Bone morphogenetic protein receptor 1α promotes osteolytic lesion of oral squamous cell carcinoma by SHH‐dependent osteoclastogenesis

Abstract Oral squamous cell carcinoma (OSCC) is an aggressive tumor that usually invades the maxilla or mandible. The extent and pattern of mandibular bone invasion caused by OSCC are the most important factors determining the treatment plan and patients' prognosis. Yet, the process of mandibular invasion is not fully understood. The following study explores the molecular mechanism that regulates the mandibular invasion of OSCC by focusing on bone morphogenetic protein receptor 1α (BMPR1α) and Sonic hedgehog (SHH) signals. We found that BMPR1α was positively correlated to bone defect of OSCC patients. Mechanistically, BMPR1α signaling regulated the differentiation and resorption activity of osteoclasts through the interaction of OSCC cells and osteoclast progenitors, and this process was mediated by SHH secreted by tumor cells. The inhibition of SHH protected bone from tumor‐induced osteolytic activity. These results provide a potential new treatment strategy for controlling OSCC from invading the jawbones.


| INTRODUC TI ON
Oral squamous cell carcinoma, which accounts for approximately 40% of all HNSCC, is an aggressive type of tumor associated with high mortality and a low response to chemotherapy. 1 The predilection sites of OSCC are the tongue and gingiva, through which OSCC invades the maxillary or mandibular bone. 2 Therefore, the extent and pattern of mandibular bone invasion caused by OSCC are the most important factors for determining the treatment plan and patients' prognosis. 3 The osteolytic destruction caused by osteoclasts and proteolytic enzymes is often the result of bone invasion by cancer cells. This process requires the cell-cell interaction in the TBME, which affects tumorigenesis, progression, and treatment response. 1 Osteoclasts derived from hematopoietic stem cells are the main effector cells responsible for bone resorption. After being recruited into the bone resorption site, hematopoietic stem cells differentiate into osteoclasts through RANKL/RANK/OPG signaling, which then secrete collagenases, matrix metalloproteinases, and other enzymes, destroying and reabsorbing mineral substances of bone, and finally resulting in loss of bone tissues.
Tumor cells usually secrete various factors that affect the bone resorption function of adjacent osteoclasts. It is believed that a potential control of those factors could regulate osteoclast generation, differentiation, and function and, in turn, constrain bone invasion or metastasis. The BMP family is widely involved in the formation, growth, and metastasis of HNSCC. [4][5][6][7] Bone morphogenetic protein receptor 1α is one of the most important receptors mediating BMP signaling; yet, its role in tumor progression is not fully understood.
Moreover, Cannonier et al discovered that hedgehog signaling has an important role during the OSCC invasion into the mandible. 8 Sonic hedgehog, a known regulator of OSCC microenvironment, can activate the status of cancer-associated fibroblasts and strengthen the differentiation and bone invasion through osteoclasts. 9,10 However, the molecular mechanisms that control SHH expression in OSCC remain unclear.
In this study, we explored the molecular mechanism that regulates the mandibular invasion of OSCC by focusing on BMPR1α on SHH signals.

| Patients
The study was approved by the Biomedical Ethics Committee

| Animals
BALB/c male nude mice, 4-6 weeks old, weighing 20-25 g, were obtained from Vital River Laboratory Animal Technology. All the animals were housed in an environment with a temperature of 22 ± 1°C, relative humidity of 50 ± 1%, and a light / dark cycle of 12/12 hours. All animal studies (including the mice euthanasia procedure) were done in compliance with Peking University institutional animal care regulations and guidelines and conducted carried out to the AAALAC and IACUC guidelines (2020/06/19, No. LA2020378).

| Anteromedial tibia tumor mouse model and tumor growth estimation
An anteromedial tibia tumor model was established to investigate the bone invasion of tumor cells in vivo. Control (sh-NC) or sh-BMPR1α WSU-HN6 cells were resuspended with a concentration of 10 6 cells/100 μL. Then, 100 μL cell suspension was injected into the anteromedial tibia of BALB/c nude mice (Vital River Laboratory Animal Technology). Mice were subjected to further analysis after 3 weeks. GDC-0449 (20 mg/kg, Selleck Chemicals) was used to inhibit SHH, while SAG (25 mg/kg; Sigma-Aldrich) was used as the agonist of SHH; both substances were injected intraperitoneally when the tumor volume reached approximately 100 mm 3 , followed by injection every 3 days. After 2 weeks, mice were killed, and the tissue was collected and analyzed ex vivo. 13 Fluorescence from the GFP-conjugated lentiviral vector was measured by IVIS Lumina III (PerkinElmer). The radiant efficiency was analyzed by Living Image. The tumor volume was calculated using the following equation 14 : W is the average distance in the proximal tibia at the level of the knee joint in the anterior-posterior and medial-lateral planes, and L is the distance from the edge of the proximal of the tumor to the distal extent of the tumor.

| Tartrate-resistant acid phosphatase staining
A TRAP staining kit (387A; Sigma-Aldrich) was used to detect TRAP + osteoclasts according to the manufacturer's instructions. To evaluate the osteoclast formation ability in vitro, a total number of TRAP + cells per hpf and the percentage of TRAP + area of five randomly selected fields were analyzed using ImageJ software. For analyzing the bone defect of the mouse tibia, anteromedial tibias of mice were scanned by micro-CT (Inveon MM Gantry-STD) with a 9 μm resolution ratio (60 kVp, 220 μA).

| Computerized tomography and micro-CT
Inveon Research Workplace (Siemens) was used to reconstruct a 3D image of the tibia. The ratio of BV to TV was used to assess bone loss.

| Bone marrow-derived macrophages and osteoclast formation
To induce osteoclast differentiation, BMMs were cultured with medium supplemented with 50 ng/mL M-CSF (R&D Systems) for 3 days (stage I), followed by 50 ng/mL RANKL (R&D Systems) and 50 ng/mL M-CSF stimulation for another 5 days (stage II). The conditioned medium from sh-NC and sh-BMPR1α WSU-HN6 cells were added at stage II to mimic the in vivo TBME. To examine the effect of SHH, rm-SHH (50 ng/mL; R&D Systems) or GDC-0449 (500 ng/mL) were added at stage II. The cells were then fixed or collected for further analysis.

| Bone resorption assay
A bone resorption assay was carried out to detect the resorption activity of osteoclasts. Bone marrow cells were plated in 24-well osteo assay plates (Corning osteo assay, CLS3987; Sigma-Aldrich) and cultured with the aforementioned conditional medium or reagents. After 10 days, the plates were observed using an Olympus microscope and analyzed by ImageJ to calculate the resorbed area. SPSS 26.0 (IBM) was used for statistical analysis. Student's t test was used to analyze the data between two groups; Tukey's test was used to analyze data among multiple groups. Survival time was compared using the log-rank test, and Kaplan-Meier was used to calculate the accumulated survival rate. The correlation test was analyzed by the Pearson test. P < .05 was considered statistically significant.

| Statistical analysis
More information on detailed materials and methods is available in Appendix S1.

| Expression of BMPR1α positively correlated with bone invasion in human OSCC
First, we examined the expression of BMP family receptors in specimens collected from patients with lower gingival OSCC ( Figure 1A).
ACVR2B, BMPR2, ALK1, and ALK2 were upregulated in OSCC tissue compared with paracarcinoma tissue, although without statistical significance. Among the seven receptors, BMPR1α and BMPR1β showed a significant increase in OSCC tissue, and BMPR1α expression level was higher than BMPR1β.
Next, the basic information of 104 cases diagnosed with lower gingival OSCC was collected, and the location and extent of jaw bone invasion were analyzed. The lesions that occurred behind the mental foramen of the mandible accounted for the vast majority of cases (88/104; Table 1).
Seventy-seven of the cases with well-stored pathological specimens were included for further analysis. Immunohistochemical analysis of BMPR1α was carried out to estimate the expression of BMPR1α ( Figure 1B). Moreover, the bone defect volume of the above patients' mandible and the proportion of bone defects caused by tumor invasion were calculated ( Figure 1C). The correlation analysis results showed that BMPR1α expression level was positively correlated with proportion of bone defects invaded by OSCC ( Figure 1D). In addition, a larger number of TRAP + osteoclasts was observed in the tumor-bone invasion site with high expression of BMPR1α ( Figure 1B).
Furthermore, survival analysis revealed that the prognosis of OSCC cases in the BMPR1α-high expression group was worse than the BMPR1α-low expression group ( Figure 1E).

| Bone morphogenetic protein receptor 1α participates in bone invasion by regulating the differentiation of osteoclasts
To explore the tumor biological role of BMPR1α, we collected the cDNA of four OSCC cell lines (WSU-HN6, SCC15, SCC25, and CAL27) and performed quantitative RT-PCR. As shown in Figure  Receptor activator of nuclear factor κ-B ligand is critical for bone invasion and is responsible for the differentiation of osteoclast differentiation. 15 In the tumor tissues of mice injected with BMPR1α knockdown tumor cells, the expression of RANKL was significantly decreased, as shown by IHC ( Figure 2C,G).

| Expression of BMPR1α in OSCC tumor cells affects the differentiation and resorption function of osteoclasts
The culture supernatants from sh-NC and sh-BMPR1α WSU-HN6 cells were used as the conditioned medium to further verify the role of BMPR1α expression level in OSCC on the differentiation of osteoclasts. The TRAP staining results showed significantly decreased osteoclast differentiation in the sh-BMPR1α medium compared to the sh-NC medium, as confirmed by the osteoclast number and percentage of TRAP + area per hpf ( Figure 3A-C).
Previous study reported that the formation of F-actin rings is required for the resorption function of osteoclasts. 16 The organized, sharply defined structures of F-actin rings, such as ruffled borders and sealing zone represent normal function of osteoclasts, 17,18 and this process is related with the osteoclast-related genes including NFATc1, ATP6v0d2, and RANKL. 19,20 Immunofluorescence staining showed that osteoclasts with OSCC cell-derived conditioned medium formed organized F-actin rings, whereas disorganized F-actin rings were found in groups induced with conditioned medium from BMPR1α knockdown OSCC cells ( Figure 3A). Thus, we speculated that the resorption function was weakened in the conditioned medium obtained from BMPR1α knockdown OSCC cells, as confirmed by bone resorption assay ( Figure 3A,D). Meanwhile, the mRNA levels of osteoclastogenic markers, such as rankl, nfatc1, cathepsin k, and atp6v0d2, were all decreased in the sh-BMPR1α WSU-HN6 group compared to the sh-NC group ( Figure 3E). To sum up, we concluded that the expression level of the BMPR1α gene in tumor cells might participate in bone invasion by regulating osteoclast differentiation.

| Bone morphogenetic protein receptor 1α regulates the expression of SHH in tumor cells of OSCC
Previous studies reported that hedgehog signaling might participate in BMP-induced signaling of mesenchymal stem cells and cancer cell activities. [21][22][23] In addition, our previous study found that IHH has a critical role in calvarial bone homeostasis and repair through regulation of osteoclast differentiation. 24 Thus, we hypothesized that BMPR1αrelated bone invasion might be hedgehog signaling-dependent. Thus, we analyzed the expression of the hedgehog family, including IHH, SHH, and Desert hedgehog in sh-NC and sh-BMPR1α WSU-HN6 cells. Bone morphogenetic protein receptor 1α knockdown or inhibition by BMP signaling significantly decreased the expression of SHH in WSU-HN6 cells ( Figures 4A and S3A). Western blot results further confirmed decreased SHH expression after BMPR1α was decreased ( Figure 4B).
Through in silico promoter analysis, we found potential sequences at the promoter of SHH that might be binding sequences of Smad4, the downstream transcription factor of BMP signaling ( Figure S3B). Next, the ChIP assay confirmed the binding of Smad4 with SHH promoter in tumor cells ( Figure 4C). After BMPR1α was inhibited, the ChIP results indicated a reduction of Smad4 occupied on the promoter of SHH (Figures 4C and S3C). Accordingly, the  Figure 4D), resulting in the decreased expression and secretion of SHH ( Figure S3D,E).
To verify the relation of BMPR1α and SHH, specimens of the aforementioned cases were subjected to IHC staining. The results verified the positive correlation of SHH with BMPR1α in OSCC tissue ( Figure 4E,F).

| Sonic hedgehog mediates BMPR1α-induced osteoclast differentiation
Next, we examined whether BMPR1α-induced osteoclast differentiation is regulated in an SHH-dependent manner. We induced BMM with sh-NC or sh-BMPR1α medium and treated sh-BMPR1α group cells with recombined mouse SHH to upregulate the hedgehog signaling. Staining with TRAP showed that the osteoclast number Immunofluorescence staining showed the F-actin rings with more organized structure, and bone resorption assay also confirmed recovered bone resorption ability after SHH stimulation in the sh-BMPR1α group ( Figure 4G,J). In addition, the mRNA levels of rankl, nfatc1, cathepsin k, and atp6v0d2 were restored ( Figure 4K).
Next, an in vivo experiment was undertaken to verify whether SHH mediates BMPR1α-induced osteolytic lesion. Sonic hedgehog agonist SAG, which helps activate hedgehog signaling, was administrated. The 3D reconstruction of mouse tibia indicated that SAG treatment prevented the bone preservation effect of BMPR1α knockdown ( Figure 4L,M). The TRAP staining verified recovered osteoclast numbers after SAG administration ( Figure 4L,N).
Immunohistochemistry also confirmed that RANKL was recovered under SAG stimulation ( Figure 4L,O). In addition, SHH stimulation directly increased the expression of related mRNA, TRAP + osteoclasts, F-actin ring formation, and bone resorption capacity of osteoclasts ( Figure S4).

| Hedgehog inhibition protects bone tissue from BMPR1α-induced osteolytic lesion
Next, we used GDC-0449, an inhibitor of hedgehog signaling targeting SMO, to test the potential therapeutic effect of SHH inhibition. 25 GDC-0449 weakened the enhanced osteoclast differentiation, F-actin ring formation, and resorption ability induced by the conditioned medium ( Figure 5A-D). In addition, the expression of osteoclast-related genes was also decreased in the hedgehog inhibition group compared to the conditioned medium group ( Figure 5E).
Next, the anteromedial tibia tumor cell implantation model was used to verify the effect of hedgehog inhibition on tumor-related osteolytic lesions. After treatment with GDC-0449, the tumor volume decreased ( Figure 5F,G). Furthermore, the 3D reconstruction of mouse tibia showed that the bone loss induced by tumor invasion was significantly ameliorated by hedgehog inhibition (Figure 5H,I).
Meanwhile, decreased osteoclast numbers were observed in the GDC-0449 group ( Figure 5H,J). In addition, IHC confirmed decreased RANKL levels in the tumor area ( Figure 5H,K). Therefore, our results prove that BMPR1α promotes osteolytic lesion of OSCC by SHH-dependent osteoclastogenesis ( Figure 6) and has a therapeutic effect on OSCC-induced bone defect by inhibiting hedgehog signaling.

| DISCUSS ION
Our data suggested that BMPR1α expression was positively correlated with OSCC tumor progression and mandibular invasion.
Furthermore, we discovered that BMPR1α expression drove osteoclast differentiation to promote bone invasion of OSCC in an SHH signaling-dependent manner, whereas targeting SHH or BMPR1α protected the bone from tumor invasion. These findings suggest an important role of BMPR1α and SHH signaling in osteoclast differentiation and OSCC-induced osteolytic destruction, thus providing a potential strategy for molecular-targeted treatment.
Oral squamous cell carcinoma is an aggressive tumor associated with poor prognosis. [26][27][28] The cross-talk between cancer The BMP family is the main growth factor with an important role in bone regeneration. 40  Bone invasion is considered a later stage of a disease, and the surgical treatment at this point can destroy jaw function and cause aesthetic problems. In the approaching era of personalized medicine, the current treatment methods targeting bone invasion environments of OSCC are provided to the patient with limited consideration of the cancer cells' origin. Our new outlook suggests delivering individual tumor microenvironment treatments based on the expression level/activity/functionality of tumorderived factors, rather than utilizing a commonly shared therapeutic umbrella approach, therefore providing targeted treatment in combination with surgical treatment. We also provided evidence that hedgehog inhibition might be used as a potential strategy for blocking BMPR1α-induced bone defect and tumor progression. The notion of "BMPR1α-SHH-associated bone remodeling" could be a step toward a specific personalized therapy for OSCC generating a different bone niche in patients afflicted with incurable bone invasion. In this study, however, we did not analyze the metastasis development related to the bone invasion. Thus, more research is needed to further investigate the relationship between BMPR1α and tumor metastasis.

F I G U R E 6
In oral squamous cell carcinoma (OSCC) tissues, overexpression of bone morphogenetic protein receptor 1α (BMPR1α) activates downstream signaling and induces expression of Sonic hedgehog (SHH) regulated by transcription factor Smad4. SHH enhances osteoclasts' differentiation and resorption activity in a paracrine way, promoting the bond defect induced by OSCC