Generation of Human Induced Pluripotent Stem Cells Using Epigenetic Regulators Reveals a Germ Cell-Like Identity in Partially Reprogrammed Colonies

Previous studies have shown that induced pluripotent stem cells (iPSCs) can be derived from fibroblasts by ectopic expression of four transcription factors, OCT4, SOX2, KLF4 and c-MYC using various methods. More recent studies have focused on identifying alternative approaches and factors that can be used to increase reprogramming efficiency of fibroblasts to pluripotency. Here, we use nucleofection, morpholino technologies and novel epigenetic factors, which were chosen based on their expression profile in human embryos, fibroblasts and undifferentiated/differentiated human embryonic stem cells (hESCs) and conventionally generated iPSCs, to reprogram human fibroblasts into iPSCs. By over expressing DNMT3B, AURKB, PRMT5 and/or silencing SETD7 in human fibroblasts with and without NANOG, hTERT and/or SV40 overexpression, we observed the formation of colonies resembling iPSCs that were positive for certain pluripotency markers, but exhibited minimal proliferation. More importantly, we also demonstrate that these partially-reprogrammed colonies express high levels of early to mid germ cell-specific genes regardless of the transfection approach, which suggests conversion to a germ cell-like identity is associated with early reprogramming. These findings may provide an additional means to evaluate human germ cell differentiation in vitro, particularly in the context of pluripotent stem cell-derived germ cell development, and contribute to our understanding of the epigenetic requirements of the reprogramming process.


Introduction
Human embryonic stem cells (hESCs) are derived from the inner cell mass (ICM) of blastocysts and are characterized as pluripotent due to their ability to self-renew and give rise to all types of cells in the body. In 2007, Yamanaka and colleagues generated induced pluripotent stem cells (iPSCs), which are similar to hESCs in terms of morphology, gene expression and the ability to form cell types of all three germ layers both in vitro and in vivo [1][2][3]. iPSCs provide a platform for studying human development and disease as well as the potential to develop innovative patient-specific therapies with decreased risk of immune rejection relative to hECCs since the patient's own cells might be used for therapy [4][5][6][7].
Initially, Yamanaka and colleagues reprogrammed fibroblasts by using four transcription factors (OCT4, SOX2, KLF4 and c-MYC) in viral vectors [3,8]. However, this method has several drawbacks and thus, recent studies have focused on eliminating the use of c-MYC and utilizing alternative methods of reprogramming, including excisable constructs, nonintegrating plasmids adenovirus, episomal and piggybac transposon vectors to circumvent the genomic integration of viral transduction and increase reprogramming efficiency [9][10]. Other DNA-free methods such as Sendai virus, mRNA, microRNA and protein reprogramming have also been explored [11][12][13][14][15]. In general, two different approaches, the introduction of novel factors or the addition of cell permeable chemical compounds, either alone or in conjunction with one another have also been successfully used to increase the reprogramming efficiency of iPSCs. For the first approach, factors typically utilized for cell immortalization such as hTERT and the SV40 large T antigen have been transfected together with the four Yamanaka factors into human fetal, neonatal and adult dermal fibroblasts [16]. While hTERT and SV40 may increase cell growth in surviving colonies, they may also induce significant cell death and aneuploidy as previously shown [17]. Other studies have shown that additional novel factors, including NANOG, LIN28, REX1, Zfp296 and Glis1 can be used as a substitution together with one or more of the Yamanaka factors to make the reprogramming process more efficient [18]. Alternatively, iPSCs have also been derived using small molecule compounds such as Valproic Acid (VPA) and 5-Aza-2´-deoxycytidine (AZA), which may substitute for c-MYC during transfection and are thought to act by inducing epigenetic remodeling [19].
Despite recent studies demonstrating increased efficiency of iPSC generation, several reports have shown that iPSCs differ epigenetically from their hESC counterparts, which can be explained, at least in part, by incomplete DNA methylation or histone modification during cellular reprogramming [20][21][22]. Here, we explored use of various combinations of different novel epigenetic factors to reprogram human fibroblasts into iPSCs. We chose the factors used for reprogramming based on their expression profile in human embryos, fibroblasts and undifferentiated/differentiated hESCs and previously derived hiPSCs. By nucleofecting fibroblasts with plasmids containing DNMT3B, AURKB and PRMT5 in conjunction with SETD7 silencing via morpholino technologies, we demonstrate an early role for these epigenetic factors in reprogramming and reveal a germ cell-like identity in partially reprogrammed colonies.

Ethics statement
Human blastocysts donated for non-stem research were obtained with written informed consent from the Stanford University Regenerative Medicine through the Ethical procurement of Nonviable or Excess cellular Waste (RENEW) Biobank as previously described [23]. De-identification and molecular analysis was performed according to the Stanford University Institutional Review Board (IRB)-approved protocol #10466 entitled "The RENEW Biobank." No protected health information was associated with each of the blastocysts.

Gene expression analysis
Pre-amplification was performed according to the manufacturer's protocol (Fluidigm Corp.; So. San Francisco, CA) using the CellsDirect One-Step qRT-PCR kit (Invitrogen) and 20X TaqMan gene expression assays (Applied Biosystems, Foster City, CA). For undifferentiated and differentiated hESC lines and conventionally generated hiPSCs, RNA was first extracted using the Qiagen RNeasy Kit (Valencia, CA) and cDNA prepared with the SuperScript III First-Strand Synthesis System (Invitrogen). Between 1 and 250ng of cDNA was pre-amplified by adding 5μl CellsDirect 2X reaction mix, 2.5μl 0.2X PPP (2.5μl each 20X TaqMan gene expression assay plus Tris-EDTA buffer for a final volume of 250μl), 1μl Platinum Taq (Invitrogen), and 0.25-1.25μl TE buffer, thermocycled for 14 cycles at 95°C for 15 sec. and 60°C for 4 min. and the resulting PCR reactions diluted 1:2 with TE buffer. Nucleofected colonies were pre-amplified with 5μl CellsDirect 2X reaction mix, 2.5μl 0.2X PPP, 1μl SuperScript III/Platinum Taq (Invitrogen), 1.5μl TE buffer and 0.1μl Superase-In (Invitrogen), thermocycled at 50°C for 15 min., 70°C for 2 min, and for 18 cycles at 95°C for 15 sec. and 60°C for 4 min. before similarly diluting. 2.25μl of pre-amplified cDNA was mixed with 2.5μl of TaqMan Universal PCR Master Mix (Applied Biosystems) and 0.25μl of Fluidigm sample loading reagent. An assay mix was prepared by adding 2.5μl 20X TaqMan gene expression assay to 2.5μl Fluidigm assay loading reagent. Gene expression analysis was performed using 96.96 Dynamic Arrays (DA) and the BioMark microfluidic system (Fluidigm Corp.). Samples were analyzed in triplicate and their geometric mean calculated for normalization to the 2 or 3 most stable housekeeping genes using the qBasePlus 1.3 analysis software (http://www.biogazelle.com) as previously described [29] and graphed using Gene-E (http:// www.broadinstitute.org/cancer/software/GENE-E/).

Human blastocyst thawing and culture
Human blastocysts were thawed using Quinn's Advantage Thaw Kit (CooperSurgical, Trumbull, CT) as previously described [30]. In brief, cryostraws were removed from the liquid nitrogen and exposed to air before incubating in a 37°C water bath. Once thawed, the blastocysts were transferred to a 0.5-mol/L sucrose solution for 10 minutes followed by a 0.2mol/L sucrose solution for an additional 10 minutes. The blastocysts were then washed in Quinn's Advantage Medium with Hepes (CooperSurgical) plus 5% Serum Protein Substitute (SPS; CooperSurgical) and transferred to a 20μl microdrop of Quinn's Advantage Blastocyst Medium (CooperSurgical) with 10% SPS under mineral oil (Sigma-Aldrich) for culturing at 37°C with 6% CO 2 , 5% O 2 and 89% N 2 for 2 hours prior to fixation.

Immunostaining
The zona pellucida (ZP) was removed from human blastocysts by treatment with Acidified Tyrode's Solution (Millipore) and ZP-free embryos were washed three times in Quinn's Advantage Medium with Hepes plus 5% SPS before fixing in 4% paraforamaldehyde (USB Corp., Cleveland, OH) for 20 minutes. Undifferentiated and differentiated hiPSCs were plated on glass 4-well chamber slides (Lab-Tek; Hatfield, PA) and similarly fixed. Following fixation, the blastocysts, undifferentiated/differentiated hiPSCs and nucleofected clones were washed three times in Phosphate Buffered Solution (PBS; Invitrogen) with 0. (1:200) at RT for 1 hour. Primary signals were detected using the appropriate 488, 568 or 647-conjugated donkey Alexa Fluor secondary antibody (Invitrogen) at a 1:250 dilution at RT for 1 hour in the dark. The blastocysts, undifferentiated/differentiated hiPSCs and nucleofected clones were washed three times with PBS-T between each primary and secondary antibody incubation as well as prior to the next primary antibody incubation. Immunofluorescence was visualized by sequential imaging, whereby the channel track was switched each frame to avoid cross-contamination between channels, using a Zeiss LSM700 URGB Zen confocal microscope (Thornwood, NY). The instrument settings, including the laser power, pinhole and gain, were kept constant for each channel to facilitate semiquantitiative comparisons between blastocysts, undifferentiated/differentiated hiPSCs and nucleofected clones.

Gene expression profiling of epigenetic reprogramming factors in human fibroblasts, iPSCs and ESCs
We first evaluated the expression of both the DNA methyltransferase (DNMT) family and the different classes of histone-modifying enzymes at the mRNA level in three different hESC lines (H9, HSF8 and HSF10), two independently generated hiPSC lines (C2 and C3), and the original adult human dermal fibroblasts (AHDF) at day 0, 7, 14 and/or 21 of spontaneous differentiation by microfluidic Q-PCR ( Figure 1A). Our results suggest that the de novo methyltransferase, DNMT3B, is highly expressed in both undifferentiated hESCs and hiPSCs, whereas its expression is almost undetectable in non-reprogrammed AHDFs and decreases as hESC and hiPSC differentiation proceeds (Figure 2A, 2B and Figure S1). Similarly, when we examined the expression of the different histone-modifying enzyme classes in AHDFs, hESCs, and the hiPSC clones we observed elevated expression of AURKB and PRMT5, enzymes that mediate the phosphorylation of serine residues and methylation of arginine residues, respectively, and their expression decreased upon differentiation ( Figure 2C, 2D and Figure S2). In contrast, high expression of the histone lysine methylation enzyme, SETD7, was detected in AHDFs, but not in undifferentiated hESCs and hiPSCS and then increased with differentiation between days 7, 14 and/or 21 ( Figure 2C, 2D and Figure S2). This suggested that the induction of DNMT3B, AURKB and PRMT5 expression in conjunction with the silencing of SETD7 expression might assist in the reprogramming of fibroblasts to a more hESC-like state.

Analysis of candidate epigenetic reprogramming factors at the protein level in human blastocysts
Based on the data obtained with hESCs and hiPSCs here and our previous study of potential reprogramming candidates in human embryos and hESCs at the mRNA level (Chavez et al., unpublished data), we next examined the expression and localization of our candidate reprogramming factors at the protein level in human embryos ( Figure 1B and Figure 3). For this purpose, human blastocysts were incubated with antibodies directed to DNMT3B, AURKB, PRMT5 and/or SETD7, stained with DAPI and analyzed by confocal fluorescent microscopy. As Figure 3A and 3B demonstrate, expression of DNMT3B, AURKB and PRMT5 was observed in human blastocysts and their expression primarily localized to the ICM of the embryo. In contrast, SETD7 expression was reduced in human blastocysts and was only detected within the trophoectoderm layer of the embryo ( Figure 3A and 3B), confirming the mRNA expression profiling of these factors in human embryos, hESCs and iPSCs.

Assessment of histone modifications mediated by candidate epigenetic factors in hiPSCs
We next sought to further confirm our results at the protein level by immunostaining hESCs and hiPSCs with certain histone modifications that AURKB, PRMT5 and SETD7 are known to mediate ( Figure 1B and Figure 4). While both Histone H3-serine 28 phosphorylation (H3-S28P) and Histone H4arginine 3 dimethylation (H4-R3me2) were highly expressed in undifferentiated hiPSCs, their expression decreased upon differentiation ( Figure 4A and 4B). However, there was a small population of differentiating hiPSCs that still expressed H3-S28P and since this epigenetic mark is associated with mitosis [31], this suggested that these cells are proliferative ( Figure 4B   Germ Cell-Like State in Partly Reprogrammed Clones PLOS ONE | www.plosone.org were detected in undifferentiated hiPSCs (indicated by white arrow) and its expression increased or was at least retained following differentiation in accordance with previous findings [32] to validate our findings at the mRNA level ( Figure 4A and 4B).

DNMT3B-GFP and SETD7-MO transfected colonies exhibit characteristics of partial reprogramming
Our results thus far suggested that the induction of DNMT3B, PRMT5 and AURKB expression and silencing of SETD7 expression in human fibroblasts might enhance the reprogramming of fibroblasts into iPSCs. To test this theory, we first nucleofected adult human fibroblasts (HUF1 and HUF5) with a plasmid encoding DNMT3B and/or a SETD7 morpholino (SETD7-MO; Figure 1C) given that morpholinos can be easily transfected into cells using nucleofection technologies [33]. For visualization and to assess transfection efficiency, the DNMT3B overexpression vector was GFP-tagged and the SETD7 morpholino 3'-carboxyfluroescein (3'-COF) labeled. As shown in Figure 5A, the presence of GFP in the nucleus and 3'COF in the cytoplasm was detected in some cells transfected with the DNMT3B plasmid or SETD7-MO 2-3 days posttransfection, respectively, indicating that nucleofection was sufficient for introducing these factors into fibroblasts. However, fewer colonies appeared when the cells were transfected with only one factor and the colonies obtained by nucleofection with DNMT3B-GFP appeared more hESC-like over those nucleofected with the SETD7-MO alone (Figure 5B). When fibroblasts were nucleofected with both DNMT3B-GFP and SETD-MO, the formation of several small clones was observed, which proliferated until day 10 ( Figure 5C) and then underwent senescence after approximately 14 days postnucleofection (Figure S3A; left). Therefore, we performed a second round of nuclefoection with DNMT3B-GFP and SETD7- MO between 2 to 7 days as previously described [34] and determined that nucloefecting cells on Day 1 and 3 was the most efficient in terms of reprogramming efficiency and number of clones (Table S1 and Figure S3B). Nucleofecting fibroblasts with DNMT3B-GFP and SETD7-MO more than two times, however, did not yield better results due to cellular detachment for nucleofection and massive cell death even in the presence of conditioned media on matrigel or mouse embryonic fibroblasts (MEFs) used as feeders (Figure S3A; right). Despite minimal proliferation, at least some of the cells in the colonies obtained by nucleofection with DNMT3B-GFP and SETD7-MO were positive for pluripotency markers such as Stage Specific Embryonic Antigen-3 (SSEA-3) and the keratin sulfate-related antigen, TRA-1-60 ( Figure 5D).

Gene expression analysis of DNMT3B-GFP and SETD7-MO transfected clones
Our next objective was to perform gene expression analysis of the clones we obtained by nucleofecting fibroblasts with DNMT3B-GFP and/or SETD7-MO by microfluidic Q-PCR ( Figure 1C). For comparison, we also included undifferentiated iPSCs derived by similarly introducing the four Yamanaka factors by nucleofection into the adult human fibroblasts and spontaneously differentiated on day 7 and 14 in the analysis. As Figure 5E demonstrates, several clones expressed high levels of DNMT3B and this expression occasionally persisted even after passaging. The expression of other DNMTs such as DNMT3A2 was also observed in some of the clones and correlated with the induction of DNMT3B, however, DNMT3L expression was not detected in any of the clones (Figure 5E). Unlike DNMT3B, the expression of SETD7 was low or absent (green), AURKB (orange) and SETD7 (red), the nuclear DNA stained with DAPI (blue) and visualized by multi-channel confocal microscopy. Differential Interference Contrast (DIC) was used to distinguish the inner cell mass (ICM) from the trophoectoderm of each embryo. Note that the expression of DNMT3B, AURKB and PRMT5 primarily localized to the ICM of human blastocysts, whereas SETD7 expression was predominantly detected in the outer trophoectoderm layer of the embryos (indicated by white arrows). in SETD7-MO-nucleofected clones, but its expression returned upon passaging (Table S2). Analogous to DNMT3B alone, however, the combination of DNMT3B over expression and SETD7 silencing had effects on the expression of other   Table S2.
In support of the SSEA-3 and TRA-1-60 partial immunostaining data (Figure 5D), we observed expression of the major pluripotency expressed gene, OCT4, at levels comparable to undifferentiated iPSCs generated with the four Yamanaka factors in several of the DNMT3B-GFP and/or SETD7-MO transfected clones ( Figure 5E). However, while certain clones also expressed additional pluripotency markers such as SOX2 and hTERT, NANOG was not expressed in the majority of clones ( Table S2), suggesting that introduction of DNMT3B-GFP and/or SETD7-MO induced partially reprogramming of adult human fibroblasts. An examination of the trophoblast marker, CDX2, and the three germ layer genes, GATA4, KDR and NCAM1 did not indicate that the colonies had formed other obvious cell types representative of extraembryonic or germ layer tissues ( Figure 5E). Notably, when we analyzed the expression of germ cell-specific genes, however, we observed high levels of early to mid germ cell markers, including Pumilio Homolog 1 (PUM1), PUM2 and STELLAR (DPPA3) in the clones ( Figure 5E and Table S2). Plating the nucleofected cells on STO feeders and culturing the colonies in Primordial Germ Cell (PGC) media containing LIF and FGF, however, did not change their morphology (data not shown). Nevertheless, a comparison of gene expression between the clones and hiPSCs differentiated in the presence of BMPs, demonstrated similar expression levels of early and mid germ cell-specific genes, but also high expression of late germ cell markers such as VASA (Mvh) and Synaptonemal Complex Protein 3 (SCP3) only in the BMP-differentiated hiPSCs. Since the addition of BMPs is generally used for more directed germ cell differentiation of pluripotent stem cell cultures, this suggested that the partially-reprogrammed clones exhibited a gene expression profile reminiscent of early stage hiPSC-derived germ cells.

Addition of AURKB, PRMT5, VPA and/or AZA for increasing reprogramming efficiency
Based on our findings that the colonies we obtained by DNMT3B-GFP and SETD7-MO transfection appeared to be only partially reprogrammed and exhibited characteristics of germ cells, we next sought to include additional candidate epigenetic factors and particular chemical compounds for increasing the reprogramming efficiency ( Figure 6A). To accomplish this, we used the neonatal human foreskin fibroblast line, HFF-1, as well as adult human fibroblasts since fibroblasts isolated from earlier developmental stages can be more easily reprogrammed [35]. In addition, we nucleofected HFF-1 and HUF5 fibroblasts with different combinations of DNMT3B-GFP, SETD7-MO as well as expression vectors for AURKB and PRMT5 (Figure 6B and Figure S3C), the other histone-modifying enzymes also shown to be highly expressed in undifferentiated hESCs and hiPSCs that decreased upon differentiation (Figure 2C, 2D and Figure S2). Finally, the HFF-1 cells were pre-treated with AZA for three days prior to transfection, nucleofected and then incubated with or without VPA (Figure 6A) based on the assumption that these cell permeable compounds help to open up the chromatin for gene activation [36]. While pre-incubation of the neonatal fibroblasts with AZA induced cell death (Figure S4A), the addition of VPA to the culture media significantly increased colony formation (Table S1 and Figure S4B). Moreover, the induction of AURKB and PRMT5 expression together with DNMT3B over expression and SETD7 silencing also resulted in reduced colony formation and cell death, which may be due to the large amount of DNA needed to transfect the HUF1 and HFF-1 fibroblasts with all four factors and (Figure 6B, Table S1 and Figure S3C). Analogous to the results observed with the adult human fibroblasts, some of the largest number of clones was obtained when the human neonatal fibroblasts were nucleofected with both DNMT3B-GFP and SETD7-MO in the absence or presence of AZA and/or VPA (Table S1 and Figure  S4A and S4B).
Analysis of gene expression in the resulting clones from HFF-1 fibroblasts revealed similar effects of DNMT3B-GFP and SETD7-MO nucleofection on the expression of DNMTs and histone-modifying enzymes, including the lack of DNMT3L expression ( Figure 6C) as in adult dermal fibroblasts. In addition, the majority of colonies also expressed relatively high levels of OCT4 as well as other pluripotency markers such as SALL4 and LIN28 to suggest that the cells were at least partially pluripotent. Unlike the clones observed with adult fibroblasts (Figure 5E), however, NANOG was expressed in almost all of the colonies obtained from neonatal fibroblasts ( Figure 6C). Moreover, while trophoblast genes such as CDX2 and GCM1 were not expressed in any of the clones, markers of the three germ layers, including GATA4, KDR, NCAM1, PAX6 and PDGDR-α, were either not expressed or expressed at relatively low levels ( Figure 6D). Notably, evaluation of the same germ cell genes in neonatal and adult fibroblasts or additional germ cell markers in HFF-1 cells suggested that the colonies appeared to be germ cell-like, unlike conventionally generated colonies (Figure 5E and Figure 6D). In particular, early germ cell-specific genes such as Chemokine (C-X-C Motif) Receptor 4 (CXCR4), IFITM3 (Fragilis), PR domain zinc finger protein 1 (PRDM1;BLIMP1), c-KIT as well as PUM1, PUM2 and STELLAR were highly expressed in several HFF-1 clones. Analogous to the BMP differentiated hiPSCs shown in Figure 5E, some of the HFF-1 colonies also expressed high levels of VASA to suggest that the neonatatal cells are indeed more easily reprogrammable (Figure 6D). Later germ cell, meiosis-related or sex-specific markers, including STRA8, GDF3, ACROSIN, DMC1, ACROSIN, BOULE and TEKT1, on the other hand, were not expressed or exhibited negligible expression in neonatal colonies ( Figure 6D).

Increasing reprogramming efficiency by the addition of NANOG and/or SV40 and hTERT
Previous studies have shown that NANOG is one of the most important pluripotency factors and an indication of complete reprogramming to pluripotency [37]. Given that we did not observe high levels of NANOG expression in the majority of clones we obtained with DNMT3B-GFP and SETD7-MO, we included NANOG with SV40 and hTERT, which have shown to increase reprogramming efficiency due to their immortalization properties [16,17], to the reprogramming factor cocktail ( Figure  6A). After 5 to 7 days post-nucleofection, several small clones formed (Table S1 and Figure S4C and S4D) and expressed high levels of almost all pluripotency genes tested, but did not proliferate ( Figure 6D). Nucleofecting neonatal fibroblasts more than two times did not assist in generating fully-reprogrammed clones as significant cell death occurred upon cellular detachment. In order to circumvent this, the HFF-1 cells were nucleofected with SETD7-MO and then transfected with DNMT3B, SV40 and hTERT using a cationic lipid reagent ( Figure 6A and Table S1). Although the numerous clones we observed early in the reprogramming process by this method did not express pro-apoptotic genes such as PDCD5, after approximately two weeks, these colonies began to undergo apoptosis (data not shown). Nevertheless, early germ cell markers were highly expressed in colonies obtained by the addition of NANOG, SV40 and/or hTERT, suggesting that the partially-reprogrammed clones exhibit a germ cell-like identity regardless of the type of transfection method used ( Figure 6D). In addition, including NANOG in the reprogramming cocktail with or without SV40 and hTERT also significantly changed the morphology of the colonies to a more embryonic germ (EG) cell-like appearance, namely a more rounded shape and less flattened edges (Figure S4C and S4D).

Discussion
Since the advent of hiPSCs a few years ago [3,8], several studies have focused on identifying and utilizing novel factors, cell-permeable compounds and/or different transfection approaches for increasing the efficiency of reprogramming human fibroblasts to pluripotency [38]. Here, we overexpressed the epigenetic factors, DNMT3B, AURKB and PRMT5 and silenced SETD7 expression based on their expression profile in human embryos, fibroblasts and undifferentiated/differentiated hESCs and conventionally generated hiPSCs, to reprogram neonatal and adult fibroblasts into hiPSCs. In conjunction with DNMT3B, AURKB and PRMT5 overexpression and SETD7 silencing, we also treated fibroblasts with the small molecule compounds, VPA and AZA, and included additional factors such as NANOG, SV40 and/or hTERT for reprogramming. From these experiments, we did not obtain fully-reprogrammed clones using these different factors and various methods, however, we observed that cells acquired a germ cell-like identity that was associated with early reprogramming.
After determining which DNMTs and histone-modifying enzymes to overexpress or silence, we transfected fibroblasts by nucleofection since morpholinos can be easily introduced into cells by this method [33] and to avoid the genomic integration associated with viral technologies [9]. Our data demonstrates that the introduction of two factors, DNMT3B-GFP and SETD7-MO, in both neonatal and adult fibroblasts was the most efficient in terms of colony number and proliferation. The decreased colony formation observed with the induction of AURKB, PRMT5, NANOG, SV40 and/or hTERT expression may be due to the excess amount of DNA required for transfection or the association of some of these factors with cell death [17]. Moreover, the number of clones obtained by first treating fibroblasts with AZA for three days prior to nucleofection in media containing VPA was much greater than with AZA pre-treatment alone, which may be explained by previous findings that AZA can induce apoptosis in mammalian cells [39].
Regardless of the approach, the colonies that formed following transfection appeared pluripotent based on expression of pluripotency-related cell surface antigens and pluripotency genes at the protein and mRNA level, respectively. In addition, while certain clones exhibited cell proliferation early in the reprogramming process, this proliferation ceased after two weeks and the cells began to undergo apoptosis. This may help explain why only certain cells, rather than all or at least the majority of cells as in typical hESC and iPSC colonies, were positive for pluripotency-related cell surface antigens. More importantly, we observed high levels of several early to mid germ cell markers in colonies obtained by different reprogramming methods from both neonatal and adult fibroblasts that were collected prior to this two-week period. Notably, the germ cell-specific gene, PRDM1/ BLIMP-1, was previously shown to be a target of LIN28, one of the four factors that Thomson and colleagues originally introduced for reprogramming somatic cells to pluripotency [40], via inhibition of the let-7 microRNA [41]. Although LIN28 is not essential for reprogramming process, it does increase reprogramming efficiency [40], and given recent findings of inverse correlation between let-7 and Prdm1 expression in mouse iPSCs [42], suggests that the induction of germ cellspecific genes is part of the reprogramming process. Based on the assumption that partially-reprogrammed iPSCs represent intermediate stages of reprogramming [43,44] and given that naturally developing germ cells are also pluripotent and proliferate only at distinct times during development [45], perhaps a germ cell-like state is associated with the transition of somatic cells to reprogrammed iPSCs as previously described [42,46,47]. In support of this, we observed high to moderate levels of LIN28 and PRDM1 expression in almost all of the clones we obtained from the neonatal fibroblasts. Finally, the addition of NANOG, which has been show to be dispensible for somatic pluripotency, but required for germ cell formation [48], to the reprogramming cocktail changed the colony morphology to a more EG cell-like appearance. Thus, our findings may provide an additional means to evaluate human germ cell differentiation in vitro, particularly in the context of pluripotent stem cell-derived germ cell development, and assist in our understanding of complete epigenetic reprogramming. Additional research should be aimed at characterizing the cell fate of the germ-like cells in vitro and in vivo using a xeno-transplant model.  )   Table S1.

Supporting Information
The reprogramming efficiency of each transfection approach. A table displaying the cell type, transfection method, reprogramming factors and treatment conditions used for each transfection approach in this study.
(DOCX )   Table S2. Comparison of gene expression ratios between cell types. A table comparing global gene expression levels of pluripotency factors, the candidate reprogramming factors (DNMT3B and/or SETD7) and germ cell markers in the original HUF5 adult dermal human fibroblasts, following transfection with DNMT3B and/or SETD7-MO with passage (P) and clone (C) numbers, and conventionally generated induced pluripotent stem cells (iPS) on Day 0, 7 and 14 of differentiation with and without Bone Morphogenetic Proteins (BMPs). (DOCX)