In this study, we used an inducible, lentiviral system to generate GFP positive, pluripotent iPS cells from MEFs derived from Oct4-GFP/R26-M2rtTA and Nanog-GFP/R26-M2rtTA mice. We established iPS cell lines that were able to differentiate in teratoma assays and could produce viable chimeras. This is in contrast to iPS cells harboring constitutively expressing lentiviral constructs that are not efficiently silenced in iPS cells. The inducible system allowed us to investigate the timing and sequence of ES cell marker gene activation using FACS analysis. We found direct in vitro reprogramming to be a gradual process encompassing the sequential activation of four pluripotency marker genes, with AP being expressed first on day 3 followed by SSEA1 on day 9. GFP expressed from the endogenous promoters of Oct4 or Nanog was first detectable on day 16 (). SSEA1−/GFP− cells were not able to activate GFP expression in the same time frame as SSEA1+/GFP− cells and the progress from the SSEA1+/GFP− state to the dox-independent, fully reprogrammed SSEA1+/GFP+ state was greatly enhanced by continued transgene expression. Together, these results suggest that SSEA1+/GFP− cells may represent a transgene-dependent intermediate state of reprogramming. These partially reprogrammed cells required continued transgene expression in order to progress towards a fully reprogrammed state, whereas transgene downregulation caused reversion to a fibroblast-like, SSEA1-negative state. Finally, we demonstrated that the generation of iPS cells requires the ectopic expression of the four transcription factors for a minimum of 12–16 days (). The removal of dox prior or on day 12 resulted in the return of colony-forming cells to a MEF-like state. The independence from transgene expression closely correlated with the reactivation of the endogenous Oct4 and Nanog loci (). We postulate that the activation of the endogenous Oct4 or Nanog may be a marker for fully reprogrammed, transgene-independent iPS cells.
Reprogramming of somatic cells: sequential marker activation and time of virus expression
The sequential activation of pluripotency marker genes found in this study is consistent with our previous observations from MLV-based reprogramming strategies where cells at day 14 after MLV-infection expressed AP and SSEA1, but not Nanog, whereas cells had activated all three markers at day 20 (Wernig et al., 2007
). Intriguingly, Nanog-neo MEFs in that study yielded neomycin resistant colonies even when the selection process was initiated at day 6 after infection, a time point well before Nanog expression could be detected by immunostaining (Wernig et al., 2007
). Similar results were reported in Okita et al., 2007
, and Maherali et al., 2007
, where puromycin resistant colonies were derived from Nanog-GFP-IRES-puro MEFs by starting puromycin selection as early as 3 days after infection. This discrepancy in timing between the appearance of antibiotic resistance and protein detection could be due to a low level of initial expression that is sufficient to render the partially reprogrammed cells drug resistant, but below the level required to visualize GFP and maintain pluripotency. This may be the result of a gradual process of gene reactivation that occurs at different times for specific markers, with AP and SSEA1 being activated earlier than Oct4 or Nanog ().
We found the time span at which transgene expression becomes dispensable for iPS cell derivation (12–16 days) to precede the appearance of GFP expression which marks the full activation of the endogenous Oct4 and Nanog loci (day 16). Since GFP detection by FACS requires a significant level of protein expression, it is likely that this method overestimates the minimum time span required for Nanog and Oct4 activation. Delayed detection of GFP expression could also explain why SSEA1+/GFP− cells sorted at day 21, but not at day 9, contained a small fraction of transgene-independent cells. These cells might have activated the Oct4 or Nanog loci but did not as yet display GFP expression levels high enough for FACS detection. Sustained transgene expression beyond the minimal time requirement increased the number of cells activating endogenous Nanog and Oct4 expression, supporting the idea of stochastic epigenetic events playing a role in four-factor reprogramming. This is consistent with our previous observation that infected cell populations continue to generate iPS colonies over a drawn-out time window (Meissner et al., 2007
). Our results suggest that individual cells either enter the reprogramming process at different time points after transgene induction or take different times to go through the reprogramming sequence. Longer transgene expression, therefore, would give more cells the chance to undergo the required stochastic epigenetic changes and, consequentially, proceed to a state of transgene independence.
In previous reports on iPS cell derivation from somatic cells, transgene expression was driven by the LTRs of MLV based vectors, which were shown to be efficiently silenced in iPS cells (Maherali et al., 2007
; Okita et al., 2007
; Wernig et al., 2007
). When we used constitutively expressed lentiviral vectors to generate iPS cell lines, we found these cells to be poorly capable to differentiate in teratoma assays. This result supports the notion that efficient transgene silencing is essential for the derivation of truly pluripotent iPS cell lines. In a recent publication, iPS cell lines generated using constitutive lentiviral vectors were reported to differentiate in teratoma assays and to contribute to various tissues of mid-gestation chimeric fetuses (Blelloch et al., 2007
). The cDNAs in that study were driven from the CMV promoter, which has been shown to undergo methylation-mediated silencing in embryonic stem cells (Hong et al., 2007
; Xia et al., 2007
), in contrast to the Ubiquitin C promoter employed in our constitutively expressed viral constructs (Lois et al., 2002
). It is possible that the differentiation capabilities observed in iPS cells harboring CMV-driven cDNAs could be a result of at least partial silencing of the viral transgenes. Notably, no viable chimeras have been reported from iPS cell lines derived using constitutive lentiviral constructs so far (Blelloch et al., 2007
Very recently, it has been reported that mouse and human iPS cells can be generated without the use of a c-Myc transgene (Nakagawa et al., 2008
; Wernig et al., 2008
; Yu et al., 2007
). However, the forced expression of c-Myc, as well as other factors such as Nanog and Lin28, has been found to have a substantial effect on the efficiency and the timing of the reprogramming process. While these reports are significant steps towards reducing the tumorigenic potential of iPS cells and their derivatives, the final solution to this problem will be the generation of transgene-free iPS cells. The ability to quantify reprogramming dynamics in a controlled system as presented here will be an invaluable tool for the development of transient reprogramming strategies.
In conclusion, the results presented in this report clarify steps involved in the generation of iPS cells by the expression of the four transcription factors Oct4, Sox2, Klf4 and c-Myc. The information on pluripotency marker activation and transgene expression provided here can be utilized as a benchmark for further analyses of the reprogramming process and should allow for the identification of factors that positively affect epigenetic reprogramming. When compared to nuclear transfer, it is obvious that reprogramming by viral transduction requires a longer period of time, and it will be important to understand the molecular basis for this difference. The determination of the minimum length of transgene expression has implications for the development of non-retroviral delivery methods of these four factors to derive genetically unmodified iPS cells. Specifically, any transient expression strategy for iPS cell generation, such as protein transduction, will need to provide protein expression at sufficient levels for a minimum of 12–16 days. The generation of transient, non-viral approaches in conjunction with a better understanding of the reprogramming process will be an important step in the development of stem cell based therapies.