphate via increased urinary output. These mechanisms may protect against muscle dysfunction due to elevated phosphate concentrations with exercise.

Limitations of the study

There are several limitations to this study, the first of which is the lack of structural receptor information that would allow better understanding of MRGPRD and its interactions with ligands such as the BAIBA enantiomers. A second limitation is that while we have tested the effects of L-BAIBA on several bone functions, it reduces the effects of hindlimb unloading on bone and muscle loss³² and enhances the effects of suboptimal anabolic loading on bone formation but by itself does not stimulate bone formation.³³ These studies have not been performed for D-BAIBA, so the physiological significance of this enantiomer is unclear. Future experiments examining the effects of D-BAIBA on bone remodeling in vivo and effects on the osteocyte phenotype in vitro remain to be performed. A third limitation is the lack of studies examining the direct effects of the enantiomers on osteoclasts. NF-κB may also affect osteoclast function and, if so, it could be important to measure other parameters of osteoclast function such as CTX-I, RANKL, and others. A fourth limitation is the relatively high dose of the enantiomers necessary to induce signaling in osteocytes in vitro and ex vivo (5–20 μM) compared to circulating levels and those induced by exercise (approximately 0.05 μM). A fifth limitation is that we have only performed short-term and not long-term treadmill exercise, which may explain why we observed an increase in L- but not D-BAIBA; however, others have described how both L- and D-BAIBA were elevated in physically active humans with endurance exercise. In addition, it would be essential to understand how resistance exercise, aerobic exercise, and the combination affect BAIBA, FGF23, and phosphate levels differently. As both L- and D-BAIBA are increased with exercise, both BAIBA enantiomers may contribute to the increase in FGF23 that occurs with endurance exercise.³⁰ Future studies would focus on acutely and chronically exercised mice to show whether BAIBA regulates FGF23 during exercise and whether feedback by normal phosphate exists to inhibit intact FGF23 bioactivity by protein processing. This would be similar to observations in other systems, where Fgf23 gene expression is stimulated (i.e., hypoferremia, inflammation), but low phosphate may feed back to induce intracellular cleavage of intact FGF23 in osteocytes.⁸³ Although we do not have the means to block the production of L-BAIBA by muscle as many of the enzymes necessary for its synthesis are also essential for other metabolites, in future studies conditional osteocyte Mrgprd-KO mice would be a useful tool to determine the crosstalk between skeletal muscle, bone, and kidney to regulate phosphate metabolism during exercise.

In conclusion, we demonstrated that BAIBA enantiomers increase FGF23 production in osteocytes through the same receptor, MRGPRD, but mediate their effects through specific Gα proteins and distinct downstream signaling pathways. The enantiomer produced by contracted muscle, L-BAIBA, regulates FGF23 through the Gαs/cAMP/PKA/CBP/β-catenin and Gαq/PKC/CREB signaling pathways at the early time point of 24 h. In contrast, D-BAIBA indirectly increases Fgf23 at a late time point by producing sclerostin via the Gαi/NF-κB signaling pathway. The exercise-induced BAIBA enantiomers add to the rising number of regulators of FGF23 gene expression and suggest that muscle not only consumes phosphate to maintain metabolism but actively participates in phosphate homeostasis by regulating Fgf23 in bone and renal excretion of phosphate by the kidneys.

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Lynda F. Bonewald (lbonewal@iu.edu).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• All data reported in this paper will be shared by the lead contact upon request. Raw data from Figures 1, 3, 5, 6, S4, and S6 were also deposited on Mendeley at DOI: https://doi.org/10.17632/tt2kthkr82.1.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mice

3-month-old male and female C57BL6/J mice were obtained from the Jackson Laboratory and aged to 4 months old at our animal facility before treadmill exercise experiments.

Mrgprd-KO mice in a C57BL/6 background were kindly provided by Dr. Anne-Marie Malfait (Rush University) with the permission of Dr. Mark Zylka (The University of North Carolina).⁸⁴ The entire open reading frame of Mrgprd was replaced with an in-frame fusion of EGFPf in the mouse model. After F3 generations of heterozygous breeding, homozygous breeding was performed to generate Mrgprd-KO mice required for the current project. Wild-type (WT) mice obtained by heterozygous breeding were also mated to generate WT control mice. Genotyping was performed using genomic DNA isolated from ear notch by polymerase chain reaction (PCR) with primers for Mrgprd and Gfp. 3-month-old Mrgprd-KO and WT male and female mice were used for ex vivo experiments, and 3 or 5-month-old WT male and female mice were used for in vivo experiments.

All animal experiments were performed according to an approved Institutional Animal Care and Use Committee protocol at Indiana University School of Medicine (IUSM), conforming to relevant federal guidelines (IACUC protocol #22051 and #22045). The IUSM animal facility is operated as a specific pathogen-free, AAALAC-approved facility. Animal care and husbandry meet the Guide for the Care and Use of Laboratory Animals by the National Research Council. Animals were group-housed and maintained on a 12-h light/dark cycle with ad libitum food and water at a constant temperature of 23°C. Daily inspections of health checks were performed by qualified veterinary staff and animal care technicians.

Cell culture of the mouse osteocyte like IDG-SW3 cell line

IDG-SW3 cells were cultured in an alpha-minimum essential medium (α-MEM, Hyclone, South Logan, UT) containing 10% fetal bovine serum (FBS, R&D systems, Minneapolis, MN), 100 U/ml penicillin (Gibco, Dublin, Ireland), 50 μg/mL streptomycin (Gibco), and 50 U/ml interferon-gamma (INF-γ, Life Technologies, Carlsbad, CA) on the type I collagen-coated flasks (Nunc, Roskilde, Denmark). Cells were maintained in sub-confluence at 33°C in a humidified atmosphere containing 95% air and 5% CO₂.

Isolation and culture of murine osteocyte-enriched bone chips

Long bones including femora, tibiae, and humeri were dissected from three-month-old WT and Mrgprd^ΔEGFPf male and female mice. Soft tissue was removed from bones and marrow was flushed out using a syringe and 27-gauge needle with Hank’s balanced salt solution (HBSS, Hyclone). Next, bones were cut into approximately 2 mm-sized pieces and digested with a series of collagenase and EDTA to remove the bone surface cells. Collagenase solution was prepared as 300 active U/ml Type I Collagenase (Sigma-Aldrich) dissolved in α-MEM and EDTA solution (5 mM, pH = 7.4, Sigma-Aldrich) was prepared in HBSS with 1% BSA (Sigma-Aldrich). All digestion steps were performed with a 4 mL solution and on an orbital shaker set to 200 rpm for 30 min in a 37°C and 5% CO₂ humidified incubator. Following the sequential digestion, bone fragments from individual mice were divided into 3 wells in a 12-well plate and cultured in α-MEM with 10% FBS and antibiotics overnight. Then, bone chips were used for experiments next day.

METHOD DETAILS

Synthesis of L- and D-BAIBA

After assessing the purity of commercially available L- and D-BAIBA and finding them only to be about 63% pure, in order to obtain pure L- and D-BAIBA, both were synthesized by the Indiana University School of Medicine Chemical Genomics Core. (S)-3-amino-2-methylpropanoic acid (L-BAIBA) and (R)-3-amino-2-methylpropanoic acid (D-BAIBA) were synthesized according to the literature.⁸⁵

General Procedures for Chemistry: All reagents and solvents were purchased from commercially available sources and used without further purification. NMR spectra were obtained on a Bruker 300 MHz NMR instrument. The chemical shifts are reported as ppm (δ) relative to the residual solvent peak. 1H NMR coupling constants (J) are reported in Hertz (Hz), and the multiplicities are indicated as follows: s = singlet; d = doublet; t = triplet; q = quartet; m = multiplet. HPLC purities under 214 nm were obtained on the Agilent 1290 LC instrument.

(S)-3-Amino-2-methylpropanoic acid (L-BAIBA)

Pd (150 mg, 10% weight on carbon) was added to a solution of Methyl (S)-3-(N, N-dibenzylamino)-2-methyl propanoate (300 mg, 1 mmol) in 6 mL MeOH, 0.6mL water and 0.15 mL AcOH. This suspension was stirred under 1 atm H₂ overnight and then filtered through Celite, and the filtrate was concentrated in vacuo. The residue was dissolved in THF (15.0 mL) and lithium hydroxide (213 mg, 5.1 mmol) in water (1.5 mL) was added. After stirring overnight, the solvent was removed in vacuo and the residue evaporated with 5mL aq. HCl (2 M). Purification via ion-exchange chromatography gives the title compound (92 mg, 88%) as a white solid. Melting point: 182°C–184oC, 1H NMR (300 MHz, d4-CD3OD): δ 2.96–2.93 (m, 2H), 2.47 (m, 1H), 1.20 (m, 3H). 13C NMR (75 MHz, d4-CD3OD): δ 179.9, 42.5, 38.6, 14.9. Purity >95% (λ 214 nm).

(R)-3-Amino-2-methylpropanoic acid (D-BAIBA)

Pd (150 mg, 10% weight on carbon) was added to a solution of Methyl (R)-3-(N, N-dibenzylamino)-2-methyl propanoate (300 mg, 1 mmol) in 6mL MeOH, 0.6 mL water and 0.15 mL AcOH. This suspension was stirred under 1 atm H₂ overnight and then filtered through Celite, and the filtrate was concentrated in vacuo. The residue was dissolved in THF (15.0 mL) and lithium hydroxide (213 mg, 5.1 mmol) in water (1.5 mL) was added. After stirring overnight, the solvent was removed in vacuo and the residue evaporated with 5mL aq. HCl (2 M). Purification via ion-exchange chromatography gives the title compound (94 mg, 90%) as a white solid. 1H NMR (300 MHz, d4-CD3OD): δ 2.96–2.91 (m, 2H), 2.47 (m, 1H), 1.20 (m, 3H). 13C NMR (75 MHz, d4-CD3OD): δ 181.1, 43.9, 39.9, 16.2. Purity 93.4% (λ 214 nm).

The purity of the chemically synthesized compounds was also verified using our previously published method.³¹ 10 μL of samples were mixed with solvents containing internal standards and centrifuged with a Max-RP 50 × 2.0 mm guard column (Phenomenex, Torrance, CA) to precipitate proteins. Supernatants were transferred to the autosampler inserts (SIL-30AC, Shimadzu Scientific Instruments, INC., Tokyo, Japan) for LC-MS/MS analysis using a triple quadrupole liquid chromatograph-mass spectrometer (Shimadzu LCMS-8050). All analyses and data processing were completed on Shimadzu LabSolutions v5.91 software. The purity of each chemical synthesis is shown in Figure S1.

In vivo animal experiments

Treadmill exercises were performed to investigate whether exercise elevates BAIBA enantiomers, intact FGF23, and phosphate excretion. We obtained 3-month-old male and female C57BL6/J mice from the Jackson Laboratory and aged them to 4 months old at our animal facility. Mice were acclimated to a rodent treadmill (Columbus Instruments, Columbus, OH) with three 15-min sessions (8 m/min with 0° incline) one week prior to the exercise sessions. We performed a treadmill exercise bout, which included 1) a 5-min warm-up session at 8 m/min with a 0° incline and 2) a 5-min exercise interval at 15 m/min with a 10° incline followed by a 1-min rest interval at 8 m/min with 0° incline. We repeated for eight exercise intervals, which was a total of 53 min. Negative reinforcement was provided by an electrical stimulus (1.2 mA) at the end of the treadmill to encourage mice to run. No mice were expelled from the study, although pre-determined exclusion criteria were set to identify mice that were unable to meet the demands of exercise. All the acclimations and exercise sessions were performed in the morning, and mice were returned to their cage with free access to food and water until euthanasia 1 h after the treadmill exercise bout.

In dose experiment, three-month-old mice were treated with control vehicle (0.9% saline) or L-BAIBA with different concentrations of 1, 10, 50, and 100 mg/kg/day for three days. 100 mg/kg/day of BAIBA enantiomers were administered for 5-day D-BAIBA treatment and 3-day L-BAIBA treatment with five-month-old mice. The enantiomers were administered in drinking water, which was replaced freshly every day for three or five days. Daily water intake and body weight were recorded to measure the average intake of BAIBA enantiomers. We conducted overnight fasting, which was intended to reduce diet effects on phosphate levels, similarly performed in human examinations.^86,87 However, we noticed significant reductions in body weight and water intake on the last day of experiments due to overnight fasting. Therefore, we did not perform overnight fasting for experiments with five-month-old young adult mice. After treatments, urine, whole blood, and hindlimbs (femora and tibiae) were harvested from mice. Soft tissues such as muscles and connective tissues were removed from hindlimb bones, and the bone surface cells were removed by scratching with a surgical blade. Both sides of epiphysis were cut, and bone marrow was flushed out using a syringe and 27G-needle with PBS (−), and then cleaned bones were snap-frozen in liquid nitrogen. Whole blood was collected by cardiac puncture under anesthesia. To measure the circulating intact FGF23 and phosphate levels, plasma was harvested using EDTA-coating tubes. The plasma and urine samples were snap-frozen in liquid nitrogen and stored at −80°C until use.

In vitro experiments

IDG-SW3 cells were seeded at a density of 4×10⁴ cells/cm² in collagen-coated 12 or 24-well plates (Corning, Corning, NY). When the cells reached complete confluency, the culture media was replaced with osteogenic media (α-MEM with 10% FBS, 100 U/ml penicillin, 50 μg/mL streptomycin, 50 μg/mL ascorbic acid, and 4 mM β-glycerophosphate) (Sigma-Aldrich, St. Louis, MO) and cells were cultured at 37°C and 8% CO₂. Culture media was replaced every three days. After 28 days of culture, cells were treated with L- and D-BAIBA (1–20 μM) for 24 or 72 h. To investigate the signaling pathways utilized by L- and D-BAIBA, Day 28 IDG-SW3 cells were pre-treated with an MRGPRD receptor antagonist, MU6840 (10, 25, and 50 μM, Key Organics, Cornwall, UK), Melittin, a Gαs inhibitor (10 and 100 nM, Tocris Bioscience, Bristol, UK), Ebselen, a Gαq inhibitor (10 and 100 nM, Tocris Bioscience), Pertussis toxin, a Gαi inhibitor (1 and 10 ng/mL, Cayman Chemical, Ann Arbor, MI), PKI5–24, a PKA inhibitor (1, 10 and 50 nM, Alfa Aesar, Lancashire, UK), Go6983, a PKC inhibitor (10 and 100 nM, Tocris Bioscience), the CBP/β-catenin inhibitor, ICG-001 (10, 100, and 1000 nM, Tocris Bioscience), and the NF-κB inhibitor, BMS-345541 (100 and 500 nM, Sigma-Aldrich) for 1 h and then, cells were treated with L-BAIBA and D-BAIBA for the indicated times.

Ex vivo experiments

To validate the results obtained with the IDG-SW3 cells, ex vivo, bone chips were treated with 100 and 200 μM L- and D-BAIBA for 24 or 72 h under shaking at 10–12 times/min. After treatment, bone fragments were washed with HBSS, snap-frozen in liquid nitrogen, and kept at −80°C until use. Frozen bone fragments were pulverized before RNA extraction.

Micro-computed tomography (μCT) analysis of femurs

To investigate the bone microarchitecture of Mrgprd-KO, formalin-fixed femurs were scanned at a 9 μm voxel size using a Bruker SkyScan 1176 high-resolution μCT system (50 kV, 500 mA, 0.5 mm Aluminum filter). Images were reconstructed using NRecon software. Reconstruction parameters were identical for all scans: 2 for smoothing (Gaussian), ring artifact correction of 5, and beam hardening correction at 20. Reconstructed images were then rotated to ensure consistent cross-sectional analysis with DataViewer and saved in a transaxial orientation. A 1 mm trabecular region of interest (ROI) was selected in the distal metaphysis of femurs, starting at 0.5 mm distal to the growth plate and extending proximally. The trabecular microarchitecture within the ROI was segmented and assessed using CTAn with a threshold value of 80 and 255. For cortical bone analysis, a 0.1 mm cortical ROI was analyzed at the midshaft of the femurs with a threshold value of 120 and 255.

Whole bone mechanical testing of femurs

To investigate bone mechanical properties of Mrgprd-KO, parameters related to whole-bone strength were measured using three-point bending tests.⁸⁸ Each femur was stripped of soft tissue, wrapped in PBS-soaked gauze, and stored at −20°C until use. They were thawed to room temperature and loaded to failure in monotonic compression using a three-point bending platen. The lower span points were spaced 10 mm apart, and the upper point contacted the femoral diaphysis at the midshaft. During each test, force, and displacement were collected every 0.01 s. From the force/displacement curves, ultimate force and energy to failure were calculated using standard equations.⁸⁹

Quantification of urine and plasma biochemical parameters

Right after the treadmill exercise, the mice were restrained over a urine collection device and lightly stroked around the belly to encourage urination. The urine was collected with a pipette and transferred to a collection tube. Whole blood was also collected from the mice, either from a facial vein or by cardiac puncture under anesthesia, using commercially available EDTA-treated collection tubes for plasma. After collecting the whole blood, it was centrifuged at 1,000 × g for 15 min in a refrigerated centrifuge, and the resulting supernatant was placed into a collection tube using a pipette. Both the urine and blood samples were kept at −80C until analysis. Urine and plasma biochemical parameters were assessed according to the previous publication.⁷⁷ Briefly, the urine and blood levels of inorganic phosphate were measured using COBAS Mira Plus automated chemistry analyzer (Roche Diagnostics, Pleasanton, CA) and phosphate assay kit (Abcam), and urine creatinine was measured by colorimetric assay (BioAssay Systems, Hayward, CA). To assess the urinary phosphate excretion, phosphate levels were normalized by urine creatinine. However, the phosphate levels in urines obtained from the treadmill exercises were not normalized by creatinine since exercise also elevated creatinine levels (Figure 1C). The concentration of intact FGF23 was measured using an ELISA kit (Quidel, San Diego, CA).

Quantification of L- and D-BAIBA

Chemicals and reagents

Aminobutyric acid standard compounds (S)-3-aminoisobutyric acid (L-BAIBA) and (R)-3-aminoisobutyric acid (D-BAIBA) were purchased from AdipoGen Life Sciences. (San Diego, CA). Isotopic internal standard (IS) compounds (±)-3-amino-iso-butyric-2,3,3-d3 acid (D,L-BAIBA-d3) were obtained from CDN Isotopes (PointeClaire, Quebec, Canada). Formic acid (reagent grade, ≥95%) and BSA were obtained from Sigma-Aldrich (St. Louis, MO). Phosphate Buffered Saline (PBS) was purchased from Fisher Scientific (Pittsburgh, PA). HPLC–MS grade acetonitrile, water, and methanol were purchased from J.T. Baker (Phillipsburg, NJ).

LC-MS/MS conditions

All components of liquid chromatography-tandem mass spectrometry (LC–MS/MS) system are from Shimadzu Scientific Instruments, Inc. (Columbia, MD). The LC system was equipped with pumps A and B (LC-30AD), and autosampler (SIL-30AC). The LC separation was conducted on a chiral SPP-TeicoShell column (150 × 4.6 mm, 2.7 μm, AZYP LLC., Arlington, TX) configured with a Synergi 4 μm Max-RP column as guard column (50 × 2.0 mm, Phenomenex, Torrance, CA). The MS/MS analysis was performed on Shimadzu LCMS-8050 triple quadrupole mass spectrometer. Quantification of L- and D-BAIBA in murine serum samples was performed using the LC-MS/MS method.^31,90 Mobile phases are methanol (A) and water containing 0.005% formic acid and 2.5 mM ammonium formate (B). The MS instrument was operated and optimized under positive electrospray (+ESI) and multiple reaction monitoring modes (MRM). The m/z transitions (precursor to product ions) and their tuning voltages were selected from published paper³¹ and further optimized based on the best MRM responses from instrumental method optimization software. All analyses and data processing were completed on Shimadzu LabSolutions V5.91 software (Shimadzu Scientific Instruments, Inc., Columbia, MD).

Sample preparation for LC-MS/MS analysis

Ten microliter mouse serum samples and same volume of IS mixture solution (1.2 μM, 0.1% formic acid in methanol, v/v) were added to 35 μL 0.1% (v/v) formic acid in methanol, followed by 20 min-shaking at room temperature and another 15 min-centrifugation at 15,000×g, 4°C to precipitate the proteins. The supernatant was directly transferred to autosampler vial and 45 μL of each sample was injected for LC-MS/MS analysis. The samples of standard calibration curves were prepared by spiking the pure standards in surrogate matrix 5% (w/v) BSA in PBS (pH7.4). The samples for ten-point calibration curves were prepared by diluting the working solution to 0.02–10.24 μM for L-BAIBA and D-BAIBA. Then ten microliters of each standard sample were taken and treated following the same preparation procedures of serum samples for LC-MS/MS analysis.

RNA isolation and real-time qPCR

Total RNA was isolated from IDG-SW3 cells and bone samples using Trizol (Invitrogen, Waltham, MA) and purified using a Direct-zol RNA miniprep kit (Zymo Research, Irvine, CA) according to the manufacturer’s protocol. The isolated RNA was treated with DNase I (Life Technologies) and used for reverse-transcription into cDNA by a high-capacity cDNA transcription kit (Life Technologies). All amplifications were performed using the Step One Plus cycler (Life Technologies) with TaqMan arrays and Power Up SYBR Green Master Mix (Applied Biosystems, Waltham, MA). The data was normalized against the housekeeping genes, Actb and B2m. Relative expression levels were determined by the ΔΔ Ct method.

Western blotting

IDG-SW3 cells were cultured on 12-well plates with the differentiation media for 28 days. Then, cells were treated with L-BAIBA and D-BAIBA (10 μM) over a 72-h time course. After treatment, culture media was removed, and cells were washed with ice-cold PBS (−) 3 times. And then, the cells were lysed in ice-cold RIPA buffer (Boston Bioproducts, Ashland, MA) containing protease inhibitor cocktail (Sigma-Aldrich, St. Louis, MO) and phosphatase inhibitor (Sigma-Aldrich). Next, the lysed cells were centrifuged at 10,000g and 4°C for 15 min, and supernatants were collected for western blotting. The amount of total protein was quantified using the Pierce BCA Protein Assay kit (Thermo Scientific, Rockford, IL). 30 or 50 mg of total protein from each sample was electrophoretically separated on 10% Mini-Protean TGX Precast Protein Gel (Bio-rad, Hercules, CA) by SDS-PAGE. The separated proteins were transferred to Immobilon-P PVDF membrane (Millipore Sigma, Burlington, MA) for 3 h at a constant 50 V. Next, the membrane was incubated with 5% BSA for 1 h at room temperature. After blocking, the membrane was incubated with an anti-GPCR TGR7 antibody (1/1,000 dilution, Abcam, Cambridge, UK), anti-β-catenin antibody (1/1,000 dilution, Cell Signaling Technology, Danvers, MA), anti-(Ser675) phospho β-catenin antibody (1/1,000 dilution, Cell Signaling Technology), anti-CREB antibody (1/1,000 dilution, Cell Signaling Technology), anti-(Ser133) phospho CREB antibody (1/1,000 dilution, Cell Signaling Technology), anti- NF-κB p65 antibody (1/1,000 dilution, Cell Signaling Technology), and anti-(Ser536) phospho NF-κB p65 antibody (1/1,000 dilution, Cell Signaling Technology) at 4°C for overnight. Next, the blotting membrane was washed with Tris-buffered saline (TBS) containing 0.1% tween 20 (Thermo Scientific) and incubated with horseradish peroxidase (HRP) conjugated goat anti-rabbit IgG (1/2,000 dilution, Invitrogen) for 1 h at room temperature. Immunoreactive bands were detected using the SuperSignal West Dura Extended duration substrate (Thermo Scientific) and visualized using a Celvin S chemiluminescence imager (Bio step, Burkhardtsdorf, Germany). Densitometry of bands was quantified using ImageJ version 1.5.3 (National Institute of Health, USA).

Quantification of intracellular cAMP

IDG-SW3 cells were seeded at a density of 4×10⁴/cm² on the 24-well plates and differentiated for 28 days. First, cells were pre-treated with 100 μM 3-isobutyl-1-methylxanthine (IBMX, Tocris Bioscience) to stabilize intracellular cAMP for 30 min at 37°C. Then, cells were cultured with L-BAIBA and D-BAIBA over a 24-h time course. Intracellular cAMP levels were measured using the LANCE cAMP kit (PerkinElmer, Waltham, MA).⁹¹ Cells were rinsed with PBS (−) and lysed with ice-cold detection buffer with gentle agitation at 4°C for 30 min to ensure cell lysis. Next, 10 μL of cell lysates were reacted with an anti-cAMP antibody labeled with ULightTM in a detection buffer for 30 min. The samples were then mixed with 20 μL of the tracer mixture containing the Eu-labeled streptavidin and the Biotin-cAMP and incubated with gentle agitation at room temperature for 1 h. The fluorescence was measured with excitation at 340 nm and emission at 615 nm. The intracellular cAMP levels were normalized by the total amount of protein quantified using the Pierce BCA assay kit.

Enzyme-linked immunosorbent assay (ELISA)

To quantify the secreted sclerostin levels from osteocytes, IDG-SW3 cells were treated with L-BAIBA and D-BAIBA for 24 h, and conditioned media were harvested. According to the manufacturer’s instructions, sclerostin levels in cultured media were measured using a Mouse/Rat Sclerostin Quantikine ELISA kit (R&D Systems).

Sclerostin blocking antibody

To investigate whether sclerostin regulates Fgf23 expression, 28 days of differentiated IDG-SW3 cells were treated with 10 μM D-BAIBA in the presence of 100–1000 pg/mL of sclerostin neutralizing antibody (AbD09097-h/m IgG2a, Bio-Rad). After 24 h, AbD09097 with the same concentrations was additionally applied. To assess the effect of AbD09097 on Fgf23 gene expression, total RNA was purified for RT-qPCR analysis described above.

QUANTIFICATION AND STATISTICAL ANALYSIS

The data were presented as mean ± standard deviation (SD). An unpaired t test was carried out for comparisons between two groups of normally distributed data, confirmed by the Shapiro-Wilk test. Outliers identified by Grubb’s test were removed. For datasets that were not normally distributed, the Mann-Whitney test was used to analyze two groups of data. For statistical comparisons between more than two groups of datasets with normal distribution, one-way analysis of variance (ANOVA) was used after Grubb’s test. It was followed by Dunnet’s or Tukey’s post hoc test. Additionally, two-way ANOVA with Tukey’s post-hoc test was used to analyze the effect of two independent variables on a dependent variable. A value of p < 0.05 was considered significant. They were labeled as follows: *: p < 0.05, **: p < 0.01, ***: p < 0.001, and ****: p < 0.0001. All statistical calculations were done using the GraphPad PRISM 10.2.0 software (La Jolla, CA).

To measure the strength of the linear relationship between two variables, the Pearson correlation analysis was performed. A p-value <0.05 was considered statistically significant. All statistical calculations were done using the GraphPad PRISM 10.2.0 software.

All the statistical details of experiments can be found in the figure legends, including the statistical tests used, exact value of n, and what n presents.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit anti-GPCR TGR7	Abcam	Ab79262; RRID: AB_2145276
Rabbit anti-(Ser675) phospho β-catenin	Cell Signaling Technology	#4176; RRID: AB_1903923
Rabbit anti-β-catenin	Cell Signaling Technology	#8480; RRID: AB_11127855
Rabbit anti-(Ser133) phospho CREB	Cell Signaling Technology	#9198; RRID: AB_2561044
Rabbit anti-CREB	Cell Signaling Technology	#9104; RRID: AB_490881
Rabbit anti-(ser536) phosphor NF-κB p65	Cell Signaling Technology	#3033; RRID: AB_331284
Rabbit anti-NF-κB p65	Cell Signaling Technology	#8242; RRID: AB_10859369
HRP conjugated goat anti-Rabbit IgG	Invitrogen	#31460; RRID: AB_228341
Sclerostin antibody AbD09097-h/m IgG2a	Bio-Rad	HCA230Z
Chemicals, peptides, and recombinant proteins
L-BAIBA	This paper	N/A
D-BAIBA	This paper	N/A
MEM Alpha Modification, with L-Glutamine	Hyclone	SH30265.01
Fetal Bovine Serum	R&D systems	S11150
Penicillin-Streptomycin (10,000U/mL)	Gibco	15140–122
L-Ascorbic Acid	Sigma	A4544
β-Glycerophosphate disodium salt hydrate	Sigma	G9422
MU6840	Key Organics	5R-0328
Melittin	Tocris Bioscience	1193
Ebselen	Tocris Bioscience	5245
Pertussis toxin	Cayman Chemical	70323–44-3
PKI5–24	Alfa Aesar	AAJ66438MCR
Go6983	Tocris Bioscience	2285
ICG-001	Tocris Bioscience	4505
BMS-345541	Sigma Aldrich	B9935
Collagenase, from Clostridium histolyticum Type IA	Sigma-Aldrich	C9891–5G
Ethylenediaminetetraacetic acid, tetrasodium salt	Sigma-Aldrich	10378–23-1
Albumin, from bovine serum	Sigma-Aldrich	A7030–500G
TRIzol reagent	Invitrogen	15596026
DNase I	Life Technologies	18068015
RIPA buffer	Boston Bioproducts	BP-115X
Protease inhibitor cocktail	Sigma-Aldrich	PPC2020
Phosphatase inhibitor	Sigma-Aldrich	P0001
Tween 20	Thermo Fisher Scientific	BP337–500
IBMX	Tocris Bioscience	2845
(S)-3-aminoisobutyric acid	AdipoGen Life Sciences	CDX-A0147
(R)-3-aminoisobutyric acid	AdipoGen Life Sciences	CDX-A0148
(±)-3-amino-iso-butyric-2,3,3-d3	CDN Isotopes	D-7229
Formic acid	Sigma-Aldrich	64–18-6
Bovine serum albumin	Sigma-Aldrich	A7906–50G
Phosphate buffered saline	Fisher Scientific	MT21040CV
HPLC-MS grade acetonitrile	J.T. Baker	75–05-8
HPLC-MS grade water	J.T. Baker	7732–18-5
HPLC-MS grade methanol	J.T. Baker	67–56-1
Ammonium formate	Honeywell Research Chemicals	540–69-2
Critical commercial assays
Mouse FGF23 ELISA	R&D systems	60–6500
Mouse/Rat Sclerostin Quantikine ELISA kit	R&D systems	MSST00
Phosphate assay kit	Abcam	ab65622
Creatinine assay kit	Bio Assay Systems	DICT-500
Direct-zol RNA Miniprep kit	Zymo Research	R2052
High-capacity cDNA transcription kit	Life Technologies	4388950
Power Up SYBR Green Master Mix	Applied Biosystems	A25742
BCA Protein Assay Kit	Thermo Fisher Scientific	J63283.QA
10% Mini-Protean TGX Precast Protein Gel	Bio-Rad	4561034
Immobilon-P PVDF membrane	Millipore Sigma	IPVH15150
SuperSignal West Dura Extended duration substrate	Thermo Fisher Scientific	34075
LANCE cAMP assay kit	PerkinElmer	AD0262
Deposited data
Raw data from Figures 1,3, 5, 6, S4, and S6	This paper	Mendeley: DOI: https://doi.org/10.17632/tt2kthkr82.1
Experimental models: Cell lines
IDG-SW3	Woo et al.45	N/A
Experimental models: Organisms/strains
Mouse: MRGPRDΔegepf	Zylka et al.84	N/A
Oligonucleotides
Taqman assay probe for Fgf23	Thermo Fisher Scientific	Mm00445621_m1
Taqman assay probe for Sost	Thermo Fisher Scientific	Mm00470479_m1
Taqman assay probe for Pdpn	Thermo Fisher Scientific	Mm00494716_m1
Taqman assay probe for Dmp1	Thermo Fisher Scientific	Mm00803831_m1
Taqman assay probe for Mepe	Thermo Fisher Scientific	Mm02525159_s1
Taqman assay probe for Phex	Thermo Fisher Scientific	Mm00448119_m1
Taqman assay probe for Tnsf11	Thermo Fisher Scientific	Mm00441906_m1
Taqman assay probe for Tnsfllb	Thermo Fisher Scientific	Mm00435454_m1
PCR primer for B2m Forward; ACAGTTCCACCCGCTCACATT	Integrated DNA Technologies	N/A
PCR primer for B2m Reverse; TAGAAAGACCAGTCCTTGCTGAAG	Integrated DNA Technologies	N/A
PCR primer for Actb Forward; GGCTGTATTCCCCTCCATCG	Integrated DNA Technologies	N/A
PCR primer for Actb Reverse; CCAGTTGGTAACAATGCCATGT	Integrated DNA Technologies	N/A
PCR primer for Mrgprd Forward; TTTTCAGTGACATTCCTCGCC	Integrated DNA Technologies	N/A
PCR primer for Mrgprd Reverse; GCACATAGACACAGAAGGGAGA	Integrated DNA Technologies	N/A
PCR primer for Gfp Forward; GCACGACTTCTTCAAGTCCGCCATGCC	Integrated DNA Technologies	N/A
PCR primer for Gfp Reverse; GCGGATCTTGAAGTTCACCTTGATGCC	Integrated DNA Technologies	N/A
Software and algorithms
ImageJ	National Institutes of Health	ver1.53
Prism8	GraphPad	ver8.0.2
Other
Celvin S 320 chemiluminescence imager	Biostep	BG56-A0320

Highlights.

• Exercise-induced L-BAIBA correlates with intact FGF23 and urine phosphate in mice

• Both L- and D-BAIBA induce Fgf23 in osteocytes in vitro and ex vivo

• Both BAIBA enantiomers use the same receptor but activate distinct pathways

• Both BAIBA enantiomers increase Fgf23 in bone and urine phosphate in mice