During the course of these studies, we were most surprised by what human ES and iPS cells do not need, and by how challenging it is to remove something from a medium once its inclusion becomes routine and accepted. Albumin, for example, has had many cell culture roles attributed to it, from lipid carrier and blocking reagent to physical protection against sheer force11,23
. However, our results clearly demonstrated that the dominant role of BSA in TeSR is to prevent a toxic effect of β-mercaptoethanol and that, in the absence of β-mercaptoethanol, BSA is not required (). β-mercaptoethanol’s inclusion in our previous TeSR medium for human ES cells traces its origin back to a positive effect on cloning efficiency of mouse EC cells described in 197824
. Because of those results, β-mercaptoethanol was included in mouse ES cell media25
and subsequently in early human ES cell media1
The finding of an interaction between BSA and β-mercaptoethanol underscores the challenge of optimizing media, as it is not sufficient to just examine the effects of the addition or subtraction of individual components. We thus reexamined all other components of TeSR medium made without BSA or β-mercaptoethanol and found that several (i.e., pipocolic acid, GABA, LiCl, chemically defined lipids, trace elements B, trace elements C, glutathione, and additional Thiamine and L-Glutamine) no longer had a positive effect. In spite of the much greater simplicity of E8 medium, it supports the long-term undifferentiated proliferation of both human ES and iPS cells comparably to TeSR ().
E8 medium supports higher reprogramming efficiencies for both viral and episomal approaches (, and Supplementary Fig 2f
). In every attempt, from both established, banked fibroblasts (5) and from independent biopsy samples (4), integration free iPS cell lines were successfully derived in E8-based medium on vitronectin (Supplementary Table 2
). With the exception of the banked neonatal fibroblasts that reprogrammed at a high efficiency, the efficiencies obtained directly from biopsy samples grown in E8-based medium were consistently higher than those of established cell lines obtained from commercial sources, suggesting that fibroblast passage history is important. Although these efficiencies could be further improved, the number of clones obtained per biopsy sample in these defined conditions using an episomal approach (60 - ~1,000 iPS cell colonies / 106
fibroblasts) already greatly exceeds what is typically used for subsequent characterization.
It is important to distinguish mechanistically between culture components that improve reprogramming itself, and those that merely improve the survival and proliferation of the resulting iPS cell clones19,26
. Because reprogramming occurs over time, it is often difficult to cleanly distinguish between these two effects, and indeed, they do not need to be mutually exclusive. We found, for example, that L-ascorbic acid promotes ES cell proliferation and expansion (). L-ascorbic acid has previously been reported to promote reprogramming27
, but our results are consistent with L-ascorbic acid only promoting iPS cell growth and survival, with reprogramming itself being driven by other factors. It is also possible that ascorbic acid leads to epigenetic modifications essential to both reprogramming and cell survival in serum-free culture conditions28
. Similarly, most derivation procedures have a splitting step before iPS colonies are visible, and reprogramming efficiency is often calculated based on the final iPS colony number. When splitting is required for the emergence of iPS cells in some situations (e.g. fibroblast overgrowth) it can lead to confusion between factors that improve reprogramming itself, and those that improve survival after splitting. Importantly, if the iPS cell colony number is only assessed after splitting, an independent clonal origin of distinct iPS cell colonies cannot be confirmed, making calculations of true reprogramming efficiency problematic. Because our method removes TGFβ and hydrocortisone from the defined medium after 5–10 days, fibroblast over-growth is inhibited and truly independent iPS cell clones can be isolated and counted without splitting.
Because the E8 medium reduces medium cost and simplifies quality control, it is now used for all routine culture of both human ES and iPS cells in this laboratory. This simplified, defined medium also provides a much cleaner background for examining specific pathways in self-renewal, cell death, and differentiation13
and it supports substantially improved reprogramming efficiencies. Although we have only demonstrated improved efficiencies for viral and episomal reprogramming approaches, these conditions should be equally useful for other non-integrative reprogramming approaches. Finally, since E8 medium is highly defined, it should also help facilitate the transfer of basic research on human pluripotent stem cells to the clinic.