Loss of androgen receptor promotes adipogenesis but suppresses osteogenesis in bone marrow stromal cells

Abstract

Gender differences have been described in osteoporosis with females having a higher risk of osteoporosis than males. The differentiation of bone marrow stromal cells (BMSCs) into bone or fat is a critical step for osteoporosis. Here we demonstrated that loss of the androgen receptor (AR) in BMSCs suppressed osteogenesis but promoted adipogenesis. The mechanism dissection studies revealed that AR deficiency suppressed osteogenesis-related genes to inhibit osteoblast differentiation from BMSCs. Knockout of AR promoted adipogenesis of BMSCs via Akt activation through IGFBP3-mediated IGF signaling, and the 5' promoter assay and chromatin immunoprecipitation assays further proved that AR could modulate IGFBP3 expression at the transcriptional level. Finally, addition of IGF inhibitors successfully masked the AR deficiency-induced Akt activation, and inhibitions of Akt, IGF1, and IGF2 pathways reversed the AR depletion effects on the adipogenesis process. These results suggested that AR-mediated changes in IGFBP3 might modulate the IGF-Akt axis to regulate adipogenesis in BMSCs.

Figure 1. ARKO mice exhibit osteoporosis and enlarged adiocytes

(A) Genotyping of WT and ARKO mice. AR^wt indicates intact AR and AR^del is mutant AR. Cre bands are cre recombinase positive. (B) Upper panel: The external genitalia of male WT and ARKO mice. Lower panel: The internal genitalia of WT and ARKO mice. Arrowheads and arrows indicate testis and bladder, respectively. (C) Representative heart images of WT and ARKO mice. Graph is the quantification of heart weight over body weight (HW/BW). (D) Representative images of trabecular bone from WT and ARKO mice. Graph is bone volume over tissue volume (BV/TV). (E) Representative images of white adipose tissues and brown adipose tissues of WT and ARKO mice. (F) Fertility test in WT and ARKO mice by measuring numbers of pups/litter number. (G) DHT concentrations in WT and ARKO mice. For all studies, n=6 from either WT or ARKO mice. *, p value < 0.05. **, p value <0.001. ***, p value <0.0001. p values were generated with the student's t test.

Figure 2. Identifying key regulators involved in AR-mediated osteogenesis and adipogenesis in BMSCs

(A) Classifications of top 10 categories of signature genes that are positively correlated or regulated by AR. Highlighted are genes previously reported to promote osteogenesis (blue color) and inhibit adipogenesis (*). (B) AR-mediated key factors in bone development. Black letters within red objects are major categories of key factors. Green letters beside the red objects are full gene names. (C) Osteogenesis determination by using alizarin red staining in WT and ARKO BMSCs. (D) Akp2, (E) IBSP, and (F) Dmp1 expression were measured in WT and ARKO BMSCs after 30 days of osteogenesis induction. (G) Alizarin red staining was performed in human bone marrow stromal cells (hBMSCs), which were manipulated with siRNA scramble control (sc) or siRNA-AR (siAR). Quantitation is shown in the graph. *, p value < 0.05, ***, p value <0.0001. p values were generated with the student's t test.

Figure 3. AR deficiency increases progenitors and adipogenic potentials of stromal cells

(A) Representative images of colonies in WT and ARKO mice. Graph on the right is quantification of colony numbers. (B) Colony forming efficiency analysis was performed in WT and ARKO ADSCs. Graph on the right is the quantification of colony numbers. (C) Adipogenesis analysis of WT and ARKO BMSCs using Oil Red O staining. (D) aP2, LPL, and PPARγ mRNA levels were determined to quantify the extents of adipogenesis in WT and ARKO BMSCs. (E) Oil Red O staining was performed in hBMSCs, which were manipulated with scramble control (sc) or siRNA-AR (siAR). Quantitation result is shown in the graph. (F) aP2, LPL, and PPARγ2 were determined in hBMSCs, which were manipulated with either sc or siAR. *, p value < 0.05. **, p value <0.001. p values were generated with the student's t test.

Figure 4. AR deficiency suppresses activation of Akt by IGFBP3 in BMSCs

(A) IGF1, IGF2, IGFBP3, IGF1R, and IGF2R were measured using qRT-PCR in WT and ARKO BMSCs. n=8 from either WT or ARKO mice. (B) IGFBP3 mRNA and (C) protein expression were examined in hBMSC-sc and hBMSC-siAR. (D-F) IGFBP3, Akt, pAkt, Erk1/2, pErk1/2, JNK, and pJNK, were determined in 3 WT and ARKO BMSCs. Quantification ratios of pAkt over Akt (pAkt/Akt), pErk1/2 over Erk1/2 (pErk/Erk), and pJNK over JNK (pJNK/JNK) were determined and shown in graphs below gel images. (G) AR, pAkt, and Akt were measured in ARKO ADSCs transfected with vector and AR plasmids. (H) IGF inhibitor effects on Akt activation were tested in WT and ARKO BMSCs. pAkt expression is dose dependently inhibited by IGF inhibitor in WT and ARKO BMSCs. *, p value < 0.05. p values were generated with the student's t test.

Figure 5. AR modulates IGFBP3 through promoter regulation

(A) Illustration of the IGFBP3 promoter region and 2 putative AREs, ARE1 located from -819∼-794 and ARE2 located from -1816∼-1802. (B) Map of the constructed pGL3-IGFBP3 luciferase. A segment of IGFBP3 promoter is from the location -2026 to +7. The two insertion sites are Kpn I and Mlu I. (C) Luciferase assays of MMTV-luciferase (MMTV) and pGL3-IGFBP3-luciferase (IGFBP3 promoter) were performed in HEK-293T cells. 1 ng pRL-Tk was used as control for luciferase. 0.5 μg vector, mAR, MMTV, and IGFBP3 promoter were transfected into HEK-293T cells either in the presence or absence of 10 nM DHT. (D) Chromatin immunoprecipitation assay was performed in WT and ARKO BMSCs. Non-immune 10% input served as positive control for IGFBP3 primers specific for ARE2. IgG was used to pull down chromatin complex as a negative control. AR antibody was used to pull down chromatin complex as experimental targets. ***, p value < 0.0001. p values were generated with the student's t test.

Figure 6. Inhibition of IGFs and Akt signaling pathways reverse the AR deficiency effects on BMSC adipogenesis

(A) Oil Red O staining was used to visualize oil droplets, and qRT-PCR was used to determine adipogenisis markers including LPL, aP2, and PPARγ2 in the differentiated adipocytes from WT and ARKO BMSCs in the presence of 10 μM LY294002 (Akt inhibitor). (B) 1 μM Chromeceptin (IGF II inhibitor) and (C) 2 μM PPP (IGF I inhibitor) were used to treat WT and ARKO BMSCs, which were under adipogenic induction and the adipogenic differentiation was monitored by Oil Red O staining and qRT-PCR to determine oil droplets and adipogenesis markers. *, p value < 0.05. **, p value < 0.001 ***, p value < 0.0001. p values were generated with the student's t test.