In vivo reprogrammed pluripotent stem cells from teratomas share analogous properties with their in vitro counterparts

Recently, induced pluripotent stem cells (iPSCs) have been generated in vivo from reprogrammable mice. These in vivo iPSCs display features of totipotency, i.e., they differentiate into the trophoblast lineage, as well as all 3 germ layers. Here, we developed a new reprogrammable mouse model carrying an Oct4-GFP reporter gene to facilitate the detection of reprogrammed pluripotent stem cells. Without doxycycline administration, some of the reprogrammable mice developed aggressively growing teratomas that contained Oct4-GFP+ cells. These teratoma-derived in vivo PSCs were morphologically indistinguishable from ESCs, expressed pluripotency markers, and could differentiate into tissues of all 3 germ layers. However, these in vivo reprogrammed PSCs were more similar to in vitro iPSCs than ESCs and did not contribute to the trophectoderm of the blastocysts after aggregation with 8-cell embryos. Therefore, the ability to differentiate into the trophoblast lineage might not be a unique characteristic of in vivo iPSCs.

Scientific RepoRts | 5:13559 | DOi: 10.1038/srep13559 into the trophoblast lineage is not characteristic of naïve pluripotent ESCs but is a primed pluripotency feature, as human ESCs preferentially differentiate into the trophoblast lineage 20 . These facts prompted us to address more characteristics from new in vivo iPSC lines. Here, we have developed a new method to generate in vivo iPSCs, which can be easily selected from reprogrammable mice. To facilitate the detection of reprogrammed cells, we generated a triple transgenic mouse carrying a transcriptional activator (rtTA; within the ubiquitously-expressed Rosa26 locus), a doxycycline-inducible polycistronic cassette encoding the 4 reprogramming factors (OSKM) within the Col1a1 locus, and Oct4-GFP. We obtained Oct4-GFP + cells from teratomas of the reprogrammable mice. The reprogrammed PSCs were established from these teratoma-derived Oct4-GFP + cells, which were morphologically indistinguishable from ESCs. However, these in vivo reprogrammed PSCs (rPSCs) were more similar to in vitro iPSCs than ESCs and did not contribute to the trophectoderm of the blastocysts after aggregation with 8-cell embryos. Therefore, differentiation ability into the trophoblast lineage might not be a unique characteristic of iPSCs derived from the in vivo milieu.

Results
Generation of reprogrammable mice with Oct4-GFP. The reprogrammable mouse is a useful tool to study the mechanisms underlying cellular reprogramming triggered by doxycycline administration 19 . We generated reprogrammable mice, which were the F1 generation of reprogrammable mice crossed with OG2 mice, which carry the Oct4-GFP (Δ PE) transgene. However, mouse embryonic fibroblasts (MEFs) from these heterozygous reprogrammable mice (OG2 +/− /RTC4 +/− ) were rarely reprogrammed to Oct4-GFP + cells after doxycycline treatment (data not shown). Reprogramming efficiency has been reported to be higher when the reprogrammable MEFs are homozygous for OKSM and Rosa26-M2rtTA transgenes (Ho/Ho) than when MEFs are heterozygous for each transgene (Het/Het) 21 . Thus, we generated a new set of reprogrammable mice homozygous for the transcriptional activator (M2rtTA; within the ubiquitously expressed Rosa26 locus) and doxycycline-inducible polycistronic cassette encoding the 4 reprogramming factors OSKM within the Col1a1 locus, and heterozygous for Oct4-GFP transgene (Δ PE) 19 . We named the reprogrammable mice rOG2 (for reprogrammable OG2) mice (Supplemental material Fig. S1).
First, we tested whether the rOG2 mice were successfully reprogrammable upon doxycycline induction. Neural stem cells (NSCs) and MEFs were obtained from rOG2 mice and cultured in doxycycline-containing ESC medium. At 10-15 days after culture with doxycycline, Oct4-GFP + cells were observed. After further passage, the Oct4-GFP + cells formed dome-like colonies, like ESCs, and were called rOG2-MEF-iPSCs (Supplemental material Fig. S3A and B). This result indicated that the reprogramming cassette and the Oct4-GFP transgene combination in rOG2 mice functioned properly in an in vitro system. Spontaneous generation of teratomas in rOG2 mice. As previously reported, some reprogrammable mice spontaneously developed aggressively growing tumors that histologically presented as largely undifferentiated teratomas 19,22 . The rOG2 mice also formed teratomas spontaneously at the age of 4 weeks without doxycycline administration (Fig. 1A). Interestingly, these teratomas contained not only differentiated cell populations of all 3 germ layers ( Fig. 1B) but also undifferentiated cell populations expressing Nanog (Fig. 1C), indicating that certain somatic cells in rOG2 mice were spontaneously reprogrammed into the pluripotent state in vivo. We dissociated the teratomas into single cells and examined them by fluorescence microscopy. Many Oct4-GFP + cells were detected in the cells from teratomas ( Fig. 1D and Supplemental material Fig. S2). The GFP + cells could be pluripotent cells that were generated in vivo. The GFP + cells were sorted by FACS and cultured on a feeder-layered dish in ESC culture medium. To our surprise, the GFP + cells formed ESC-like colonies in ESC culture conditions (Fig. 1E). These ESC-like cells might be iPSCs induced by the in vivo environment (rOG2-T-rPSCs).
Characteristics of the in vivo iPSCs. By clonal expansion, we established 2 in vivo reprogrammed PSC (rPSC) lines, designated as rOG2-T-rPSCs #1 and #2, from the FACS-sorted GFP + cells ( Fig. 2A). Immunocytochemistry analysis showed that these rOG2-T-rPSCs were positively stained for core pluripotency markers, Oct4, Sox2, and Nanog (Fig. 2B). qRT-PCR analysis also showed that in vivo rPSCs expressed endogenous pluripotency markers Oct4 (endo), Nanog (endo), and Rex1 at levels similar to those of in vitro iPSC and ESCs (less than twofold; Fig. 2B,C). Next, we investigated the DNA methylation status in Oct4 and Nanog promoter regions of rOG2-T-rPSCs #1 and #2. Oct4 and Nanog promoter regions were completely demethylated in rOG2-T-rPSCs #1 and #2, as shown in rOG2-MEF-iPSCs and ESCs (Fig. 2D). The rOG2-T-rPSCs could differentiate into all 3 germ layers in vitro via embryoid body formation (Fig. 2E) and could form germline chimeras (Fig. 2F). Thus, PSCs reprogrammed in vivo from reprogrammable mice possess pluripotent characteristics, including the overexpression of pluripotency marker genes, DNA demethylation in Oct4 and Nanog promoter regions, and in vitro and in vivo differentiation potential. Next, to test the developmental potential to trophoblast lineage, we aggregated the in vivo iPSCs with 8-cell embryos and cultured them until the blastocyst stage. However, we could not observe the contribution of in vivo iPSCs to the trophectoderm of blastocysts (0/200) ( Fig. 2G and Supplemental material Fig. S5). However, in vivo rPSCs were more similar to in vitro iPSCs than ESCs. Scatter plot analysis showed that the r 2 values (square of linear correlation coefficient) between ESCs and rOG2-T-riPSCs #1 and #2 cells were 0.96-0.97 (Supplemental material Fig. S4). The expression of pluripotency markers and Oct4 target genes in rOG2-T-rPSCs #1 and #2 was also closer to the rOG2-MEF-iPSCs than ESCs (Fig. 3B,C).

Discussion
We established in vivo reprogrammed PSC lines from teratomas formed in reprogrammable mice containing the Oct4-GFP marker. These in vivo rPSCs expressed pluripotency markers, displayed an epigenetic status similar to that of ESCs, and could differentiate into all 3 germ layers in vitro and in vivo. Although gene expression profiles of in vivo rPSCs were similar to those of ESCs, they were closer to in vitro rPSCs than ESCs. This result is in conflict with reports of in vivo rPSCs being closer to ESCs than other iPSCs generated in vitro at the transcriptome level 18 . Abad et al. found that in vivo iPSCs were so potent that they differentiated into the trophoblast lineage 18 . However, we did not observe in vivo rPSCs contributing to the trophectoderm of blastocysts after aggregation with 8-cell embryos (Fig. 2G). A recent report showed that only ESCs expressing 2-cell-specific genes had the ability to contribute to both embryonic and extraembryonic tissues 23 . Moreover, ESCs grown in 2i-containing medium had a better potential to differentiate into trophoblast and extraembryonic endoderm than those grown in conventional ESC culture medium 24 . Therefore, the ability of iPSCs to differentiate into the trophoblast lineage might be related more to culture environment than to the in vivo milieu. Stable pluripotent stem cells do not reside in vivo; instead, they exist transiently during early embryonic development. Stable pluripotent stem cells were found to be established from pre-implantation embryo or post-implantation epiblast cells (5.5-6.5 dpc) were cultured in an in vitro system 25 . Therefore, the in vivo environment might not always favor pluripotential reprogramming, although pluripotency might be exhibited. Most importantly, trophoblastic differentiation is not a feature of pluripotency in mice. Thus, it is not reasonable to estimate the quality of pluripotent cells because of trophoblastic differentiation potential.
We used homozygous reprogrammable mice for in vivo rPSCs because MEFs from heterozygous reprogrammable mice were rarely reprogrammed to Oct4-GFP + cells after doxycycline treatment in vitro. Only 1 copy of the reprogramming gene set might not be sufficient for the successful reprogramming of MEFs. Reprogramming efficiency is much lower in MEFs containing 1 copy of each transgene (Het/Het) than in MEFs that were homozygous for OKSM and Rosa26-M2rtTA transgenes 21 . It was recently reported that doxycycline-inducible reprogrammable MEFs by Oct4 and Tet1 could not be reprogrammed in traditional induction medium but were reprogrammed in specific optimal medium condition 26 . Therefore, the doxycycline-inducible reprogrammable system might need at least 2 copies of the gene set or suitable medium conditions for successful reprogramming.
Notably, the in vivo rPSCs in this study were derived from reprogrammable mice without doxycycline treatment. Some reprogrammable mice spontaneously formed aggressively growing tumors containing undifferentiated cells 19 . This phenotype might be attributed to the leaky expression of the transgene in an undefined cell type. The underlying mechanism is unclear; however, it is possible that the in vivo microenvironment influenced the regulation of doxycycline-inducible transgenes. Recently, we found a clue for the reactivation of integrated transgenes in an in vitro system. Integrated reprogramming factor genes, which were inactive in the iPSC state, were spontaneously re-activated when the iPSCs were differentiated into NSCs in vitro 27 . The reactivation of transgenes was closely correlated with the change in the levels of DNA methyltransferases during the differentiation of iPSCs. These results indicate that somatic cells could be reprogrammed into pluripotent cells not only in vitro but also in vivo 19,22 .
As shown by in vivo iPSC generation through teratoma formation, the process of iPSC derivation shares many characteristics with cancer development. During reprogramming, differentiated somatic cells acquire properties of self-renewal along with unlimited proliferation and exhibit global alterations of the transcriptional program, which are also critical events during carcinogenesis 28 . Partial reprogramming in vivo can bring about cancer development 29,30 . Ohnishi and colleagues showed that premature termination of reprogramming by transient expression of reprogramming factors led to tumor formation in vivo. These tumor cells could be fully reprogrammed into iPSCs by further induction of reprogramming factors. Therefore, tumor formation in vivo could be a result of incomplete reprogramming. In the present study, we showed that once completely reprogrammed, in vivo rPSCs formed through tumor formation possessed pluripotency and resembled in vitro iPSCs, which do not contribute to the trophectoderm.

Isolation of in vivo rPSCs from rOG2 teratomas. Pieces of teratomas were washed and chopped
in PBS containing 10 × penicillin/streptomycin/glutamine. Collected tissues were centrifuged at 900 rpm for 3 min. For single-cell dissociation, tissues were pipetted in 0.25% trypsin and incubated at 37 °C for 5 min. This step was repeated thrice. Dissociated tissues were then filtered through a 70 μ m mesh cell strainer for single-cell purification.
Briefly, the amplified products were verified by electrophoresis on 1% agarose gel. The desired PCR products were used for subcloning using the TA cloning vector (pGEM-T Easy Vector; Promega). The reconstructed plasmids were purified, and individual clones were sequenced (Solgent Corporation).
Microarray-based analysis. Total RNA was isolated using an RNeasy Mini Kit (Qiagen) and digested with DNase I (RNase-free DNase, Qiagen) according to the manufacturer's instructions. Total RNA was amplified, biotinylated, and purified using the Ambion Illumina RNA amplification kit (Ambion) according to the manufacturer's instructions. Biotinylated cRNA samples (750 ng) were hybridized to each MouseRef-8 v2 Expression BeadChip (Illumina), and signals were detected using Amersham fluorolink streptavidin-Cy3 (GE Healthcare Life Sciences) by following the Illumina bead array protocol. Arrays were scanned with an Illumina BeadArray Reader confocal scanning system according to the manufacturer's instructions.
Raw data were extracted using the software provided by the manufacturer (Illumina GenomeStudio v2011. 1, Gene Expression Module v1.9. 0). Array data were filtered on the basis of p-values of < 0.05 in at least 50% of the samples. The selected probe signal value was logarithmically transformed and normalized using the quantile method. Comparative analyses were carried out using the local pooled error test and fold change. False discovery rate was controlled by adjusting the p-values using the Benjamini-Hochberg algorithm. Hierarchical clustering was performed using average linkage and Pearson distance as a measure of similarity.