MiR‐103‐3p targets the m6A methyltransferase METTL14 to inhibit osteoblastic bone formation

Impaired osteoblast function is involved in osteoporosis, and microRNA (miRNA) dysregulation may cause abnormal osteoblast osteogenic activity. However, the influence of miRNA on osteoblast activity and the underlying mechanisms remain elusive. In this study, miR‐103‐3p was found to be negatively correlated with bone formation in bone specimens from elderly women with fractures and ovariectomized (OVX) mice. Additionally, miR‐103‐3p directly targeted Mettl14 to inhibit osteoblast activity, and METTL14‐dependent N6‐methyladenosine (m6A) methylation inhibited miR‐103‐3p processing by the microprocessor protein DGCR8 and promoted osteoblast activity. Moreover, miR‐103‐3p inhibited bone formation in vivo, and therapeutic inhibition of miR‐103‐3p counteracted the decreased bone formation in OVX mice. Further, METTL14 was negatively correlated with miR‐103‐3p but positively correlated with bone formation in bone specimens from elderly women with fractures and OVX mice. Collectively, our results highlight the critical roles of the miR‐103‐3p/METTL14/m6A signaling axis in osteoblast activity, identifying this axis as a potential target for ameliorating osteoporosis.


| INTRODUC TI ON
Osteoporosis is a common disease among elderly women and is characterized by reduced bone mass and abnormal microarchitecture, resulting in fragility fractures (Rachner et al., 2011). Emerging evidence indicates that osteoblasts directly affect the entire bone remodeling process and that impaired osteoblast activity plays fundamental roles in bone metabolic disorder (Harada, & Rodan, 2003).
Moreover, the mechanisms that mediate impaired osteoblast activity remain unclear and merit further research. microRNAs (miRNAs) are small single-stranded noncoding RNA molecules involved in silencing and post-transcriptional regulation of gene expression, thereby mediating many biological processes (Rigoutsos, & Furnari, 2010;Sun, & Lai, 2013). Many miRNAs have been characterized to regulate osteoblast activity and osteoblastic bone formation (Inose et al., 2009;Wang et al., 2013;Xu et al., 2018). Our previous study revealed that miR-103-3p could inhibit L-type calcium channel currents and osteoblast proliferation under simulated microgravity conditions, primarily by suppressing Cav1.2 expression .
Other researchers showed that miR-103-3p was downregulated under cyclic mechanical loading conditions and found that this alteration was adversely correlated with osteoblast differentiation and bone formation in response to mechanical stimulation (Zuo et al., 2015). In addition, we found that mature miR-103-3p was evolutionarily conserved among several species and highly expressed in bone tissue ( Figure S1a,b). However, the influence of miR-103-3p on osteoblastic bone formation in postmenopausal osteoporosis and the possible underlying mechanisms have not been confirmed.
In this study, we tested the expression of miR-103-3p and bone formation marker genes in bone specimens from elderly women with fractures and in OVX mice and found that miR-103-3p was negatively correlated with bone formation. We present in vitro evidence demonstrating that miR-103-3p directly targets Mettl14 to functionally inhibit osteoblast activity and that METTL14-dependent m 6 A methylation provides negative feedback to regulate miR-103-3p processing by the microprocessor protein DGCR8, thereby modulating osteoblast activity. Moreover, we demonstrate that miR-103-3p inhibits bone formation in vivo and that therapeutic inhibition of miR-103-3p counteracts the decrease in bone formation in OVX mice.
Further, we confirmed that METTL14 is negatively correlated with miR-103-3p but positively correlated with bone formation in bone specimens from elderly women with fractures and in OVX mice.

| High miR-103-3p expression is correlated with reduced bone formation capacity
To examine the expression pattern of miR-103-3p and bone formation marker genes in human bone tissues, we collected bone specimens from 24 elderly female patients with fractures ( Figure 1a and Tables S1 and S2) and examined miR-103-3p, ALP (alkaline phosphatase), BGLAP (osteocalcin), and COL1α1 (collagen type 1α1) expression in these bone specimens by q-PCR. The q-PCR analysis showed that miR-103-3p levels in osteoporosis patients (T ≤ −2.5) F I G U R E 1 miR-103-3p levels are negatively correlated with bone formation capacity in human bone specimens, mouse bone specimens, and primary mouse osteoblasts. (a) Schematic diagram illustrating the experimental design. (b) Real-time PCR analysis of the bone mineral density-related changes in miR-103-3p levels in bone specimens from patients. The relative miRNA levels were normalized to the mean value of the T > −2.5 group. U6 small nuclear RNA was used as the internal control. (c) Real-time PCR analysis of the bone mineral densityrelated changes in the mRNA levels of the bone formation marker genes ALP (left), BGLAP (middle), and COL1α1 (right) in bone specimens from patients. The relative mRNA levels were normalized to the mean value of the T > −2.5 group. Human GAPDH mRNA was used as the internal control. (d) Correlation analysis between the miR-103-3p level and ALP (left), BGLAP (middle), and COL1α1 (right) mRNA levels in bone specimens from patients. (e) Schematic diagram illustrating the experimental design. (f) Real-time PCR analysis of the changes in miR-103-3p levels in bone specimens from Sham or OVX mice. The relative miRNA levels were normalized to the mean value of the Sham group. U6 small nuclear RNA was used as the internal control. (g) Real-time PCR analysis of the changes in the mRNA levels of the bone formation marker genes Alp (left), Bglap (middle) and Col1α1 (right) in bone specimens from Sham or OVX mice. The relative mRNA levels were normalized to the mean value of the Sham group. Mouse Gapdh mRNA was used as the internal control. (h) Correlation analysis between the miR-103-3p level and Alp (left), Bglap (middle), and Col1α1 (right) mRNA levels in bone specimens from Sham or OVX mice. (i) Schematic diagram illustrating the experimental design. (j) Real-time PCR analysis of Alp (left), Bglap (middle), and Col1α1 (right) mRNA levels in primary mouse osteoblasts during osteoblast maturation (n = 4). (k) Real-time PCR analysis of miR-103-3p levels in primary mouse osteoblasts during osteoblast maturation (n = 4). The n value for each group is indicated at the bottom of each bar in the graphs. All data are the mean ± SD. *p < 0.05, **p < 0.01. One-way analysis of variance (ANOVA) with a post hoc test was performed, and the significance of differences between two groups was determined with Student's t test. For statistical correlation, Pearson's correlation coefficient was used F I G U R E 2 miR-103-3p inhibits osteoblast activity in vitro. (a) Real-time PCR analysis of miR-103-3p levels in primary mouse osteoblasts after treatment with 300 μM agomir-103-3p, antagomir-103-3p, or the corresponding negative controls for 48 h (n = 3). (b) WST-8 assay of changes in cell growth in each group at 24-96 h after treatment with 300 μM agomir-103-3p, antagomir-103-3p or the corresponding negative controls (n = 4). (c) EdU incorporation assay of the proliferation of primary mouse osteoblasts in each group after treatment with 300 μM agomir-103-3p, antagomir-103-3p or the corresponding negative controls for 48 h (n = 3). Cells were staining with the nucleic acid dye Hoechst (blue) and EdU (green). Scale bar, 10 μm. (d) Real-time PCR analysis of the changes in the mRNA levels of the osteoblast differentiation marker genes Alp (left), Bglap (middle), and Col1α1 (right) in primary mouse osteoblasts after treatment with 300 μM agomir-103-3p, antagomir-103-3p, or the corresponding negative controls for 48 h (n = 4). (e) ALP activity (left) and the amount of BGLAP protein (middle) and collagen I (right) in the supernatant of primary mouse osteoblasts after treatment with 300 μM agomir-103-3p, antagomir-103-3p, or the corresponding negative controls for 48 h (n = 4). (f) Representative images of ALP staining of primary mouse osteoblasts after treatment with 300 μM agomir-103-3p, antagomir-103-3p, or the corresponding negative controls for 48 h (n = 3). Scale bar, 10 mm. (g) Alizarin red staining of calcium deposition in primary mouse osteoblasts after treatment with 300 μM agomir-103-3p, antagomir-103-3p, or the corresponding negative controls in osteogenic medium for 21 days. Scale bar, 10 mm. All data are presented as the mean ± SD. **p < 0.01. One-way ANOVA with a post hoc test was performed, and the significance of differences between two groups was determined with Student's t test were higher than in control patients (T > −2.5) ( Figure 1b) and that the levels of ALP, BGLAP, and COL1α1 in osteoporosis patients (T ≤ −2.5) were lower than in control patients (T > −2.5) (Figure 1c).
Mature miR-103-3p is evolutionarily conserved among several species ( Figure S1a). Thus, we further investigated the relationship between the miR-103-3p level and bone formation marker genes in bone tissues from OVX mice (Figure 1e and Figure S1c-e). We found high levels of miR-103-3p and low levels of Alp, Bglap, and Col1α1 in bone tissues from OVX mice (Figure 1f,g). Consistently, we found that miR-103-3p levels were negatively correlated with Alp, Bglap, and Col1α1 levels in these mouse samples (Figure 1h).
To further verify the relationship between miR-103-3p and osteoblast differentiation marker genes in osteoblasts, we derived and cultured primary osteoblasts from postnatal mouse calvarias
To test the influence of miR-103-3p on osteoblast differentiation, we examined the levels of osteoblast differentiation markers and ALP staining in primary mouse osteoblasts. Compared to those in the corresponding control treatment groups, the Alp, Bglap, and To examine the effects of miR-103-3p on osteoblast matrix mineralization, we performed an Alizarin red staining assay and demonstrated that agomir-103-3p decreased mineral deposition whereas antagomir-103-3p increased mineral deposition compared to that in the corresponding control treatment groups ( Figure 2g). To determine the effects of miR-103-3p on osteoclast activity, an in vitro osteoclastogenesis assay was performed. The data revealed that miR-103-3p had no effect on osteoclast activity ( Figure S2e-g).

| miR-103-3p directly targets Mettl14 to functionally inhibit osteoblast activity
To investigate the specific mechanism by which miR-103-3p regulates osteoblast activity, TargetScan, miRDB, and miRBase were used to predict the potential targets of miR-103-3p. The results showed that among the candidate target genes, Mettl14 has a miR-103-3p binding site in its 3′ untranslated region (UTR) To explore the potential role of miR-103-3p in m 6 A modification in osteoblasts, we examined the level of m 6 A in the total RNA and found that agomir-103-3p resulted in decreased m 6 A levels in osteoblasts and that antagomir-103-3p reversed the reduction in m 6 A levels ( Figure 3f). The results from the Empty vector and mock transfection are also shown. All data are the mean ± SD. *p < 0.05, **p < 0.01. One-way ANOVA with a post hoc test was performed, and the significance of differences between two groups was determined with Student's t test F I G U R E 4 METTL14-dependent m 6 A methylation regulates miR-103-3p processing by the microprocessor protein DGCR8 and modulates osteoblast activity in vitro. (a) Coimmunoprecipitation (IP) of the METTL14-interacting protein DGCR8. Western blotting with anti-DGCR8 and anti-METTL14 antibodies and immunoglobulin G (IgG) antibody was used as a control for IP. (b) IP of DGCR8, METTL14, and associated RNA from control MC3 T3-E1 cells or METTL14-overexpressing MC3 T3-E1 cells. The cells were UV-cross-linked before IP. Western blotting or immunoblotting was conducted using the antibodies described above. (c) Real-time PCR analysis of pri-miR-103-3p binding to DGCR8 in IP assay of DGCR8-associated RNA from control and METTL14-overexpressing MC3 T3-E1 cells (n = 3). Pri-let-7e was used as a positive control. (d) Real-time PCR analysis of the pri-miR-103-3p m 6 A modification level determined by IP of m 6 A-modified miRNA in control or METTL14-overexpressing MC3 T3-E1 cells (n = 3). Pri-let-7e was used as a positive control. (e) The effect of oeMettl14, siRNA-Mettl14, or the corresponding negative controls on the m 6 A content in total RNA (n = 3). (f) Real-time PCR analysis of pri-miR-103-3p (left), pre-miR-103-3p (middle), and miR-103-3p (right) in METTL14 knockdown and overexpression MC3 T3-E1 cells (n = 3). (g) Western blot analysis of the overexpression efficiency of Mettl14 plasmid after treatment of primary mouse osteoblasts with a specific Mettl14 plasmid (oeMettl14) or an empty vector (n = 3). (h) WST-8 assay of changes in primary mouse osteoblast growth at 24-96 h after treatment with oeMettl14, siRNA-Mettl14 or the corresponding negative controls (n = 3). (i) ALP activity (left) and the amount of BALP protein (middle) and collagen I (right) in the supernatant of primary mouse osteoblasts after treatment with oeMettl14, siRNA-Mettl14 or the corresponding negative controls (n = 3). (j) Representative images of ALP staining of primary mouse osteoblasts after treatment with oeMettl14, siRNA-Mettl14, or the corresponding negative controls (n = 3). Scale bar, 10 mm. (k) Alizarin red staining of calcium deposition in primary mouse osteoblasts after treatment with oeMettl14, siRNA-Mettl14, or the corresponding negative controls in osteogenic medium for 21 days. Scale bar, 10 mm. All data are the mean ± SD. *p < 0.05, **p < 0.01. One-way ANOVA with a post hoc test was performed, and the significance of differences between two groups was determined with Student's t test To examine whether miR-103-3p functionally targets Mettl14 to regulate osteoblast activity, we knocked down Mettl14 expression using siRNA (Figure 3g). We found that when we cotransfected cells with Mettl14 siRNA-NC and agomir-103-3p or antagomir-103-3p, there was no change in the effects of miR-103-3p on osteoblast proliferation ( Figure S3a). However, when we cotransfected cells with siRNA-Mettl14 and agomir-103-3p or antagomir-103-3p, the effects of miR-103-3p on osteoblast proliferation were blocked ( Figure S3b). Similar data were also obtained for osteoblast differentiation ( Figure S3c and Figure 3h).

F I G U R E 3 miR-103-3p directly targets
To further confirm that miR-103-3p-mediated regulation of osteoblast activity is Mettl14 dependent, we used the WT Mettl14 3′UTR to block the binding of endogenous miR-103-3p to Mettl14 ( Figure 3i). The time-dependent growth curve was shifted upward in cells transfected with the WT Mettl14 3′UTR compared to cells treated with the empty vector ( Figure S3d). Similar data were also obtained for osteoblast differentiation (Figure 3j,k).
Ribonuclease treatment weakened this coprecipitation, suggesting that the interaction of METTL14 and DGCR8 might be mediated in part by RNAs (Figure 4a). Additionally, we found that methylated RNA bound by DGCR8 was increased in METTL14-overexpressing (oeMETTL14) cells (Figure 4b). These findings indicated that METTL14 regulated pri-miRNA processing by manipulating recognition and binding of pri-miRNAs by DGCR8. When we immunoprecipitated DGCR8 from the vector and oeMETTL14 groups and used q-PCR to examine pri-miRNAs bound to DGCR8, we found that the level of pri-miR-103-3p bound to DGCR8 was increased in oeMETTL14 cells (Figure 4c). Moreover, we discovered that the level of pri-miR-103-3p modified by m 6 A was increased in oeM-ETTL14 cells when m 6 A was immunoprecipitated from RNAs of the vector and oeMETTL14 groups (Figure 4d). To further test the role of METTL14/m 6 A modification in miR-103-3p regulation, we first examined the level of m 6 A in the total RNA and found that oeMETTL14 resulted in increased m 6 A levels in osteoblasts and that siRNA-METTL14 decreased the m 6 A level (Figure 4e). We then assessed the expression of pri-miR-103-3p, pre-miR-103-3p, and mature miR-103-3p in siRNA-METTL14 or oeMETTL14 cells. Our data showed that the level of pri-miR-103-3p was increased in the siRNA-METTL14 group and decreased in the oeMETTL14 group. The levels of pre-miR-103-3p and mature miR-103-3p were decreased in the siRNA-METTL14 group and increased in the oeMETTL14 group (Figure 4f). We found a similar role of METTL14 in hFOB 1.19 cells ( Figure S4a). Taken together, these results indicate that the presence of m 6 A enhanced the recognition of pri-miR-103-3p by DGCR8 and the subsequent processing of pri-miR-103-3p into mature miR-103-3p.

| METTL14 promotes osteoblast proliferation, differentiation, and matrix mineralization
To explore the role of METTL14 in osteoblast activity, we trans- Masson's trichrome staining analysis of osteoid formation in the distal femur of mice after treatment with agomir-103-3p or its negative control (n = 8). Scale bars, 10 μm. Arrows indicate osteoids. (l) Three-point bending test analysis of mouse femur biomechanical properties after treatment with agomir-103-3p or its negative control. The n value for each group is indicated at the bottom of each bar in the graphs. All data are presented as the mean ± SD. *p < 0.05, **p < 0.01. One-way ANOVA with a post hoc test was performed, and the significance of differences between two groups was determined with Student's t test

| miR-103-3p inhibits bone formation in vivo
To determine the effects of miR-103-3p on bone formation, we constructed an adeno-associated virus that expressed green fluorescent protein (GFP) and agomir-103-3p or antagomir-103-3p as an in vivo delivery system. To investigate the effectiveness of the delivery system in vivo, we injected the delivery system via periosteal injection into the marrow cavity of the femur with an in-house pressurized injector and tracked GFP expression in live mice with a whole-animal fluorescence imaging system Wang, Deng, et al., 2019;Xu et al., 2018) at 1, 2, and 3 weeks postinjection (Figure 5a).
No fluorescence was detected in the control group, and the GFP fluorescence signals in the right femur lasted up to 3 weeks postinjection (Figure 5a). To further confirm the expression of miR-103-3p in the bone tissue of mice, q-PCR analysis was performed and showed that the level of miR-103-3p was increased and that this increase lasted for 3 weeks after a single periosteal injection; in addition, the variation in miR-103-3p levels was greater in bone than in other tissues (Figure 5b-d and Figure S5a).
To further test the effects of miR-103-3p on bone formation in vivo, we performed pulsed periosteal injections of agomir-103-3p (agomir) or agomir-NC (NC) with the delivery system in mice at 6 months ( Figure 5e). We chose the Alp, Bglap, and Col1α1 levels as indicators of osteoblast activity in vivo, as described previously (Wang et al., 2013).
q-PCR analysis showed that the Alp, Bglap, and Col1α1 levels were significantly lower than those in the NC group, and there were no differences between the control and NC groups (Figure 5f).
To analyze the bone mass and trabecular architecture of the mice,  Figure S5e). Through tartrate-resistant acid phosphatase (TRAP) staining, we also found that agomir-103-3p had no effect on osteoclast activity in vivo ( Figure S5f-h).
Masson's trichrome staining showed less osteoid staining in distal femurs from agomir mice than in those from NC mice ( Figure 5k). To identify the effects of miR-103-3p on the mechanical properties of the femurs, we performed a three-point bend test and found that the maximum load, stiffness, and elasticity modulus were decreased in the agomir group compared with the NC group (Figure 5i).
Further experiments showed that both Mettl14 mRNA and protein levels were decreased in agomir mice but increased in antagomir mice ( Figure S5i). Furthermore, agomir-103-3p resulted in decreased m 6 A levels, whereas antagomir-103-3p increased the level of m 6 A in mouse femurs ( Figure S5j).

| Inhibition of miR-103-3p promotes bone formation in OVX mice.
To test the efficiency of the delivery system with antagomir-103-3p in vivo, q-PCR analysis was performed and showed that the level of miR-103-3p was decreased and that this effect could last for 3 weeks postinjection; the variation in miR-103-3p levels was greater in bone than in other tissues ( Figures S5a and S6a,b). To examine the therapeutic effects of antagomir-103-3p on bone formation in OVXinduced osteoporotic mice, we performed pulsed injections of antagomir-103-3p (OVX+antagomir) or antagomir-NC (OVX+NC) with the delivery system in estrogen-depleted mice at 6 months after OVX (Figure 6a).
We found that the Alp, Bglap, and Col1α1 levels were significantly lower in OVX mice or OVX mice treated with a negative control antagomir (OVX+NC) than in OVX+antagomir mice (Figure 6b).
We also found that the cortical bone thickness was lower in the OVX and OVX+NC mice compared to sham mice and higher in the OVX+antagomir group compared to the OVX+NC group ( Figure   S6c,d). Additionally, trabecular bone mass was notably reduced (lower BMD and BV/TV) and that the trabecular architecture was significantly impaired (lower Tb.Th, Tb.N, and Conn.D and higher Tb.Sp and SMI) in OVX and OVX+NC mice compared to sham mice, whereas the trabecular bone mass was substantially increased, and the trabecular architecture was markedly improved in the  Figure S6f). Using TRAP staining, we also found that antagomir-103-3p had no effect on osteoclast activity in vivo ( Figure S5f-h). In addition, Masson's trichrome staining showed less osteoid staining in the distal femur in OVX and OVX+NC mice than in sham mice, whereas there was more osteoid staining in OVX+antagomir mice than in OVX+NC mice ( Figure 6g). Next, we performed a three-point bend test to test for changes in the mechanical properties of the femurs and identified that the maximum load, stiffness, and elasticity modulus were decreased in OVX and OVX+NC mice compared to sham mice, whereas these three biomechanical parameters were increased in OVX+antagomir mice compared to OVX+NC mice (Figure 6h). Our data also showed that both Mettl14 mRNA and protein levels were decreased in OVX and OVX+NC mice compared to sham mice, whereas both Mettl14 mRNA and protein levels were increased in OVX+antagomir mice compared to OVX+NC mice ( Figure S6 g). We found a similar role of antagomir-103-3p in regulating m 6 A levels in mice with OVX-induced osteoporosis ( Figure S6h).

| Correlations between the miR-103-3p/ METTL14/m 6 A axis and bone formation capacity
To clarify the relationship between the miR-103-3p/METTL14/m 6 a axis and bone formation capacity, we examined both Mettl14 mRNA and protein levels and m 6 A levels in bone specimens collected from F I G U R E 6 Therapeutic inhibition of miR-103-3p counteracts the decrease in bone formation in mice with OVX-induced osteoporosis. s trichrome staining analysis of osteoid formation in the distal femurs of mice after treatment with antagomiR-103-3p or its negative control (n = 8). Scale bars, 10 μm. Arrows indicate osteoids. (h) Three-point bending test analysis of mouse femur biomechanical properties after treatment with antagomiR-103-3p or its negative control. The n value for each group is indicated at the bottom of each bar in the graphs. All data are presented as the mean ± SD. *p < 0.05, **p < 0.01. One-way ANOVA with a post hoc test was performed, and the significance of differences between two groups was determined with Student's t test 24 elderly female patients with low-energy fractures (Figure 1a and Tables S1 and S2). Our results showed that the METTL14 protein levels, METTL14 mRNA levels, and m 6 A content in total RNA were lower in osteoporosis patients (T ≤ −2.5) than in control patients (T ≥ −2.5) ( Figure S7a). Then, we found that METTL14 mRNA levels were positively correlated with the m 6 A content in total RNA ( Figure   S7b). We also found that miR-103-3p levels were negatively correlated with METTL14 mRNA levels and the m 6 A content in total RNA from these human samples ( Figure S7c). In addition, we showed that both the METTL14 mRNA levels and m 6 A content in total RNA were positively correlated with ALP, BGLAP, and COL1α1 levels in human bone tissue ( Figure S7d,e).
We further confirmed the miR-103-3p/METTL14/m 6 a axis in bone tissues from OVX mice and found that the METTL14 protein levels, Mettl14 mRNA levels, and m 6 A content in total RNA were lower in OVX mice than in sham mice ( Figure S7f). Consistently, we found that Mettl14 mRNA levels were positively correlated with the

| DISCUSS ION
In the present study, we found that the miR-103-3p level was negatively correlated with bone formation in bone specimens from elderly women with fractures as well as in OVX mice. Then, we found that miR-103-3p directly targeted Mettl14 to functionally inhibit osteoblast activity and that METTL14 provided negative feedback to regulate miR-103-3p processing by DGCR8 and promote osteoblast activity. Moreover, we presented in vivo evidence demonstrating that miR-103-3p inhibited bone formation under physiological conditions and that inhibition of miR-103-3p promoted bone formation in OVX mice. In addition, we confirmed that METTL14 was negatively correlated with miR-103-3p but positively correlated with bone formation in bone specimens from women with fractures and OVX mice. As summarized in Figure S7  A series of miRNAs have been verified to regulate osteoblast function and osteoblastic bone formation (Inose et al., 2009;Wang et al., 2013;Xu et al., 2018). We found a close association between elevated miR-103-3p and reduced bone formation, evidenced by the negative correlation between the miR-103-3p level and ALP, BGLAP, and COL1α1 mRNA levels in bone specimens from elderly women with fractures and in OVX mice and primary mouse osteoblasts.
These results prompted us to investigate whether miR-103-3p plays a role in osteoblast activity. miR-103-3p has been reported to be adversely correlated with osteoblast differentiation and bone formation in response to mechanical stimulation (Zuo et al., 2015). Our previous study showed that miR-103-3p inhibits osteoblast proliferation under simulated microgravity conditions, a form of mechanical unloading .
Consistently, we observed that miR-103-3p inhibited osteoblast activity in mouse primary osteoblasts and hFOB 1.19 cells. More importantly, we found that miR-103-3p had no effect on osteoclast activity. Therefore, these data suggest that the aberrantly elevated miR-103-3p in bone specimens may contribute to a reduction in osteoblast activity and suppression of bone formation.
Some mechanisms by which miR-103-3p regulates gene expression are already known (Favereaux et al., 2011;Trajkovski et al., 2011). Specifically, miR-103-3p directly targets caveolin-1 to regulate insulin sensitivity (Trajkovski et al., 2011). Favereaux et al. showed that miR-103-3p is directly involved in chronic pain via targeting of Cav1.2 (Favereaux et al., 2011). In the present study, we predicted the potential targets of miR-103-3p by using miRNA target prediction software to investigate how miR-103-3p regulates osteoblast activity. Notably, the 3'UTR of Mettl14 possesses a 7-nt perfect match site for the miR-103-3p seed region predicted by prediction tools. Moreover, we confirmed that Mettl14 is a direct target of miR-103-3p in MC3 T3-E1 mouse preosteoblasts, as indicated by the luciferase assay and RIP analysis. miR-103-3p negatively regulated METTL14 expression and the m 6 A level in total RNA from primary mouse osteoblasts. Further functional studies verified that miR-103-3p inhibits osteoblast activity by directly targeting Mettl14 in osteoblasts. Therefore, our present study found for the first time that Mettl14 is a new target of miR-103-3p in osteoblasts.
Previous studies have shown that altered m 6 A levels are involved in regulation of miRNAs (Alarcón et al., 2015;Han et al., 2019;Ma et al., 2017;Peng et al., 2019;Wang, Deng, et al., 2019;Yan et al., 2020;Zhang, Bai, et al., 2019). Alarcón and colleagues showed that m 6 A modification promotes the recognition of pri-miRNA sequences and reported that DGCR8 is involved in an initiation event in miRNA biogenesis during METTL3-dependent m 6 A methylation (Alarcón et al., 2015). Other researchers have demonstrated that upregulation of METTL3/m 6 A modification promotes pri-miR-25 (Zhang, Bai, et al., 2019), pri-miR-221/222 (Han et al., 2019), pri-miR-143-3p (Wang, Deng, et al., 2019), and pri-miR-1246  maturation (decreasing the expression of pri-miRNA but increasing the expression of pre-miRNA and miRNA). Recently, Ma et al. re-ported that METTL14, another important m 6 A methyltransferase regulator, positively modulates the pri-miR-126 process in a DGCR8dependent manner (Ma et al., 2017). In the present study, we showed that METTL14-dependent m 6 A methylation regulated miR-103-3p processing by the microprocessor protein DGCR8. However, some reports have demonstrated that METTL3/m 6 A suppresses the expression of pre-miR-320 and miR-320 (Yan et al., 2020). This variation in results may be due to the use of different species and cell sources or different disease models.
Accumulating evidence has shown that m 6 A modification not only modulates nearly all aspects of RNA metabolism, such as splicing, structure, stability, translation, and export, but is also involved in many human diseases, including major depressive disorder, obesity-related traits, type 2 diabetes mellitus, and cancers (Wang, & He, 2014;Zhao, & He, 2015;Zhao et al., 2017). Recently, several studies have focused on the involvement of m 6 A modification in bone metabolism (Guo et al., 2011;Li et al., 2019;Sachse et al., 2018;Shen et al., 2018;Tian et al., 2019;Wu et al., 2018;Yao et al., 2019;Yu et al., 2020;Zhang, Riddle, et al., 2019;Zhang et al., 2020). Wu and colleagues demonstrated that METTL3-mediated m 6 A modification regulated the fate of bone marrow mesenchymal stem cells (Wu et al., 2018). Other studies have shown that the m 6 A demethylase FTO is essential for normal bone growth and functions to protect osteoblasts from genotoxic damage (Zhang, Riddle, et al., 2019). However, the biological significance of the m 6 A methyltransferase METTL14 in bone formation or osteoporosis has not been confirmed. In this study, we showed for the first time that METTL14 promotes osteoblast proliferation, differentiation, and matrix mineralization.
Recently, several studies have tested the efficacy of miRNA-modulating compounds in combating osteoporosis. Wang and colleagues demonstrated that therapeutic silencing of miR-214 could increase bone formation and counteract osteoporosis (Wang et al., 2013).
Another study showed that inhibition of miR-31a-5p prevented bone loss in aged rats . Our previous studies revealed that blocking the expression of miRNA-132-3p reverses disuse osteopenia in mice (Hu et al., 2020). We also showed that delivery of miR-33-5p partially rescued bone loss in hindlimb-unloaded mice . In the present study, we showed that mature miR-103-3p is evolutionarily conserved among several species and highly expressed in bone tissue. Moreover, we demonstrated that miR-103-3p inhibited osteoblast activity and bone formation by targeting Mettl14 in vivo under physiological conditions. More importantly, the therapeutic inhibition of miR-103-3p partially counteracted bone loss in OVX mice. In addition, miR-103-3p had no effect on osteoclast activity in vitro and in vivo. These results demonstrate that therapeutic silencing of miR-103-3p may promote bone formation and partially rescue bone loss in postmenopausal osteoporosis by exerting an anabolic effect on osteoblast activity. To reveal the relationship between the miR-103-3p/METTL14/m 6 a axis and bone formation capacity, we measured METTL14 expression levels in bone specimens and found that METTL14 was negatively correlated with miR-103-3p but positively correlated with bone formation in bone specimens from elderly women with fractures and OVX mice.
It should be noted that our study has some limitations. In the present study, we observed a regulatory effect of miR-103-3p/ METTL14/m 6 a on osteoblast activity but did not investigate the effect and mechanism of METTL14 in vivo. We are working to acquire a sufficient number of osteoblast-specific miR-103-3p and METTL14 knock-out and knock-in mice to further verify the existence of the miR-103-3p/METTL14/m 6 a axis in vivo and test the mechanism of METTL14 in bone formation. In addition, unfortunately, due to the experimental material constraints, we did not get or synthesize the targeted delivery systems in our laboratory, and we would further test the effects of miR-103-3p modulators on osteoblast in vivo in a future study.
In summary, our results illustrate that miR-103-3p plays critical roles in postmenopausal osteoporosis by inhibiting osteoblast activity and bone formation and reveal a previously unrecognized signaling axis involving miR-103-3p/METTL14/m 6 A in osteoblasts.
Moreover, our work suggests that this signaling axis may be a potential target for ameliorating postmenopausal osteoporosis.

| E XPERIMENTAL PROCEDURE S
The detailed procedures were provided in the Appendix S1.
Tianyi Zhang, and the technical staff of Nanjing Arrival Biotechnology Co., Ltd. for providing valuable suggestions for this work. This work was supported by grants from the National Natural Science

CO N FLI C T O F I NTE R E S T
The authors declare no competing financial interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
We provided part of raw data and ethics approval statements in the Appendix S1. If necessary, we provided all the original data.