Hypoxia Markedly Influences Pluripotent Stem Cell Differentiation to Mesoderm, Ectoderm, and Endoderm and Reduces Subsequent Teratoma Formation

Efficient differentiation of pluripotent stem cells (PSC) to desired cell types for cell replacement therapy remains a challenge, and potential tumor formation by residual PSC in differentiated populations is problematic. Most PSC research is performed in high, non-physiological O₂, but cells during embryonic development are exposed to much lower O₂. Here we report a wide-ranging study showing that physiological O₂ markedly influences differentiation to all three germ layers, with enhanced differentiation to cardiomyocytes and insulin-producing cells in particular, and reduces residual PSC and tumor formation.

We differentiated mouse and human embryonic stem cells (mESC and hESC) and mouse induced pluripotent stem cells under controlled cellular O₂ environments through adhesion culture on highly O₂-permeable silicone rubber membranes. Low O₂ drastically increased cardiomyocyte differentiation from these PSC. Best results were acquired by differentiation of mESC for 6 d at 5% O₂ followed by 20% O₂ for 15 d, resulting in up to 57% cardiomyocytes and 304 cardiomyocytes generated per initial mESC without purification, factors of 5 and 9 higher, respectively, than differentiation entirely at 20% O₂. Low O₂ culture for the first 6 d increased by 3x expression of Mesp1/2, which helps restrict mesodermal cells to the cardiovascular lineage, but did not increase Brachyury T expression, an early mesodermal marker, compared to 20% O₂, thereby suggesting that low O₂ acted by restricting the fate of early mesoderm towards cardiomyocytes.

Low O₂ decreased the fraction and number of Nestin+ cells (ectoderm) by 3x for mESC, but enhanced expression of endodermal genes Sox17, Foxa2, Hnf4a, and Pdx1, stimulating us to examine differentiation of hESC to Insulin+ cells using a protocol developed by ViaCyte, Inc (San Diego, CA). Differentiation to definitive endoderm (DE) was modestly enhanced by culture at 3-8% O₂, but performing differentiation entirely at low O₂ was detrimental for producing Insulin+ cells. Differentiation at 5% O₂ to produce DE with all subsequent steps at 20% O₂ achieved a population with 14% Insulin+ cells, 2x higher than differentiation entirely at 20% O₂. These cells passively secreted C-peptide but were not glucose responsive.

Low O₂ for extended periods after differentiation was complete drastically reduced the amount of residual PSC within, and the tumorigenic potential of, differentiated cell populations. Pluripotency marker (Oct4 and Nanog) expression and the fraction and number of Oct4+ cells was reduced by up to four orders of magnitude at low compared to high O₂. Upon implantation into immunocompromised mice, low O₂-differentiated cells did not form tumors or did so slower than high O₂-differentiated cells, consistent with reduced residual PSC. Cell sorting of SSEA-4+ cells after extended low O₂ culture of hESC further reduced residual PSC and tumor formation rate.

These findings establish that culture O₂ is important for almost every aspect of PSC differentiation. We used highly-O₂ permeable silicone rubber culture dishes for accurate control of cellular O₂ exposure. By modulating O₂ during different stages of differentiation, we achieved substantial increases in cardiomyocyte and Insulin+ cell yields. Extended culture at low O₂ further decreased residual PSC by several orders of magnitude. O₂ control, alone or combined with other methods could be applied to future cell therapy protocols to generate and increase the safety of differentiated cells.