Piezo1 opposes age‐associated cortical bone loss

Abstract As we age, our bones undergo a process of loss, often accompanied by muscle weakness and reduced physical activity. This is exacerbated by decreased responsiveness to mechanical stimulation in aged skeleton, leading to the hypothesis that decreased mechanical stimulation plays an important role in age‐related bone loss. Piezo1, a mechanosensitive ion channel, is critical for bone homeostasis and mechanotransduction. Here, we observed a decrease in Piezo1 expression with age in both murine and human cortical bone. Furthermore, loss of Piezo1 in osteoblasts and osteocytes resulted in an increase in age‐associated cortical bone loss compared to control mice. The loss of cortical bone was due to an expansion of the endosteal perimeter resulting from increased endocortical resorption. In addition, expression of Tnfrsf11b, encoding anti‐osteoclastogenic protein OPG, decreases with Piezo1 in vitro and in vivo in bone cells, suggesting that Piezo1 suppresses osteoclast formation by promoting Tnfrsf11b expression. Our results highlight the importance of Piezo1‐mediated mechanical signaling in protecting against age‐associated cortical bone loss by inhibiting bone resorption in mice.


| INTRODUC TI ON
Advanced age causes bone loss in both cancellous and cortical compartments (Almeida et al., 2007;Piemontese et al., 2017;Riggs et al., 2008). The cancellous bone loss with age is associated with decreased bone formation and a low rate of bone remodeling (Almeida et al., 2007). In contrast, age-associated cortical bone loss is mainly due to increased osteoclast number and bone resorption at the endocortical surface (Piemontese et al., 2017). Our previous studies showed that Tnfsf11 encoding RANKL, a cytokine that is essential for osteoclast formation, increases with age in cortical bone (Piemontese et al., 2017). In addition, OPG, a decoy receptor for RANKL, decreases with age (Piemontese et al., 2017). More importantly, blockade of osteoclast formation by deletion of Tnfsf11 in osteocytes prevents the increase in osteoclast number and cortical bone loss with age in mice (Kim et al., 2020). These studies demonstrate the importance of increased osteoclast formation in ageassociated cortical bone loss.
In humans and rodents, the loss of bone mass caused by aging is often associated with a decline in muscle mass and physical activity (Curtis et al., 2015;Hamrick et al., 2006). These declines lead to reduced mechanical loading on bone. Mechanical stimulation such as that derived from physical activity promotes bone formation by increasing osteoblast number and activity (Bailey & Brooke-Wavell, 2010;Robling & Turner, 2009;Turner et al., 1998). On the contrary, loss of mechanical stimulation causes bone loss by promoting osteoclast formation and bone resorption (Xiong et al., 2011). Therefore, both osteoblasts and osteoclasts are regulated by mechanical stimulation. In addition, the anabolic response to mechanical loading in bone decreases with age indicating a reduced ability to sense and/or transduce mechanical stimulation in old bone (Holguin et al., 2014;Karinkanta et al., 2007). Taken together, these findings suggest that decreased mechanical stimulation contributes to ageassociated bone loss.
Piezo1, a mechanosensitive ion channel, is capable of sensing various mechanical stimulation including membrane stretch, matrix rigidity, and fluid shear stress and plays an important role in mechanotransduction in many organs (Coste et al., 2010;Gudipaty et al., 2017;Li et al., 2014Li et al., , 2021Miyamoto et al., 2014). We and others have deleted the Piezo1 gene in osteoblast lineage cells and demonstrated that Piezo1 plays a critical role in bone homeostasis (Hendrickx et al., 2021;Li et al., 2019;Sun et al., 2019;Zhou et al., 2020). In our studies, we found that cancellous bone mass was decreased in young adult mice lacking Piezo1 in Dmp1-Cre targeted cells. We also observed decreased cortical thickness and periosteal circumference in these mice. More importantly, the skeletal response to mechanical loading was blunted in the conditional knockout mice (Hendrickx et al., 2021;Li et al., 2019). These results demonstrate that Piezo1 expression in Dmp1-Cre targeted cells is essential for bone homeostasis and mediates the skeletal response to mechanical stimulation. In present study, we determined whether Piezo1 expression in osteoblast lineage cells plays an important role in age-associated bone loss. We found that mice lacking Piezo1 in osteoblast lineage cells lose more cortical bone with age than control mice and that this is associated with a more profound increase in osteoclast number at the endocortical surface.

| Piezo1 expression decreases with age
To determine whether Piezo1 expression changes with advanced age, we compared Piezo1 expression in tibia cortical bone of 6-month-old versus 24-month-old mice and found that Piezo1 expression is significantly reduced in cortical bone of 24-month-old mice (Figure 1a). This decrease was observed in both the C57BL/6J and DBA mouse strains ( Figure 1a). In addition to mice, the expression of Piezo1 was also compared between young and old human subjects in the cortical bone of femoral neck. The results showed a similar decrease in Piezo1 expression in the elderly human subjects (73.8 ± 3.2 years old) compared to the young ones (47.1 ± 7.01 years old) (Figure 1b).
To further validate the findings, RNAscope, an RNA in situ hybridization technique, was performed on the mouse and human bone samples. The results indicated decreased Piezo1 mRNA in cortical bone from old murine femur and human femoral neck (Figure 1c,d), which is consistent with the gene expression analysis. Overall, the results provide compelling evidence of a correlation between aging and decreased Piezo1 expression in bone. The findings suggest that F I G U R E 1 Piezo1 expression decreases with age. (a) mRNA levels of Piezo1 in 6-and 24-month-old C57BL/6J and DBA mice (n = 9-11, here and throughout, values are mean ± SD). (b) qPCR of Piezo1 mRNA in human cortical bone isolated from human femoral neck removed during hip replacement surgeries from young (n = 9 with mean age of 47.1 ± 7.01 years old) and aged patients (n = 13 with mean age of 73.8 ± 3.2 years old). (c) Quantification and representative images of Piezo1 mRNA in situ hybridization using RNAscope on femoral cortical bone of 6-and 24-month-old C57BL/6J mice. Red dots indicate positive Piezo1 signals. Scale bar, 100 μm. (d) Quantification and representative images of Piezo1 mRNA in situ hybridization on human cortical bone of femoral neck from young and old individuals. Red dots indicate positive Piezo1 signals. *p < 0.05 using Student's t test. Percentage of Piezo1 + osteocytes % * * the reduction in Piezo1 expression may play a role in the age-related changes in bone.

| Deletion of Piezo1 in osteoblasts and osteocytes exacerbates cortical bone loss with age
To determine whether Piezo1 expression in osteoblast lineage cells plays a role in age-associated bone loss, we generated Piezo1 conditional knockout mice in which the Piezo1 gene is deleted from Dmp1-Cre transgene targeted cells, hereafter referred to as Dmp1- Cre; Piezo1 f/f mice. Littermates homozygous for the conditional allele but lacking the Dmp1-Cre transgene were used as controls.
These mice were aged to 6 and 24 months of age, and the bones of 24-month-old mice were then analyzed and compared with bones from 6-month-old mice by micro-CT analysis.
As expected, control mice exhibited age-associated cancellous bone loss at 24 months of age compared to 6-month-old mice ( Figure 2a; Figure S1a). In contrast, conditional knockout mice, which had lower cancellous bone mass than controls at young age, did not further lose cancellous bone with age ( Figure 2a; Figure S1a). Analysis of femoral cortical bone revealed that control mice had lower cortical thickness at 24 versus 6 months of age ( Figure 2b). Interestingly, old Dmp1-Cre; Piezo1 f/f mice displayed a more profound decrease in cortical thickness with age compared to control mice ( Figure 2b). Consistent with this, 24-month-old Dmp1-Cre; Piezo1 f/f mice displayed spontaneous tibial fracture (4 out of 12), whereas no fracture was observed in either young mice or old controls ( Figure 2c). These results demonstrate that even starting with lower cortical thickness, mice lacking Piezo1 in osteoblasts and osteocytes lose more cortical bone with age.
To identify the sites of the cortical bone loss, we measured periosteal and endosteal circumference in the midshaft of femur.
Micro-CT analysis revealed that the periosteal circumference increased with age in control mice ( Figure 2d). Despite the lower periosteal circumference established during growth, adult Dmp1-Cre; Piezo1 f/f mice displayed a comparable increase in periosteal circumference with age as control mice (Figure 2d). The increase of total cross-sectional area with age in femoral midshaft was also comparable between two genotypes ( Figure S1b). Endosteal circumference was increased in both control and Dmp1-Cre; Piezo1 f/f mice at 24 months of age ( Figure 2e). However, this increase was greater in conditional knockout mice than that in control mice (Figure 2e).
In line with this, the increase of medullary area with age was more profound in conditional knockout mice, whereas the increase of cortical area was blunted in these mice ( Figure S1b). Consistently, the increase in polar moment of inertia with age measured in femoral midshaft was blunted in the Dmp1-Cre; Piezo1 f/f mice (Figure 2f).
Like cortical thickness, Dmp1-Cre; Piezo1 f/f mice developed more cortical porosity with age than control mice as shown in histological sections and by micro-CT analysis (Figure 2g,h). Taken together, these results demonstrate that Piezo1 deletion in Dmp1-Cre targeted cells increased endocortical expansion and cortical porosity in old mice, suggesting that the greater loss of cortical bone with age in conditional knockout mice is due to increased endocortical bone resorption.

| Deletion of Piezo1 in osteoblasts and osteocytes increases osteoclast formation at endocortical surface
To understand the cellular basis of the skeletal phenotype, we measured osteoclast number and surface at the femoral endocortical surface. TRAP staining showed that osteoclast number and surface in endosteum were dramatically increased in aged mice than that in young mice in both genotypes (Figure 3a

| Deletion of Piezo1 in osteoblasts and osteocytes suppresses Tnfrsf11b expression
To further understand the mechanisms by which Piezo1 regulates osteoclast formation, we analyzed the expression levels of genes known to play a crucial role in osteoclastogenesis, including Tnfsf11 and Tnfrsf11b in Piezo1 knock-down MLO-Y4 cells. Our results revealed that Tnfrsf11b expression was dramatically decreased in Therefore, our results indicate that Piezo1 controls osteoclast formation via upregulation of Tnfrsf11b expression.

| Piezo1 controls Tnfrsf11b expression via Ca 2+ / CaM/mTOR signaling pathway
To further investigate the mechanisms by which Piezo1 regulates Tnfrsf11b expression, we used MLO-Y4 cells as a model.
We first examined whether calcium influx is required for Yoda1 induced Tnfrsf11b expression since Piezo1 is a calcium permissive ion channel and Yoda1 induces calcium influx in MLO-Y4 cells . We cultured MLO-Y4 cells in calcium free medium and treated these cells with Yoda1 for 2 h. We found that Yoda1 induced Tnfrsf11b expression was completely blunted in the absence of extracellular calcium ( Figure 5a). We then determined whether intracellular calcium is required for Yoda1 induced Tnfrsf11b expression. We treated MLO-Y4 cells with BAPTA, an intracellular calcium chelator, and found that Yoda1 induced Tnfrsf11b expression was blocked without free intracellular calcium (Figure 5b).

| DISCUSS ION
Loss of mechanical stimulation has been implicated in age-associated bone loss (Javaheri & Pitsillides, 2019). In the current study, we investigated the role of Piezo1, a mechanosensor, in age-associated bone loss. Our studies showed that the increased endocortical bone resorption that occurs with old age is associated with a decline of Piezo1 expression in bone. Moreover, our results indicate that Piezo1 plays an important role in preventing age-associated increase in endocortical bone resorption by maintaining OPG expression.
Therefore, we provide evidence that Piezo1 signaling opposes ageassociated cortical bone loss.
Although Piezo1 knockout mice initially showed a thinner cortex at young age, they still experienced age-related cortical bone loss.
This suggests that the underlying driving force for age-associated cortical bone loss is not dependent on Piezo1 signaling. However, our study also found that Piezo1 knockout mice experience more cortical bone loss with age than control mice, indicating that Piezo1 plays an important role in age-associated cortical bone loss as a defensive factor. This is consistent with previous studies showing that hindlimb unloading, which reduces mechanical stimulation, causes cortical bone loss in old rats (Perrien et al., 2007). Together, these studies suggest that bones in old age still experience some degree of mechanical loading, which can protect them from age-related cortical bone loss. Additionally, our study found that Piezo1 expression decreases with age in bone, which may contribute to the decreased skeletal response to loading seen with age. Our data provide evidence supporting the idea that reduced mechanical stimulation due to decreased physical activity and/or decreased response to loading contributes to age-related cortical bone loss.
The protective effects of mechanical loading on cortical bone may be attributed to increased bone formation, decreased bone resorption, or both. Studies in young and adult mice have found that mechanical loading increases bone formation Lynch et al., 2011), but this ability is greatly reduced in older mice (Holguin et al., 2014;Meakin et al., 2015). Our previous research conducted using RANKL conditional knockout mice has shown that age-related cortical bone loss is primarily due to endocortical bone resorption rather than decreased bone formation (Kim et al., 2020). Therefore, it is possible that mechanical loading alleviates age-related cortical bone loss by regulating bone resorption. Our findings are in line with this idea, as we observed that Piezo1 conditional knockout mice experienced more cortical bone loss with age than control mice due to a more profound increase in endocortical bone resorption. These data suggest that the primary role of Piezo1 signaling in older bones is to harness osteoclast formation in the endosteum and prevent cortical thinning. Whether the effect of Piezo1 signaling on old bone is dependent on mechanical stimulation needs further investigation.  (Jepsen & Andarawis-Puri, 2012). The factors that affect periosteal expansion are not fully understood, but it is thought to be related to changes in hormones, bone remodeling, and mechanical stimulation (Allen et al., 2004;Orwoll, 2003;Parfitt, 2002). In previous studies, we found that Piezo1 deletion in osteoblast lineage cells greatly reduced periosteal bone formation and cortical bone accrual in young growing mice . However, aged Piezo1 knockout mice showed increased endocortical bone resorption but normal periosteal bone formation compared to wildtype mice. These data indicated that Piezo1-mediated mechanical signaling controls periosteal bone formation in young but not old mice and that periosteal bone formation is not related to bone resorption. Consistent with this, we found in our previous studies that periosteal expansion was not affected in aged RANKL conditional knockout mice in which bone remodeling was greatly suppressed (Kim et al., 2020). More importantly, the bones of RANKL knockout mice are mechanically underloaded since they have much higher bone mass with similar body weight comparing to wildtype mice. Together, these findings suggest that mechanical stimulation plays a minimal role in periosteal bone formation in old mice.

F I G U R E 5 Piezo1 controls
Our study revealed that Piezo1 plays a role in regulating osteoclast formation through its ability to stimulate OPG expression.

Elevated production of RANKL by senescent osteocytes has been
shown to be a key mechanism responsible for increased bone resorption and age-related cortical bone loss (Kim et al., 2020). OPG, acting as a decoy receptor for RANKL, helps counteract this effect, but its expression decreases with age (Piemontese et al., 2017), which could contribute to age-associated bone resorption. We observed a correlation between decreased Piezo1 expression, reduced OPG expression, increased osteoclast formation on the endocortical surface, and further cortical bone deterioration of aged bones.
Our results highlight the intricate interplay between age, mechanical loading, and bone resorption, and suggest that lack of mechanical stimulation in aged mice could exacerbate the decline in OPG expression, leading to greater bone resorption and cortical bone degradation.
In contrast to cortical bone, we did not observe a significant cancellous bone loss associated with aging in Piezo1 conditional knockout mice. One explanation could be that the cancellous bone mass in mice lacking Piezo1 is already extremely low in young age so that there is not much bone to lose with age in these mice. The second reason could be that Piezo1 signaling primarily affects bone resorption in old age which is related to cortical bone loss while cancellous bone loss is mainly associated with decrease in bone formation. Our results are in line with previous evidence that different aging mechanisms underlie the loss of cortical versus cancellous bone and the difference may be related to the ways the two compartments respond to mechanical loading.
In summary, our findings suggest that mechanical loading primarily promotes bone formation in cortical bone in young growing mice, resulting in increased bone mass and size. However, in old mice, mechanical loading primarily inhibits bone resorption and prevents cortical thinning and porosity. Consistent with this, the role of Piezo1 shifts from promoting bone formation to suppressing bone resorption as mice age. These findings shed light on the intricate mechanisms underlying age-related bone loss and emphasize the importance of mechanical stimuli in preserving skeletal integrity during aging. This knowledge could inform strategies for preventing age-related bone disorders.

| Osteoclast culture
To demonstrate the ability of Piezo1 knockdown MLO-Y4 cells to support osteoclast formation, we utilized osteoclast coculture system which has been used to generate osteoclasts in vitro previously (Xiong et al., 2018). Bone marrow cells were isolated from the femurs and tibias of 3-month-old male C57BL/6J mice and were cultured in a petri dish in αMEM supplemented with 15% FBS and

| Femoral organ culture
Female mice at 4 months of age were euthanized in a CO 2 chamber. Femurs were dissected, and both ends were removed in a culture hood. Bone marrow was then flushed out using PBS, and the periosteal surface was scraped to remove periosteal cells. Femoral shafts were then cultured in a 12-well-plate with 1 mL of α-MEM supplemented with 10% FBS and 1% PSG for 24 h. We then treated femur shafts with 2 μM Yoda1 (Sigma) or DMSO for 3 days. Femur shafts were then collected for RNA isolation and qPCR analysis.

| Skeletal analysis
Tibial X-rays were obtained using an UltraFocus X-ray machine

| Quantitative PCR
Organs and whole bones were harvested from animals, removed of soft tissues, and stored immediately in liquid nitrogen. We prepared osteocyte-enriched cortical bone by removing the ends of femurs or tibias and then flushing the bone marrow with PBS.
We then scraped the bone surface with a scalpel and froze them in liquid nitrogen for RNA isolation. We isolated total RNA using TRIzol (Life technologies), according to the manufacturer's instructions and prepared cDNA using High Capacity first strand cDNA synthesis kit (Life Technologies). We performed quantitative RT-PCR using the following Taqman assays from Applied Biosystems: Actb (Hs03023943_g1). We calculated relative mRNA amounts using the ∆ Ct method (Livak & Schmittgen, 2001).

| Quantification and statistical analysis
GraphPad Prism 9 software (GraphPad, San Diego) was used for statistical analysis. Two-way analysis of variance (ANOVA) or Student's t test were used to detect statistically significant treatment effects, after determining that the data were normally distributed and ex-

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.