RUNX2 promotes the suppression of osteoblast function and enhancement of osteoclast activity by multiple myeloma cells

Data from a public database were collected to analyze the expression of RUNX2 in MM patients and its correlation with survival. Compared with that in normal plasma cells from healthy donors, the expression of RUNX2 was elevated in MM patient plasma cells (Fig. 1A). Relapsed MM exhibited a higher level of RUNX2 than newly diagnosed MM at baseline (Fig. 1B–D). Kaplan–Meier survival curves were generated for the high and low RUNX2 expression groups, which confirmed that high RUNX2 expression was correlated with shorter overall survival time (Fig. 1E, F). The results indicate that RUNX2 is a poor prognostic factor in terms of survival and relapse in MM.

To confirm the correlation of RUNX2 expression and myeloma bone destruction (MBD), we compared RUNX2 expression between samples with different degrees of MBD. MM patients with MBD showed higher RUNX2 levels than those without MBD (Fig. 2A). Subsequently, we collected 30 bone marrow biopsy sections and 18 fresh bone marrow fluid samples from newly diagnosed MM patients and divided them into mild and severe MBD groups according to the median level of bone destruction. Immunohistochemical (IHC) staining of bone biopsies was assessed on the background of CD138-positive staining to allow comparison of samples with similar tumor cell densities (Fig. 2C). According to both the IHC staining score and qPCR analysis, RUNX2 expression was higher in the severe MBD group than in the mild MBD group (Fig. 2D, E). These results indicate that the degree of MBD is related to the expression level of RUNX2 in MM, and the aggravation of bone destruction is accompanied by an increase in RUNX2 expression, confirming that RUNX2 is a molecule that promotes the development of MBD.

Control and RUNX2 knock-in (k/in) 5TGM1 cells were incubated with serum-free medium for 24 h, and the supernatant was collected as conditioned medium. Subsequently, MC3T3-E1 cells were cocultured with concentrated conditioned medium (CCM) and assessed after 24 h. The RUNX2 k/in CCM group exhibited a higher cell viability rate and a lower apoptosis rate than the control group (Fig. 3A, B, p  < 0.05). The expression of CCND1, which is correlated with cell proliferation, was decreased, and that of CASP3, which is involved in cell apoptosis, was increased in the RUNX2 k/in the CCM group (Fig. 3C). To evaluate the differentiation of osteoblasts, we further detected the expression of differentiation-regulating genes, including OSX, OCN, OPN, DLX2, and ALP, which decreased after RUNX2 k/in CCM incubation (Fig. 3E). Communication with osteoclasts is another important aspect of osteoblast biological function in bone remodeling [22]. Osteoblasts induce osteoclast differentiation and development by secreting nuclear factor-kappa B ligand (RANKL), and this effect is attenuated by the secretion of osteoprotegerin (OPG), which is a soluble RANKL decoy receptor [23]. Compared with control CCM incubation, RUNX2 k/in CCM incubation contributed to the increased expression of RANKL and decreased expression of OPG in MC3T3-E1 cells (Fig. 3F). The results suggest that myeloma cells with upregulated RUNX2 not only inhibit the proliferation and differentiation of osteoblasts but also promote the differentiation of osteoblasts to osteoclasts, which indirectly suppresses osteoclast activity.

RAW264.7 cells were cocultured with control CCM and RUNX2 k/in CCM for 24 h. Compared with control CCM, RUNX2 k/in CCM increased the viability and suppressed the apoptosis of RAW264.7 cells (Fig. 4A, B). The expression of CCND1 was increased and the expression of CASP3 was decreased in the RUNX2 k/in CCM group (Fig. 4C). Representative genes related to osteoclast differentiation, such as TRAF6, NFATC1, and MMP-9, were significantly upregulated after RUNX2 k/in CCM incubation (Fig. 4E). These results support that RUNX2 promotes osteoclast proliferation and differentiation mediated by MM cells.

RNA-sequencing analysis was employed to explore the downstream transformation following RUNX2 upregulation with RUNX2 k/in and control 5TGM1 cells. A total of 1064 upregulated genes and 1633 downregulated genes were identified with an absolute log2-fold change > 1 (Fig. 5A). GSEA was performed to enrich functional terms. The upregulated Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were involved in the spliceosome, while pathways in cancer and focal adhesion were downregulated (Fig. 5B). Gene Ontology (GO) enrichment analysis showed that the downregulated terms included cell adhesion, cell differentiation, positive regulation of cell motility and regulation of cell migration, and the upregulated terms were mostly related to cellular homeostasis and the immune response (Fig. 5C–D). The enrichment results indicate novel RUNX2 pathways related to bone destruction in MM. To verify this, we analyzed the expression of genes related to osteogenesis, and LCN2 and M-CSF were upregulated following RUNX2 overexpression (Fig. 5E, G). Protein‒protein interaction (PPI) analysis supported the communication between LCN2, M-CSF, and RUNX2 (Fig. 5F). Soluble LCN2 was reported to impair osteoblast differentiation and stimulate RANKL production by osteoblasts [24]. M-CSF promotes osteoclast proliferation and acts as a costimulatory molecule with RANKL to induce osteoclast differentiation [25]. These two molecules may participate in the regulation of RUNX2 in osteogenesis, but further evidence is needed.

Myeloma-bearing mice were constructed by tibia injection with control and RUNX2 k/in 5TGM1. After 4 weeks, an in vivo imaging system (IVIS) was used to detect tumor growth and invasion, of which RUNX2 k/in group was significantly advanced compared with the control group (Fig. 6A). The affected tibias were removed after execution and assessed by micro-CT (Fig. 6B). Further statistical analysis revealed a lower ratio of bone volume/total volume, lower trabecular thickness, lower trabecular number, and higher trabecular spacing in the RUNX2 k/in group than in the control group (Fig. 6C). It has been confirmed that RUNX2 upregulation in myeloma cells aggravates bone destruction in vivo.

Gene expression profiles and clinical information of microarray datasets, including GSE24080, GSE118985, GSE77539, GSE4204, GSE31161, and GSE124435, were obtained from the Gene Expression Omnibus (GEO) database (http://​www.​ncbi.​nlm.​nih.​gov/​geo/​). The raw count data were preprocessed with normalization and log2 transformation. Transcript sequencing data (TPM) of MM were downloaded from the MMRF-CoMMpass project on the Genomic Data Commons Data Portal (https://​portal.​gdc.​cancer.​gov/​). Box plots and Kaplan–Meier survival curves were constructed with R packages “ggboxplot” and “survminer” in R Studio (version 4.2.1).

Thirty bone marrow paraffin sections and 15 fresh bone marrow fluid samples from newly diagnosed MM patients and clinical data were collected in the First Affiliated Hospital of Sun Yat-sen University with approval from the ethics committee (No. IIT-2021-799).

Bone marrow paraffin sections were subjected to immunohistochemical staining with anti-CD138 antibody (Abcam: ab181789) and anti-RUNX2 antibody (CST: #12,556). To narrow bias brought by tumor cell density, we calculated the ratio of RUNX2/CD138 average optical density (AOD) by Image ProPlus software (version 6.0) as the relative protein level of RUNX2.

CD138-positive plasma cells were collected from fresh bone marrow fluid through immunomagnetic bead sorting (Miltenyi MACS bead isolation kit). RNA was extracted from CD138-positive plasma cells and subsequently used to assess RUNX2 expression.

The murine MM cell-line 5TGM1, preosteoblast cell-line MC3T3-E1, and preosteoclast cell-line RAW264.7 were kind gifts from Zhangxingding laboratory, Sun Yat-sen University. RUNX2 k/in and control 5TGM1 cell lines were constructed by lentiviral transfection and identified by quantitative polymerase chain reaction (qPCR) analysis and western blot analysis. Cells were cultured with medium supplemented with 10% FBS and 100 μg/mL penicillin‒streptomycin at 37 °C in 5% CO₂.

RUNX2 k/in 5TGM1 cells and control 5TGM1 cells were cultured with serum-free medium at a density of 106 cells/mL for 24 h. Then, the cells were removed, and the liquid supernatant was concentrated with a 100 kD protein concentrator (Millipore: UFC9100) and filtered with a 0.22 µM pore filter at 4 °C [44]. Concentrated conditioned medium (CCM) was obtained after 10 concentration steps.

MC3T3 E1 and RAW264.7 were seeded for 24 h, and then the medium was replaced with serum-free medium (BLANK), control CCM, and RUNX2 k/in CCM. After 24 h of incubation, cell viability was detected by the CCK-8 assay (Vazyme: A311-01), apoptosis was detected with the Annexin V-APC/PI Apoptosis Kit (KeyGEN Biotech: KGA1030), and differentiation was detected by quantitative PCR according to the manufacturer’s instructions.

qPCR was used to assess the expression of genes after total RNA was extracted and reverse transcribed using the PrimeScript cDNA Synthesis Kit (Takara: #6210). Quantitative PCR was performed using a SYBR Green Mater Mix (Vazyme: Q111-02) according to the manufacturer’s instructions. The PCR primers are listed as follows Table 1.

Total RNA was extracted from control and RUNX2 k/in 5TGM1 cells. At least 1 µg of RNA from 3 biological replicates of each group was used for high-throughput RNA sequencing after quality control. Sequencing data were processed at the Beijing Genomics Institute (BGI). Transcripts per kilobase of exon model per million mapped reads (TPM) values of genes were used in the subsequent analysis. Gene Set Enrichment Analysis (GSEA) was employed to evaluate KEGG and GO term enrichment in the control and RUNX2 k/in groups with R package “clusterProfiler” in R Studio (version 4.2.1). PPI networks were constructed with 421 upregulated genes that were screened with log2-fold change > 1.5 in the STRING website (https://​cn.​string-db.​org/​) and Cytoscape software (version 3.9.1).

All procedures involving animals were approved by the Sun Yat-sen University Animal Ethics Committee (No. SYSU-IACUC-2022-000,403). C57BL/6 mice (4 weeks old) received a right tibia injection of control or RUNX2 k/in 5TGM1-GFP cells (10⁶ cells). On day 28, the animals were detected by an in vivo imaging system (IVIS) and then sacrificed. The affected tibias were assessed by microcomputed tomography (micro-CT). The 3D model of bone trabecula was constructed from 0.2 mm below the tibia platform, and trabecula parameters, including bone volume/total volume (BV/TV), trabecular thickness (Tb. Th), trabecular number (Tb. N), and trabecular spacing (Tb. Sp) were assessed.

Statistical analysis was performed with R Studio (version 4.2.1) and GraphPad Prism software (version 9.0). The Mann–Whitney test and Student’s t test were used to determine significance as appropriate after the normality test. Values are presented as the mean ± standard error of the mean. A p value < 0.05 was considered statistically significant. All statistical tests were two-sided.