We used immunohistochemistry to measure

We used immunohistochemistry to measure the expression of RUNX2, SP7, and GFP in the induced 2.3 kb Col1a1-GFP mESCs. RUNX2, SP7, and the 2.3 kb Col1a1-GFP were highly expressed in the induced order ldv on day 19. RUNX2 and SP7 were colocalized in nuclei, and 2.3 kb Col1a1-GFP was largely observed in the cytoplasm and nuclei (Figure 2C). The average percentages of RUNX2-, SP7-, and GFP-positive cells were 78% ± 3%, 66% ± 5%, and 45% ± 1%, respectively (Figure 2D). The different percentages of those markers in our culture are consistent with the physiological process of osteoblast development. The percentages enable us to estimate the final yields of each osteoblastic population, given that 900,000 ± 150,000 cells/cm2 were obtained on day 19 of the culture from 100,000 cells/cm2 of input. Pluripotency markers were hardly expressed in the induced cells on day 19 (Figure S2D). The average percentages of OCT4-, NANOG-, and SOX2-positive cells were 2.3% ± 0.6%, 1.5% ± 0.7%, and 9.0% ± 3.1%, respectively (Figure S2D). In addition, BRACHYURY (T), a mesoderm maker, was ubiquitously expressed in the induced cells on day 5 (Figure S1C). von Kossa staining and alizarin red staining revealed the uniform formation of calcified cell clusters by days 19 and 23 (Figures 2E and 2F). These results indicate that the present strategy induced both the expression of osteoblast-related genes and the calcification of matrix, two key features of osteoblasts. To confirm that the present strategy could be useful for investigating osteoblast development using gene-manipulated mESCs in vitro, we examined whether Runx2−/− mESCs cultured using the present strategy could molecularly recapitulate osteoblast phenotypes in Runx2−/− mice (Figure S3A). Runx2−/− mESCs showed similar transient upregulation of T to Runx2 mESCs, but Sp7 and Bglap were hardly upregulated in Runx2 mESCs on day 19 compared to day and day 5. Importantly, the expression of the early osteoblast marker gene Ibsp showed a 212-fold upregulation in Runx2−/− mESCs on day 19, although the level was lower than that observed in Runx2 mESCs. In alizarin red staining, the calcification level of Runx2−/− mESCs was lower than that of Runx2 ones on both days 19 and 23 (Figures 2F, S3B, and S3C). These observations were consistent with the bone phenotypes of Runx2−/− mice (Komori et al., 1997; Tu et al., 2012). Thus, the present strategy can at least partially recapitulate physiological osteoblast development and will be useful for analyzing osteoblast development using gene-manipulated ESCs in vitro. Because the direct differentiation of mouse iPSCs (miPSCs) and human iPSCs (hiPSCs) into osteogenic cells has been previously reported by Bilousova et al. (2011), Kao et al. (2010), and Levi et al. (2012), we applied the present strategy to 2i-adapted miPSCs established from fibroblasts of mice expressing GFP (CAG-GFP miPSCs) (Okabe et al., 1997) (Figure 3A). Nanog was downregulated throughout the culture compared to day 0. T and Mixl1 were transiently upregulated by mesoderm induction on day 5. The osteoblast-related genes Runx2, Sp7, Col1a1, Ibsp, and Bglap were upregulated by the osteoblast induction (Figure 3B). RUNX2 and SP7 proteins were highly expressed in the induced cells on day 23, and their signals were merged with that of GFP (Figure 3C), suggesting that CAG-GFP miPSCs were differentiated into cells expressing osteoblast-related proteins. Calcified cell clusters were formed on days 19 and 23 (Figure 3D). In the comparison of calcification levels between mESCs and miPSCs, both cell types were calcified at a similar level on day 19; mESCs then showed more calcification on day 23 (Figure 3E). Thus, the present strategy efficiently differentiates both mESCs and miPSCs into osteoblasts by sequentially treating the cells with four small molecules, CHIR, Cyc, SAG, and TH, under chemically defined conditions.