We used the Nanog-GFP reporter system for sensitive and specific identification of iPS cells3
. When the three factors devoid of c-Myc were introduced into Nanog-GFP, p53-wild-type MEF, we obtained 11 ± 8 (n=4) GFP-positive colonies from 5000 transduced fibroblasts (). From Nanog-GFP, p53-heterozygous mutant MEF, we observed 58 ± 56 GFP-positive colonies. In contrast, from Nanog-GFP, p53-null fibroblasts, we obtained significantly more GFP-positive colonies (275 ± 181) than from wild-type MEF.
iPS cell generation from p53-null MEF by four or three reprogramming factors
By using a flow cytometer, we plated one Nanog-GFP cell (p53 wild-type, heterozygous mutant, or homozygous mutant), which was transduced with the three factors five days before the re-plating, into a well of 96-well plates. Twenty-three days after the re-plating, we observed GFP-positive colonies in few wells per a 96-well plate with p53 wild-type or heterozygous fibroblasts (). By contrast, we observed GFP-positive colonies in 7 ± 4 (n=4) wells per 96-well plate with p53-null fibroblasts. These data showed that loss of p53 function significantly increased the efficiency and that up to 10% of transduced cells can become iPS cells even without c-Myc.
We performed the same experiment with the four factors, including c-Myc. We observed GFP-positive colonies in zero or one well per a 96-well plate with p53 wild-type fibroblasts (). By contrast, we observed GFP-positive colonies in 6 ± 7 and 16 ± 10 (n=4) wells per a 96-well plate with p53-heterozyous and p53-null fibroblasts, respectively. These data showed that addition of the Myc retrovirus further increased the efficiency of iPS generation up to 20%.
We also tested the effect of a dominant negative p53 mutant P275S13
on the generation of iPS cells. When P275S was introduced into Nanog-GFP, p53-heterozygous MEF, we observed a substantial increase in the number of GFP-positive colonies (). In addition, we placed complementary DNAs (cDNA) encoding the wild-type p53 or transactivation-deficient mutants (D278N14
) into the pMXs retroviral vector16
and introduced it together with the retroviruses encoding Oct3/4, Sox2 and Klf4 into Nanog-GFP, p53-null MEFs. Wild-type p53 significantly decreased the number of GFP-positive colonies (). The transactivation-deficient p53 mutants, in contrast, did not show significant effects. These data confirmed that loss of p53 is responsible for the observed increase in the efficiency of direct reprogramming.
We expanded p53-null, GFP-positive clones generated by the three or four factors (six and three clones, respectively). All the clones showed morphology similar to that of mouse ES cells at passage two (S-Figure 1a). Clones generated by the three factors expressed endogenous Oct3/4, endogenous Sox2, and Nanog at comparable levels to those in ES cells (S-Figure 1b). The total expression levels of Oct3/4 and Sox2 were also comparable to those in ES cells, indicating that transgenes were effectively silenced (S-Figure 1c). When transplanted into nude mice, all the six clones gave rise to teratomas containing tissues derived from the three germ layers (S-Figure 1d). These data confirmed pluripotency of iPS cells generated by the three factors from p53-null MEFs.
We found that the expressions of the endogenous Oct3/4 and Sox2 were low in p53-null cells generated by the four factors including c-Myc. (S-Figure 1b). In contrast, the total expression levels of Sox2, Klf4 and c-Myc were markedly higher in these cells than in the remaining iPS cells and ES cells (S-Figure 1c), indicating that retroviral expression remained active in these cells. Consistent with this observation, tumors derived from these cells in nude mice largely consisted of undifferentiated cells, with only small areas of differentiated tissues (S-Figure 1d). Furthermore, the p53-null cells generated by the four factors were not able to maintain ES cell-like morphology after passage five (S-Figure 1e). Thus the c-Myc transgene, in the p53-null background, suppresses retroviral silencing and inhibits acquirement and maintenance of iPS cell identity.
We next tried to generate iPS cells from terminally differentiated somatic cells by the four factors in a p53-null background (). We isolated T lymphocytes from Nanog-GFP reporter mice that were either p53 wild-type or p53-null. We activated T cells by anti-CD3/CD28 antibody and transduced with the four retroviruses. From p53 wild-type T lymphocytes, we did not obtain any GFP-positive colonies. In contrast, we obtained 11 GFP-positive colonies from p53-null T cells (2 × 106 cells), from which three iPS cell lines were established.
These GFP-positive cells were expandable and showed morphology similar to that of mouse ES cells (). They were positive for alkaline phosphatase and SSEA1, makers of mouse ES cells (). They also expressed ES cell marker genes, including Rex1, and Nanog (). In contrast, they did not express T cell specific genes such as FasL, GzmA, GzmB and Ifng. As was the case in iPS cells derived from p53-null MEF, silencing of the transgenes in these cells was not complete. Nevertheless, we obtained four adult chimeric mice from these iPS cells (). As we predicted from the p53-null background and incomplete silencing of the c-Myc retroviruses, three of the four chimeric mice died within seven weeks. We confirmed the rearrangement of the T cell receptors in these iPS cells, various tissues from the chimeras, and the tumor observed in the chimeras (). The intensity of the rearranged bands in tumors was as strong as in iPS cells, indicating that the tumor was derived from iPS cells. These data demonstrated that the four factors could generate iPS cells even from terminally differentiated T cells, when p53 is inactivated.
We then examined whether p53 deficiency increased efficiency of human iPS cell generation. To this end, we introduced the dominant negative mutant P275S or a p53 carboxy-terminal dominant-negative fragment (p53DD)17
into adult human dermal fibroblasts (HDF) together with three or four reprogramming factors. We found that the numbers of iPS cell colonies markedly increased with the two independent p53 dominant negative mutants (S-Figure 2a & b). In another experiment, we examined effects of shRNA against human p53 (shRNA-2)18
. We confirmed that the shRNA effectively suppressed the p53 protein level (S-Figure 2c) in HDF. When co-introduced with the four reprogramming factors, the p53
shRNA markedly increased the number of human iPS cell colonies (S-Figure 2d). Similar results were obtained in the experiments with three reprogramming factors (S-Figure 2e). In contrast, suppression of Rb did not enhance iPS cell generation. A control shRNA containing one nucleotide deletion in the antisense sequence (shRNA-1) did not show such effects (S-Figure 2d & e). Co-introduction of the mouse p53 suppressed the effect of the shRNA. When transplanted into testes of SCID mice, these cells developed teratomas containing various tissues of three germ layers (S-Figure 2f). These data demonstrated that p53 suppresses iPS cell generation not only in mice, but also in human.
To elucidate p53 target genes that are responsible for the observed enhancement of iPS cell generation, we compared gene expression between p53 wild-type MEF and p53-null MEF by DNA microarrays, and between control HDF and p53 knockdown HDF. In MEF, 1590 genes increased and 1485 genes decreased >5 fold in p53-null MEF. In HDF, 290 genes increased and 430 genes decreased >5 fold by p53 shRNA. Between mouse and human, eight increased genes are common, including v-myb myeloblastosis viral oncogene homolog (MYB) and a RAS oncogenes family, RAB39B (S-Table 1). Twenty-seven decreased genes were common between the two species, including p53, cyclin-dependent kinase inhibitor 1A (p21, Cip1), BTG family, member 2 (BTG2), zinc finger, matrin type 3 (ZMAT3), and MDM2.
We transduced four increased genes and seven decreased genes by retroviruses into HDF, together with either the four reprogramming factors alone, or with the four factors and the p53 shRNA. Among the four increased genes, none mimicked the effect of p53 suppression (). Among seven decreased genes, MDM2, which binds to and degrades the p53 proteins, mimicked p53 suppression. Two genes, p53 derived from mouse, and p21 effectively counteracted the effect of the p53 shRNA (). Forced expression of p21 markedly decreased iPS cell generation from p53-null MEF (not shown). In wild-type MEF, the combination of the four factors markedly increased p21 protein levels (). When c-Myc was omitted, p21 proteins still increased, but to a lesser extent than with the four factors. In p53-null MEF, these increases of p21 proteins by either the three or four factors were not observed. These data highlighted importance of p21 as a p53 target during iPS cell generation.
Effect of c-Myc and p53 suppression on p21
Finally, we generated iPS cells from wild-type or p53-null MEF, both containing the Nanog-GFP reporter, by repeated trasnfection of two expression plasmids, one containing the cDNAs of Oct3/4, Klf4, and Sox2 connected with the 2A polypeptides and the other one with the c-Myc cDNA (). We plated 1.3 × 105 MEF and transfected the two plasmids together daily for seven days. On 28th days after the initial transfection, we did not observe any Nanog-GFP positive colonies from wild-type cells (). In contrast, ~100 GFP-positive colonies emerged from p53-null cells. We randomly picked up 12 colonies and found that seven of them did not contain plasmid integrations (). By microinjecting these integration-free iPS cells, we obtained adult chimeric mice (). It remains to be determined whether germline transmission can be obtained.
Effect of p53 suppression on plasmid-mediated mouse iPS cell generation
Our data showed that p53 and p21 suppress iPS cell generation. Suppressive effects of these tumor suppressor gene products on cell proliferation, survival, or plating efficiency should contribute to the observed effect (S-figure 3). In addition, they may have direct effects on reprogramming. Permanent suppression of p53 would lower quality of iPS cells and cause genomic instability. Nevertheless, transient suppression of p53 by siRNA or other methods may be useful in generating integration-free iPS cells for future medical applications.