Through the systematic comparison of the transcriptional profiles and epigenetic signatures of iPSCs before and after excision of a single copy of a polycistronic reprogramming cassette we have gained novel insights into the molecular events that occur during iPSC generation. Here, we provide evidence indicating that i) persistent transgene expression prevents iPSCs from attaining an ESC-like transcriptional program; ii) iPSCs reprogrammed with and without c-Myc exhibit highly similar gene-expression profiles; iii) reprogramming results in aberrant imprinting of the Dlk1-Dio3 locus in some but not all iPSC clones generated with our polycistronic cassette; and iv) Klf4 binds strongly to the Gtl2 promoter and shows decrease binding upon removal of the reprogramming transgenes.
Lentiviral vectors seem to resist methylation-dependent silencing in mouse pluripotent cells 
. Consequently, incomplete transgene silencing following reprogramming has been shown to negatively affect the differentiation potential of iPSCs, possibly by antagonizing the transcriptional programs triggered by developmental cues 
. We found that transgene-carrying iPSCs display a transcriptional pattern that distinguishes them from transgene-free iPSCs. Moreover, the degree of transcriptional dysregulation is correlated with residual transgene activity and relatively small (1.5- to 2.5-fold) increases in exogenous reprogramming factor expression appear to elicit genome-wide transcriptional changes that can be identified by PCA. It should be noted that transgene-carrying iPSCs have activated the endogenous pluripotency regulators Nanog
and do not represent an intermediate partially reprogrammed state. Our findings suggest that rather than being associated with a random perturbation of the genome, residual expression of the reprogramming factors in iPSCs induces a genetic program that supports their self-renewal and their ability to differentiate, albeit at a reduced efficiency 
. In line with this notion, transcriptional differences between iPSCs and ESCs are attenuated following withdrawal of exogenous reprogramming transgenes. Conceivably, these changes reflect the fine-tuning of the regulatory circuitry underlying pluripotency that occurs only after exogenous expression of the Yamanaka factors is withdrawn.
Pluripotency can be induced in the absence of c-Myc overexpression, albeit at a low efficiency and delayed kinetics 
. Furthermore, the experimental evidence suggests that the main role of this factor is to suppress somatic cell-specific gene expression during the initial stages of reprogramming 
. Consistent with these observations, we found that the addition of c-Myc to the reprogramming cocktail has little to no effect on the transcriptional pattern of iPSCs, even under continuous transgene expression. Notably, evidence exists that iPSCs produced in the absence of c-Myc behave differently. For example, OKS-iPSCs displayed reduced competence for germline transmission compared to OKSM-iPSCs 
, but exhibited enhanced in vitro
cardiogenic potential in another study 
. Apart from residual expression of c-Myc during differentiation or in vivo
development, the reasons for these differences remain unclear. However, because c-Myc might participate in chromatin remodeling during reprogramming 
, we cannot exclude a potential effect of this factor on the epigenome of reprogrammed cells that would be manifested at the time these cells are coaxed to differentiate. A comprehensive analysis of the epigenetic profiles of OKS-iPSCs vs OKSM-iPSCs could shed light on the mechanisms behind these observations.
Epigenetic modification of the Gtl2
locus during reprogramming seems to be important for the generation of iPSCs with full developmental potential 
. Our results indicate variability in the frequencies of mis-imprinted iPSC clones derived by reprogramming dermal fibroblasts. Importantly, all iPSC lines were derived from male mice, thus ruling out the possibility that the epigenetic differences might be due to the lower global and DMR-specific methylation previously reported for female ESCs 
. Specifically, we found that reprogramming of C57BL/6 fibroblasts by means of a single polycistronic vector gave rise to hypermethylated Gtl2OFF
iPSC clones, while the same experimental approach yielded mostly clones with low to normal Gtl2
methylation levels when C57BL/6×129/sv fibroblasts were transduced. Moreover, two clones underwent Gtl2 silencing after transgene excision, suggesting that the reprogramming factors may be directly involved in this process. In support of this idea, binding sites for Oct4 and Klf4 were identified in the Dlk1-Dio3
region, and Chip-qPCR analysis revealed increased recruitment of Klf4 near the TSS of Gtl2
in transgene-carrying iPSCs. Our findings imply an active role of Klf4 (and possibly Oct4) in establishing the methylation status of Gtl2
and suggest that, when present at supraphysiological levels, Klf4 may protect this region from cytosine methylation through a mechanism similar to that described for the imprinted Igf2
. These results are in concordance with recent studies emphasizing the role of Klf4 in establishing appropriate Gtl2 imprinting 
. In addition, some of the epigenetic modifiers identified as differentially expressed in our microarray analysis could also play a role. For example, Dnmt3a, which was found to be downregulated in transgene-carrying iPSCs, methylates Gtl2/Dlk1
DMRs in vivo
. Another possible alternative is that Oct4 and Klf4 binding promotes the activity or recruitment of DNA methyltransferases thus altering the methylation status of the Dlk1-Dio3 locus. A recent study by Stadtfeld et al. provides experimental support for this last hypothetical model by demonstrating that hypermethylation of the Dlk1-Dio3 IG-DMR occurs late in the reprogramming process and is catalyzed by Dnmt3a 
. Moreover it is suggested that the absence of ascorbic acid results in loss of the histone acetylation active marks at the chromatin state thus facilitating the recruitment of Dnmt3a and the resulting hypermethylation of Dlk1-Dio3 locus 
. In combination our data and these recent findings suggest that Oct4 and Klf4 binding at the Dlk1-Dio3 locus results in chromatin alterations marked by hypomethylation of the locus when the reprogramming genes are constitutively expressed. Removal of the reprogramming cassette results in resolution of the DMRs in this locus to a normal (50% methylated) or aberrantly imprinted (hypermethylated) state. Indeed, different combinations/stoichiometries of reprogramming factors appear to have an effect on the epigenetic status of Gtl2
as recently shown 
. Lastly, further studies are needed to address the possibility that aberrant imprinting is affected by cellular changes that accompany the subcloning and expansion of iPSCs.
In summary, we demonstrate that residual expression of exogenous reprogramming factors has a pervasive effect on the transcriptional program of mouse iPSCs and may also influence epigenetic signatures associated with full pluripotency. Although retroviral and lentiviral vectors undergo silencing in human pluripotent cells, our findings suggest that variegation effects as well as potential reactivation of the transgenes during differentiation could have a negative impact on the biological properties of iPSCs. Indeed, even small variations in the levels of pluripotency factors appear to have a profound effect on the early cell fate choices of ESCs 
. Collectively, our data demonstrate the importance of generating iPSCs that are free of reprogramming transgenes for both research and therapeutic applications.