Introduction
This study was initiated to examine whether the family proteins of the four transcription factors could generate iPS cells. Mouse embryonic fibroblasts (MEFs) containing a green fluorescent protein (GFP)-IRES-Puror transgene driven by the Nanog gene regulatory elements were used3. Nanog is specifically expressed in mouse ES cells and preimplantation embryos5, 6 and can serve as a selection marker during iPS cell induction. Introduction of the aforementioned four factors reprograms MEFs to GFP-expressing colonies of iPS cells. Nanog-selected iPS cells are indistinguishable from ES cells in morphology, proliferation and gene expression, and give rise to germline-competent chimeras2, 3, 4.
Oct3/4 belongs to the octamer-binding (Oct) family of transcription factors, which contain the POU domain7. The closest homologs of Oct3/4 are Oct1 and Oct6. We introduced Oct3/4, Oct1 or Oct6, together with the remaining three factors, into the Nanog-reporter MEFs by retroviral delivery. With Oct3/4, many GFP+ colonies were observed (Fig. 1a). In contrast, no GFP+ colonies were obtained with Oct1 or Oct6, thus indicating the inability of these two homologs to induce iPS cells. Sox2 is one of the SRY-related HMG-box (Sox) transcription factors, characterized by the presence of the high-mobility group (HMG) domain8. We tested Sox1, Sox3, Sox7, Sox15, Sox17 and Sox18. GFP+ colonies were obtained with Sox1, and fewer GFP+ colonies were obtained with Sox3, Sox15 and Sox18 (Fig. 1a). Sox18, however, failed to expand the cells. Klf4 belongs to the family of Krüppel-like factors (Klfs), zinc-finger proteins that contain amino acid sequences similar to those of the Drosophila embryonic pattern regulator Krüppel9. Klf1, Klf2 and Klf5 were tested, and GFP-expressing colonies were obtained with Klf2 (Fig. 1a). Klf1 and Klf5 also generated iPS cells, but with a lower efficiency. c-Myc has two related proteins: N-Myc and L-Myc10. GFP+ colonies emerged with both N-Myc and L-Myc (Fig. 1a). Therefore, some, but not all, family proteins of the four factors can reprogram MEFs to iPS cells.
Figure 1: Effect of family factors and omission of Myc on the generation of iPS cells from Nanog-reporter MEFs.
(a) Generation of iPS cells with family genes from MEFs by Nanog selection. The number of GFP+ colonies is shown. The results of three independent experiments are shown with different colors (white, gray and black). "4 factors" indicates the combination of Oct3/4, Sox2, Klf4 and c-Myc. (b) The effect of puromycin selection timing on iPS cell generation. Shown are GFP+ colonies observed 28 d after transduction of the four factors or the three factors devoid of Myc. (c) The effect of puromycin selection timing on the percentage of GFP+ colonies per all colonies.
Full size image (55 KB)We also tested the family proteins for their ability to reprogram MEFs in which 
geo (a fusion between -galactosidase and the neomycin-resistance gene) was knocked into the Fbxo15 locus11. Similar results to those with the Nanog-based selection were obtained: Sox2 could be replaced by Sox1 and Sox3, Klf4 by Klf2, and c-Myc by N-Myc and L-Myc. The cells generated by the family proteins were expandable and showed a morphology indistinguishable from that of ES cells (not shown). They gave rise to teratomas in nude mice (see Supplementary Fig. 1 online). Therefore, some family proteins are capable of inducing iPS cells from both Nanog-reporter MEFs and Fbxo15-reporter MEFs.
Unexpectedly, a few ES cell–like and GFP+ colonies from Nanog-reporter MEFs were obtained without the Myc retrovirus (Fig. 1a). This was in contrast to a previous study in which no GFP+ colonies could be obtained without Myc3. One difference between the two studies is the timing of the drug selection. In the earlier study, puromycin selection was initiated 7 d after the transduction, whereas in the present experiment the selection was started 14 d after. This suggested that iPS cell generation without Myc is slower than with Myc.
To test this possibility, we transduced Nanog-reporter MEFs either with the four factors or with three factors devoid of Myc, and started puromycin selection 7, 14 or 21 d after transduction (Fig. 1b). With the four factors, GFP+ colonies were observed in all of the conditions. The colony numbers substantially increased when puromycin selection was delayed. Without Myc, no GFP+ colonies were observed when selection was initiated 7 d after transduction. In contrast, GFP+ colonies did emerge even without Myc when selection was started 14 or 21 d after transduction. In each condition there were fewer colonies with the three factors than with the four factors. Nanog-selected iPS cells generated without the Myc retrovirus expressed ES-cell marker genes at similar levels to those in ES cells (Supplementary Fig. 2 online), and gave rise to adult chimeras when transplanted into blastocysts (Supplementary Table 1 online). Another difference is that fewer GFP- colonies and fewer background cells were observed with the three factors devoid of Myc than with the four factors (Fig. 1c). Therefore, the omission of Myc resulted in a less efficient but more specific induction of MEFs to iPS cells.
In the present study, it was also possible to generate a few iPS cells without Myc from MEFs in which geo was knocked into the Fbxo15 locus11 (Fig. 2a). This is again in contrast to the original report, in which no iPS cells were obtained without Myc1. In the two experiments, G418 selection was initiated with the same timing: 3 d after the transduction. However, the colonies appeared 14–21 d after transduction of the four factors, whereas 
30 d were required without the Myc retrovirus. Another difference was that in the present study the retroviral transfection efficiency was increased by preparing each of the four or three factors separately in an independent Plat-E12 plate. This yielded a significant increase in the number of iPS cell colonies compared with the original work, in which all four factors were prepared in a single Plat-E plate. This result is consistent with the notion that generation of iPS cells without Myc is a slower and less efficient process than with Myc.
Figure 2: Generation of iPS cells without the Myc retrovirus from MEFs containing the Fbxo15 reporter and the constitutively active GFP transgene.
(a) Morphology of iPS cells generated without the Myc retrovirus, using phase contrast microscopy. The bar indicates 500 
m. (b) RT-PCR analyses of ES-cell marker genes in ES, MEF and iPS cells generated without Myc. (c) Chimeras derived from iPS cells induced without Myc (clones 142B-6 and 142B-12). (d) The incidence of tumor-associated death in chimeras derived from iPS cells, which were generated with or without Myc.
Fbxo15-selected iPS cells generated with the four factors express lower levels of ES-cell marker genes than do ES cells1. They cannot produce adult chimeras when microinjected into blastocysts. In contrast, iPS cells generated without Myc expressed ES-cell marker genes at similar levels to those in ES cells even with the Fbxo15 selection (Fig. 2b). Furthermore, adult chimeras with a high contribution from these iPS cells were obtained (Fig. 2c and Supplementary Table 1).
We examined tumorigenicity in these mice derived from the Myc-
iPS cells, which were selected for either Nanog or Fbxo15 (Supplementary Table 1). We found that six of 37 chimeras derived from iPS cells with the four factors died of tumors within 100 d after birth. By contrast, all 26 chimeras generated without Myc survived this period (Fig. 2d). Thus, the omission of the Myc retrovirus significantly reduced the risk of tumorigenicity in chimeras (P < 0.05 by 
2 test). Future study is required to determine whether these mice develop tumors later in life.
Next, we examined whether the omission of Myc would result in efficient isolation of iPS cells without drug selection. The four or three factors were introduced into adult tail tip fibroblasts (TTFs) containing the Nanog reporter, but puromycin selection was not applied. To visualize transduced cells, we used DsRed retrovirus together with the four or three factors. Thirty days after the retroviral transduction, the dishes with cells transduced with the four factors were covered with numerous GFP- colonies and background cells (Fig. 3a and Supplementary Table 2 online). Using fluorescence microscopy, we found small portions of these colonies (4, 132 and 424 colonies in three independent experiments) to be GFP+. Of note, the GFP+ colonies were negative for DsRed, which was consistent with the retroviral silencing observed in Nanog-selected iPS cells3. In contrast, with the three factors devoid of Myc, a small number (7, 21 and 43 in three independent experiments) of discrete colonies was observed with few background cells. Approximately half of them expressed GFP in a patchy manner. DsRed was detected in only a small portion of some colonies, indicating that it was largely silenced. No overlap was observed between GFP and DsRed expression. Most of these colonies were expandable and produced iPS cells, which became positive for GFP and negative for DsRed at passage 2. Therefore, the omission of Myc resulted in a more specific generation of iPS cells, in which Nanog-GFP is activated and the retroviruses are silenced.
Figure 3: The efficient isolation of iPS cells without drug selection.
(a) Morphology of iPS cells induced from adult TTFs containing the Nanog-GFP-IRES-Puror reporter. Cells were transduced with either the four factors or the three factors devoid of Myc, together with DsRed, and then were cultured for 30 d without drug selection. Expression of the Nanog reporter (Nanog-GFP) and the DsRed retrovirus (Tg-DsRed) was examined by fluorescence microscopy. The bar indicates 500 m. (b) Morphology of iPS cells induced from adult TTFs, which contained a DsRed transgene driven by a constitutively active promoter (ACTB, -actin gene) but lacked the Nanog- or Fbxo15- selection cassettes. The cells were transduced with either the four factors or the three factors devoid of Myc, together with GFP, and then cultured for 30 d without drug selection. The expression of the GFP retrovirus (Tg-GFP) was examined by fluorescence microscopy. The bar indicates 500 m. (c) RT-PCR analyses of ES-cell marker genes in ES cells and in iPS cells generated from TTFs without drug selection (8 clones with four factors and 12 clones with three factors). (d) Chimeras derived from iPS cells, which were generated from adult TTFs without drug selection or the Myc retrovirus.
We next attempted to generate iPS cells from adult TTFs that did not have selection markers but had the DsRed transgene driven by a constitutively active promoter13. The four factors or the three factors devoid of Myc were introduced. In addition, a GFP retrovirus was introduced to monitor silencing. After 30 d without drug selection, 1,000 colonies emerged from 0.5 
105 cells transduced with the four factors. Most of them were positive for GFP, indicating that retroviral silencing did not take place in these cells. In contrast, only 16 colonies (Fig. 3b) emerged from 3.5 105 cells transduced with the three factors devoid of Myc. Most of these colonies expressed no GFP, and the remaining colonies expressed GFP in small areas. All of these colonies were expandable and showed ES cell–like morphology at the second passage. They were all negative for GFP, thus indicating retroviral silencing. RT-PCR showed that these cells expressed ES-cell marker genes at comparable levels to those in ES cells (Fig. 3c). In addition, RT-PCR confirmed the retroviral silencing of Klf4 and the absence of the Myc transgene in iPS cells generated with the three factors. Furthermore, when transplanted into blastocysts, these cells gave rise to chimeras (Fig. 3d and Supplementary Table 1). Therefore, by omitting MYC, high-quality iPS cells can be efficiently generated from adult TTFs without drug selection.
We have recently reported the generation of iPS cells from adult human dermal fibroblasts (hDFs) by the same four factors14. We therefore examined whether human iPS cells can be generated without the MYC retrovirus. As we described, from 5 104 hDFs transduced with the four factors, we obtained 10 human (h)ES cell–like colonies and 100 non-hES cell–like colonies (Fig. 4a). Without MYC, we obtained no colonies at all. From 5 105 hDFs transduced with the four factors, we obtained >500 non-hES cell–like colonies. We occasionally observed a few hES cell–like colonies, but it was difficult to isolate these cells because of the high number of background cells. Without MYC, we obtained 0–5 hES cell–like colonies with a few background cells. The hES cell–like colonies without MYC were expandable and were positive for hES-cell markers, including SSEA-3, SSEA-4, TRA-1-60, TRA-1-81, TRA-2-49/6E and NANOG (Fig. 4b). RT-PCR showed that the cells expressed hES-cell marker genes such as the endogenous OCT3/4, the endogenous SOX2, NANOG and REX1 at similar levels to those in hES cell line H9 and human embryonic carcinoma cell line NTERA-2 (Fig. 4c). Furthermore, these cells differentiated into various cells that were positive for 
-smooth muscle actin (SMA), III-tubulin and -fetoprotein (AFP) (Fig. 4d). Thus, as is the case with mouse iPS cells, the omission of MYC resulted in a more specific induction of adult human fibroblasts to iPS cells, but with lower efficiency.
Figure 4: Generation of human iPS cells without the MYC retrovirus.
(a) The number of hES cell–like colonies and total colonies from either 5 104 or 5 105 hDFs, which had been transduced either with the four factors or with the three factors without MYC. (b) The morphology and immunostaining of hES-cell markers in human iPS cells generated without MYC (clone 253G1). Bars, 100 m. (c) The expression of ES-cell marker genes in human iPS cells derived from hDFs without the MYC retrovirus (253G) or with c-MYC (253F). (d) Embryoid body–mediated differentiation of human iPS cells generated without the MYC retrovirus (clone 253G1). Bars, 100 m.
Our study does not argue that Myc is dispensable for iPS cell generation. We found that MEFs expressed c-Myc from the endogenous gene at 20% of the levels observed in mouse ES cells (Supplementary Fig. 3 online). This expression continues in iPS cells. Thus, Oct3/4, Sox2 and Klf4 may recruit endogenous Myc proteins to induce reprogramming. The endogenous expression levels may increase during the course of reprogramming. Recently, another group reported the generation of human iPS cells from neonate fibroblasts by a different combination of OCT3/4, SOX2, NANOG and LIN28 (ref. 15). Whether the same four factors work with adult human fibroblasts remains to be determined. In addition, LIN28 may activate endogenous Myc proteins by enhancing translation of insulin-like growth factor-2 (ref. 16). Further studies are required to determine the roles of Myc and the other factors in the reprogramming process.
In summary, we have demonstrated that iPS cells can be generated from mouse and human fibroblasts without the Myc retrovirus. Mice derived from iPS cells that had not been transduced with the Myc retrovirus showed a significantly reduced incidence of tumorigen-icity compared with those derived from cells transduced with the four factors. Without the Myc retrovirus, however, the efficiency of iPS cell generation decreased substantially. In approximately half the experiments, we were not able to obtain human iPS cells when transduction with the Myc retrovirus was omitted. It will be important to find factors or small molecules that can enhance the efficiency of iPS cell generation without Myc for the generation of disease- and patient-specific iPS cells.
Methods
Plasmid constructions.
The coding regions of family genes were amplified by RT-PCR with the primers listed in Supplementary Table 3 online, subcloned into pDONR201 or pENTR-D-TOPO (Invitrogen), and recombined with pMXs-gw by the LR reaction (Invitrogen).
Induction of iPS cells with drug selection.
The induction of iPS cells was performed as previously described1, 3 with some modifications. Briefly, MEFs, which contained either the Nanog-GFP-IRES-Puror reporter or the Fbxo15-geo reporter, or both, were seeded at 1.3 and 8.0 105 cells/well in 6-well plates and 100-mm dish, respectively, with SNL leukemia inhibitory factor feeder cells17. The transduced cells were cultivated with ES medium containing leukemia inhibitory factor18. Selection with G418 (300 g/ml) or puromycin (1.5 g/ml) was started as indicated. Twenty-five to 30 d after transduction, the number of colonies was recorded. Some colonies were then selected for expansion.
iPS cell generation without drug selection.
TTFs were isolated from adult Nanog-reporter mice or adult DsRed-transgenic mice13. Retrovirus-containing supernatant was prepared in the separated method. For the four-factor transduction, retrovirus-containing supernatants for Klf4, c-Myc, Oct3/4, Sox2 and DsRed, were mixed with the ratio of 1:1:1:1:4. When the fibroblasts were transduced with the three factors, retrovirus-containing supernatants for Klf4, Oct3/4, Sox2, Mock and DsRed were mixed with the ratio of 1:1:1:1:4. With DsRed transgenic mice, the GFP retrovirus was used instead of DsRed. For transfection, TTFs were seeded at 8.0 105 cells per 100-mm dish, which did not have feeder cells. TTFs were incubated in the virus/polybrene -containing supernatants for 24 h. Four days after transduction, TTFs transduced with the three factors were reseeded at 3.5 105 cells per 100-mm dish with SNL feeder cells and cultured with ES medium. TTFs transduced with the four factors were reseeded at 0.5 105 cells per 100-mm dish with feeder cells. Thirty to 40 d after transduction, the colonies were selected for expansion.
Characterization of iPS cells.
RT-PCR and teratoma formation were performed as previously described1, 3. For the chimera experiments, 15–20 iPS cells were injected into BDF1-derived blastocysts, which were then transplanted into the uteri of pseudo-pregnant mice.
Generation of human iPS cells.
DFs from the facial dermis of a 36-year-old Caucasian female were purchased from Cell Applications, Inc. Induction of iPS cells was performed as described14.
Note: Supplementary information is available on the Nature Biotechnology website.
Author contributions
M.N., M.K., K.Tanabe, K.Takahashi, T.A. and K.O. generated and characterized iPS cells. T.I. performed the chimera experiments. Y.M. prepared plasmids. N.T. characterized iPS cells. S.Y. supervised the study and wrote the manuscript.
