Adult human dermal fibroblasts were transduced with retroviruses expressing M3
O-SKM or OSKM. Fibroblasts were subcultured onto feeder cells three days later, when addition of bFGF or LIF was also started. Control culture was prepared without bFGF and LIF. ESC-like colonies were defined as iPSC colonies when they were double-positive for the pluripotency markers NANOG and TRA1-60 (). In the presence of bFGF, M3
O-SKM induced iPSC colonies over 20 times more efficiently than OSKM on day 10 as previously reported 
(). When bFGF was omitted or replaced with LIF, iPSC colonies were not observed with OSKM until day 16 at the earliest. However, iPSC colonies did appear around day 7 when M3
O-SKM was used without either cytokine. Addition of LIF did not increase the efficiency of iPSC colony formation. When NANOG and SSEA4 were used as makers for iPSC colonies, similar results were obtained ().
Expression of pluripotency markers by iPSCs prepared with LIF or bFGF.
We picked up 20–48 iPSC colonies from each culture condition between day 10 and 12 when colonies were available. These clones were then expanded on feeder cells in culture with the same cytokine to compare their efficiency of establishing pluripotent cell lines in the absence of signaling inhibitors. More than 90% of iPSC clones prepared with M3O-SKM in the presence of bFGF (F-iPSCs) remained undifferentiated for 120 days (25–30 passages) as monitored by the expression of TRA1-60 and NANOG. In the absence of bFGF and LIF, however, all iPSC colonies prepared with M3O-SKM morphologically differentiated by day 15 when the transgenes were suppressed although they remained undifferentiated if bFGF was added from day 15 onwards. In addition, most of the iPSC colonies prepared with M3O-SKM in the presence of LIF (L-iPSCs) differentiated around the same time. However, a few L-iPSC clones retained morphologically undifferentiated colonies scattered among numerous differentiated colonies beyond day 15. These apparently undifferentiated colonies were also unstable, with central undifferentiated parts being surrounded by differentiated derivatives (). The undifferentiated state of the central parts was verified by the expression of NANOG and SSEA4 (). The undifferentiated parts could be manually picked up during each subculture and expanded without substantial contamination with differentiated cells until day 200 at least.
Characterization of colony morphology, gene expression patterns, and sensitivity to an FGF receptor inhibitor in L-iPSCs obtained with M3O-SKM.
L-iPSCs established with this procedure were characterized with several approaches. First, suppression of the M3O-SKM transgenes was verified by comparing the levels of total mRNA (derived from endogenous genes and transgenes) encoding OCT4, SOX2, KLF4 and c-MYC and those derived from endogenous genes. This was done using PCR primers specific to a 3′end untranslated region and coding region of each gene. Quantitative RT-PCR indicated that the relative expression levels of the four genes fell within a 2-fold range as compared with the levels in ESCs for both total and endogenous genes, suggesting suppression of the transgenes (). Similar results were obtained with F-iPSCs. Suppression of the M3O transgene was also confirmed using a primer pair that spans the border between the M3 domain and OCT4 ().
To rule out the possibility that autocrine bFGF was compensating for the lack of exogenous bFGF in L-iPSC culture, we blocked the FGF receptor with SU5402, an inhibitor of the FGF receptor tyrosine kinase 
between day 8 and 10. The effectiveness of this treatment was verified by the observation that the number of F-iPSC colonies decreased to less than 40% of the control culture treated with the solvent dimethyl sulfoxide for both OSKM and M3
O-SKM (). However, SU5402 had no effect on the number of L-iPSC colonies, suggesting that bFGF did not play a predominant role in the formation of L-iPSC colonies. In addition, L-iPSCs cultured for 120 days expressed the pluripotency markers OCT4, NANOG, TRA1-60, TRA1-81 and alkaline phosphatase (). Expression levels for seven additional genes enriched in pluripotent cells were similar to those in ESCs as shown by quantitative RT-PCR (). Karyotype was normal in L-iPSCs (). L-iPSCs also formed teratomas containing differentiated tissues derived from all three germ layers following injection into immunocompromized mice, proving pluripotency of these cells (). L-iPSCs required the continuous presence of LIF for self-renewal and could not form ESC-like colonies after subculture without LIF.
Characterization of gene expression patterns, karyotype, and teratoma formation in L-iPSCs prepared with M3O-SKM.
To understand whether L-iPSCs belonged to the naïve or primed type, we compared them with F-iPSCs, which are primed pluripotent stem cells. First, L-iPSC colonies were flat and morphologically indistinguishable from F-iPSC colonies. Second, L-iPSCs did not form colonies after single-cell dissociation with trypsin during subculture, similar to F-iPSCs. Third, F-iPSCs and L-iPSCs expressed REX1
), STELLA (DPPA3
at similar levels (). Although FGF5
is known to be expressed in primed pluripotent stem cells as described above, its expression level was much lower than that in fibroblasts. Fourth, the status of X chromosome inactivation was investigated by staining L-iPSCs with an antibody against trimethylation of lysine 27 on histone H3 (H3K27me3), whose enrichment is commonly observed in inactive X chromosome. Generally, only one X chromosome is active in female human F-iPSCs 
. Almost all L-iPSCs showed one H3K27me3-positive area in each nucleus similar to the original fibroblasts, suggesting that one of the X chromosomes was inactivated in L-iPSCs (). On the basis of the high similarity between F-iPSCs and L-iPSCs, we concluded that L-iPSCs belonged to the primed type pluripotent stem cell.
Characterization of L-iPSCs by using gene expression analyses and immunofluorescence staining of H3K27me3.
The similarity between F-iPSCs and L-iPSCs was further tested by examining if LIF in the culture medium of L-iPSCs could be replaced with bFGF on day 90 without disrupting self-renewal. After culture for an additional 50 days, the colonies, now called LF-iPSCs, maintained an ESC-like morphology and expressed OCT4, NANOG, TRA1-60, TRA1-81 and alkaline phosphatase (). This cytokine switch diminished the tendency for differentiation; LF-iPSC colonies could be maintained in an undifferentiated state more easily than L-iPSCs. LF-iPSCs also preserved pluripotency as shown by teratoma formation following injection into immunocompromized mice (). In an opposite experiment, bFGF was replaced with LIF for F-iPSCs, but the cells did not form any ESC-like colonies after subculture. Thus, unlike F-iPSCs, the pluripotency of L-iPSCs could be sustained by both the LIF and bFGF signaling pathways.
Characterization of LF-iPSCs with immunofluorescence staining and a teratoma formation assay.
Finally, genome-wide transcription patterns were compared among ESCs, F-iPSCs, L-iPSCs, LF-iPSCs and parent fibroblasts with a microarray analysis. Overall gene expression patterns were quite similar among ESCs, two F-iPSC clones, two L-iPSC clones and two LF-iPSC clones (). Hierarchical cluster analysis of genome-wide expression profiles demonstrated that L-iPSCs, F-iPSCs, LF-iPSCs and ESCs were very closely clustered (). mRNAs encoding OCT4, SOX2, NANOG, KLF4, REX1, STELLA, FGF5 and T were all expressed at similar levels among these pluripotent stem cells, with any differences being less than 1.5-fold ().
Genome-wide gene expression patterns comparing various pluripotent stem cells.
Our study established L-iPSCs in the absence of sustained overexpression of transgenes and signaling inhibitors, unlike previously reported methods of making LIF-dependent human iPSCs. This was possible because of the usage of the transcriptionally augmented M3
O. Reflecting differences in methods, L-iPSCs were also phenotypically different from previously reported LIF-dependent iPSCs and more closely related to standard bFGF-dependent iPSCs. This result has three important implications for our understanding of the regulation of pluripotency and self-renewal in iPSCs. First, dependency on LIF or bFGF is not the primary determinant of naïve or primed pluripotency. The activity of transgenes and modulations to cytokine signaling systems appear to be more critical in determining the type of pluripotent stem cell. Second, human and mouse cells are fundamentally different in terms of their response to LIF because this cytokine induces different types of pluripotent stem cells: primed-type human iPSCs and naïve-type mouse iPSCs 
. Third, this study suggests that the diverse signaling pathways of LIF and bFGF are potentially converged at the level of OCT4-target genes. LIF regulates its target genes primarily through activation of the JAK kinase and resulting phosphorylation and DNA binding of STAT3 
. The main signaling pathway for bFGF is activation of MEK/ERK, leading to DNA binding of a variety of transcription factors 
. There are no well-established overlapping target genes between these two signaling systems. In future studies, a detailed comparison of the target genes of OCT4 and M3
O could potentially shed light on this novel mechanism underlying pluripotency and self-renewal.