Systematic in vitro and in vivo characterization of Leukemia‐inhibiting factor‐ and Fibroblast growth factor‐derived porcine induced pluripotent stem cells

SUMMARY Derivation and stable maintenance of porcine induced pluripotent stem cells (piPSCs) is challenging. We herein systematically analyzed two piPSC lines, derived by lentiviral transduction and cultured under either leukemia inhibitory factor (LIF) or fibroblast growth factor (FGF) conditions, to shed more light on the underlying biological mechanisms of porcine pluripotency. LIF‐derived piPSCs were more successful than their FGF‐derived counterparts in the generation of in vitro chimeras and in teratoma formation. When LIF piPSCs chimeras were transferred into surrogate sows and allowed to develop, only their prescence within the embryonic membranes could be detected. Whole‐transcriptome analysis of the piPSCs and porcine neonatal fibroblasts showed that they clustered together, but apart from the two pluripotent cell populations of early porcine embryos, indicating incomplete reprogramming. Indeed, bioinformatic analysis of the pluripotency‐related gene network of the LIF‐ versus FGF‐derived piPSCs revealed that ZFP42 (REX1) expression was absent in both piPSC‐like cells, whereas it was expressed in the porcine inner cell mass at Day 7/8. A second striking difference was the expression of ATOH1 in piPSC‐like cells, which was absent in the inner cell mass. Moreover, our gene expression analyses plus correlation analyses of known pluripotency genes identified unique relationships between pluripotency genes in the inner cell mass, which are to some extent, in the piPSC‐like cells. This deficiency in downstream gene activation and divergent gene expression may be underlie the inability to derive germ line‐transmitting piPSCs, and provides unique insight into which genes are necessary to achieve fully reprogrammed piPSCs. 84: 229–245, 2017. © 2016 Wiley Periodicals, Inc.


INTRODUCTION
The development of mouse (Takahashi and Yamanaka, 2006) and human induced pluripotent stem cells (iPSCs) (Takahashi et al., 2007;Yu et al., 2007) ignited the hope that patient-specific pluripotent stem cells would soon be used for therapy (Singh et al., 2015). Mice are widely used as biomedical models, despite the fact that the findings often fail to translate successfully to humans (McGonigle and Ruggeri, 2014), hence the urgent need to generate more human-like animal models in which the potentials and risks of iPSC-based therapy can be evaluated.
The pig is considered an excellent model for several devastating human diseases, such as diabetes, cardiovascular, and nervous system diseases. Miniature pigs were successfully used as a cell-replacement model for amyotrophic lateral sclerosis (ALS), as they were able to receive human spinal stem cells (Boulis et al., 2011;Glass et al., 2012). A number of genetically modified pigs have also been produced as potential disease models that faithfully recapitulate human disease characteristics (Flisikowska et al., 2014). Genetic modifications are currently introduced into pigs by cloning and somatic cell nuclear transfer (SCNT) (Du et al., 2007), but only 1À5% of the reconstructed embryos yield live piglets, resulting in a costly and labor-intensive process (Schmidt et al., 2010;Gao et al., 2011;Lee et al., 2013). Further stymying the efficiency of establishing genetic pig models, a considerable proportion of cloned piglets exhibit epigenetic abnormalities, resulting in a perinatal mortality rate of approximately 50% (Schmidt et al., 2015).
Porcine pluripotent stem cells provide an alternative method to produce genetically modified pigs based on germ line-transmitting chimeric embryos since stem cells are easier to manipulate, divide rapidly, and do not show signs of cell senescence. A strong focus on establishing porcine iPSCs (piPSCs) is underway, both to refine the pig as a model for human iPSC-based therapy and to produce genetically modified porcine disease models by means of germ line-transmitting chimeras (Ezashi et al., 2009;Telugu et al., 2010;West et al., 2010;Montserrat et al., 2011;Liu et al., 2012;Kwon et al., 2013). In contrast to mice (Okita et al., 2007) and rats (Hamanaka et al., 2011), evidence of germ-line contribution of piPSCs to chimeric embryos is challenging. Only a single report described such germ-line transmission, and in this instance, the viability of the porcine offspring was severely compromised (West et al., 2011).
Murine embryonic stem cells (ESCs) can be classified into na€ ıve (ground) and primed states: na€ ıve ESCs originate from the inner cell mass whereas primed ESCs originate from the epiblast (also referred to as epiblast stem cells) . Na€ ıve murine ESCs, and their murine iPSC counterparts, rely on Leukemia inhibitory factor (LIF) stimulation to retain pluripotency whereas primed mouse EpiSC, as well as human ESCs, rely on Fibroblast growth factor 2 (FGF2) stimulation. Two small-molecule inhibitors are also widely used to block Mitogen-activated protein kinase (MEK) and Glycogen synthase kinase 3-beta (GSKB3) activity (Telugu et al., 2010). The growth factor (s) or conditions that effectively maintain piPSC pluripotency, however, are unknown. Application of the two inhibitors together with LIF allowed for the derivation of piPSCs, which show similarities to na€ ıve state stem cells, including expression of pluripotency markers and spontaneous differentiation into all three germ layers, but failed to show germ-line contribution in chimeras or result in the birth of live offspring (Telugu et al., 2010;Ezashi et al., 2012;Fujishiro et al., 2013). Conversely, piPSCs were successfully established using basic FGF (West et al., 2010), indicating uncertainty as to which growth factor is the most beneficial for establishing induced stem cells.
Here, we present a systematic comparison of two piPSC lines, both derived from a neonatal fibroblast line expressing the green fluorescent protein Venus (Garrels et al., 2011). The cells were reprogrammed using a lentiviral construct containing a doxycycline-inducible promoter expressing the porcine coding sequences of OCT4 (also known as POU5F1), SOX2, c-MYC, and KLF4 (pOSMK). The addition or removal of doxycycline allows for the regulation of exogenous gene expression. The lines were derived using either LIF or FGF in combination with PD0325901 (a MEK-inhibitor) and CHIR9902 (a GSKB3 inhibitor), denoted as ''2i.'' Characterization of the resultant piPSC lines included assessment of pluripotency marker expression by immunocytochemistry, quantitative reversetrancription PCR, and transcriptome analyses, as well as teratoma formation and chimera contribution.

Generation and Characterization of LIF and FGF piPSCs
A lentiviral construct in which expression of the porcine sequences of OCT4, SOX2, c-MYC, and KLF4 are under the control of a doxycycline-inducible TetO promoter (TetO-pOSMK) (Fig. S1) was simultaneous transduced with a second lentivirus carrying the reverse tetracycline-controlled transactivator (rtTA) gene, which allows for doxcycyclne-dependent activation of the pluripotency factors, into porcine neonatal fibroblasts derived from piglets ubiquitously expressing the green fluorescent protein ''Venus'' (Garrels et al., 2012). Either LIF or FGF, in combination with 2i, was then used to generate and maintain na€ ıve-like and primed-like piPSC, respectively.

Molecular Reproduction & Development
colonies (denoted as LIF and FGF piPCSC, respectively): LIF piPSCs grew as small, dome-shaped colonies (Fig. 1A) whereas the primed (FGF-derived) piPSCs grew as flat, monolayer colonies (Fig. 1A). By passages 15À20, the morphologies of the two types of colonies flattened and became morphologically more similar.
Quantitative real-time PCR (qPCR) was used to profile the expression of key stem cell markers compared to the parental neonatal fibroblasts (Fig. 1B). For example, markers of na€ ıve pluripotency include LIN28 (Hanna et al., 2010) and NROB1 (Hall and Hyttel, 2014). Greater than 120-fold increases were observed in the expression of OCT4 (Fig. 1B), KLF4, c-MYC, NANOG, LIN28, and NROB1, under LIF and FGF conditions (Fig. 1B). The Figure 1. Assessment of pluripotency markers in LIF piPSC and FGF piPSC. Evaluation of the LIF piPSCs and FGF piPSCs. A: Morphological survey by brightfield microscopy (BF) and alkaline phosphatase staining (AP). Pluripotency-associated protein expression was evaluated via NANOG, SSEA3, and SSEA4 (each in red). Venus fluorescence (green) demonstrated the source of the cells. Hoechst (blue) was used as a nuclear counterstain. Scale bars, 50 mm (blue) or 100 mm (white). B: Relative increase in mRNA abundance of the key pluripotency factors NANOG, OCT4, LIN28, c-MYC, NROB1, and KLF4 in Venus piPSCs lines. Expression of individual samples was normalized to GAPDH (Glyceraldehyde 3-phosphate dehydrogenase), and the overall change in gene expression was scaled to the gene expression in the parental porcine neonatal fibroblasts. C: Comparison of pluripotency marker expression in LIF piPSCs versus FGF piPSCs (LIF / FGF ratio). increased transcript abundance of OCT4, KLF4, and c-MYC could be the sum of both exogenous and endogenous sources, whereas the increased expression of NANOG, NROB1, and LIN28 is exclusively endogenous. Normalization of LIF piPSC transcript abundance to that of FGF piPSC demonstrated comparable or slightly reduced expression of OCT4, NANOG, c-MYC, SOX2, and STAT3 under LIF conditions, whereas LIN28 expression was significantly increased (Fig. 1C). Abundance of the LIF receptor was similar in both piPSC lines, whereas isoforms of the FGF receptor were reduced to half under LIF compared to FGF conditions. The observed expression profiles were further confirmed by our RNA-sequencing data (Supplementary Information).
Doxycycline withdrawal from the piPSC media resulted in differentiation of both the LIF and FGF piPSCs, with no apparent difference between them. Withdrawl of LIF or FGF in the presence of doxycycline showed less dramatic results, and the cells generally maintained colony and cell morphology (Fig. S2). LIF piPSCs displayed a karyotype of 38, XXY in all 20 analyzed metaphases; conversely, 15 of the 20 metaphase spreads of FGF piPSC were normal with 38, XY, while 5 showed a gain of DNA on chromosome 9 (38, XY, plus [9]).
In summary, both LIF and FGF piPSCs shared pluripotency features, but also exhibited subtle differences in gene expression related to their na€ ıveand primed-like states. Cells under both culture conditions remained dependent on transgene expression to maintain pluripotency.

Differentiation of LIF and FGF piPSCs in Embryoid Bodies and of LIF piPSCs in Teratomas
The pluripotency of LIF and FGF piPSCs was evaluated by in vitro embryoid body formation. Both LIF and FGF piPSCs possessed the competence to differentiate into all three germ layers, as represented by the expression of smooth muscle actin (mesoderm), beta-3 tubulin (ectoderm), and alpha-fetoprotein (endoderm) (Fig. 2AÀF).
The differentiation capacity of both LIF and FGF piPSCs was assessed next by asking if they could give rise to teratomas in vivo. An initial experiment was conducted with 10 non-obese diabetic-severely compromised immunodeficient (NOD-SCID) mice, without doxycycline in their drinking water. Six mice were injected with LIF piPSC and four with phosphate-buffered saline, and monitored for 4 weeks. None of the mice had developed tumors at that point, so the experiments were halted. The next iteration of this study involved six NOD-SCID mice subcutaneously injected, in the flank, with LIF piPSCs, FGF piPSCs, or parental neonatal fibroblasts as a control group, all provided with doxycycline in their drinking water. Three of the six LIF piPSC-injected mice developed teratomas after 14 days, whereas none of the FGF piPSC-or fibroblast-injected mice did so. When doxycycline administration was concluded after 7 weeks, the teratomas regressed in size. These teratomas were isolated at a diameter of approximately 5 mm for analysis. Hematoxylin-and-eosin staining revealed cuboidal epithelial tissue (potential mesoderm), smooth muscle cells (mesoderm), and thyroid-like follicles (endoderm) (Fig. 2GÀI). Immunohistochemistry identified b-III tubulin-positive (ectoderm) (Fig. 2J), smooth muscle actin-positive (mesoderm) (Fig. 2K), and SOX17-positive (endoderm) (Fig. 2L) tissue coexpressed with Venus, which verified the porcine origin of the cells.

In Vitro Contribution to Chimeric Embryos
We further assessed the pluripotency of LIF and FGF piPSCs by injecting 15 cells into parthenogenic blastocysts, and tracing their fate during embryogenesis using live-cell imaging, differential trophectoderm and inner cell mass staining, and transmission electron microscopy. Development of the porcine parthenogenic blastocysts was monitored for 48 hr following injection of either LIF or FGF piPSCs or the parental neonatal fibroblasts, using Venus expression to track them within the embryos (Fig. 3A). The different groups of embryos displayed similar survival rates, as indicated by the presence of developing and hatching blastocysts, in all three experiments performed (Fig. 3C). No statistically significant difference between the groups was observed with respect to the numbers of embryos with Venus-positive cells when the embryos were cultured without doxycycline; however, the LIF piPSC group exhibited significantly more embryos with Venus fluorescence when the embryos were cultured with doxycycline (P ¼ 0.0015, odds ratio ¼ 4.1; confidence interval ¼ 1.7; 9.7). Differential staining revealed that the majority of the LIF and FGF piPSCs localized to the trophectoderm of the blastocyst (Fig. 3B), and that 1À5 Venus-positive cells were visualized per embryo (n ¼ 20). The LIF piPSCs were readily recognizable in the blastocyst by their small, dark (electrondense) lipid droplets and elongated mitochondria, as opposed to the embryonic cells displaying large, electron-lucent lipid droplets and an abundance of rounded mitochondria ( Fig. 3D and E). LIF piPSCs were observed both outside and inside the blastocyst cavity, which was resealed as soon as 2 hr following injection by tight junctions in the trophectoderm. Tight junctions were also observed between the LIF piPSCs and the trophectoderm at the sites of integration into the latter cell compartment (Fig. 3E).

In Vivo Contribution to Chimeric Embryos
LIF piPSCs appear to possess higher competence to contribute to chimeric embryos, so we set up two embryotransfer experiments to evaluate the ability of these cells to contribute to chimeric embryonic development in vivo. In the first experiment, 45 chimeric Day-5.5 embryos, produced by injecting Day-4.5 in vivo embryos with approximately 15 LIF piPSCs, were transferred to a surrogate sow. A total of 16 embryos and their surrounding membranes were isolated on Day 32 (Fig. S3A). The embryos contained no Venus fluorescence, Venus mRNA, or Venus DNA in the evaluated organs. Yet, two sets of embryonic membranes contained Venus DNA (Fig. S3B), suggesting that LIF piPSCs had integrated into the trophectoderm. In the second experiment, 110 chimeric Day-5.5 embryos, produced by injecting Day-4.5 in vivo embryos with approximately 15 LIF piPSCs, were transferred into three sows that received doxycycline in their food from 2 days before until 4 days after transfer to ensure expression of the exogenous rerogramming genes until the period when initial cell differentiation occurs in the embryos at gastrulation, around Day 11. Two of the three sows became pregnant, and the  microscopy (Venus) pictures of three live, hatching chimeric blastocysts imaged with a Nikon Biostation (20Â magnification). The chimeras contain Venus-expressing LIF piPISCs. B: Confocal image (40Â magnification) of a 7-day-old embryo after differential staining for CDX2 (red)-positive trophectoderm versus Hoechst (blue)-positive-but-CDX2-negative inner cell mass cells; the Venus (green)-positive cells are Venus-expressing LIF piPlSC (white arrow). C: G-plot distribution of hatched and fluorescent in vitro chimeras 48 hr after injection of FGF piPSC (Venus FGF); FGF piPSC with doxycycline (Venus FGF DOX); LIF piPSC (Venus LIF); LIF piPSC with doxyxycline (Venus LIF DOX); or parental neonatal fibroblasts (Venus Fibro). DÀE: Electron micrograph presenting a section of an in vitro chimera 2 hr after injection of LIF piPlSCs. Box is magnified in ''E,'' showing the interface between the trophectoderm and an piPSC. Black arrows, small electron-dense lipid droplets specific for piPSCs; black arrowheads, tight junctions; L, large lipid droplets specific for embryonic cells; N, nucleus; TE, trophectoderm.

Molecular Reproduction & Development
embryos were harvested at Day 32. None of the collected embryos displayed the presence of Venus DNA; embryonic membranes were not collected in the second experiment.

Analyses of the Transcriptome of LIF and FGF piPSCs in Comparison With Pluripotent Cell Populations From Porcine Embryos
Global analysis of gene expression was used to compare the transcriptome of parental porcine neonatal fibroblasts, LIF piPSCs, FGF piPSCs, and three distinct populations of pluripotent cells obtained from in vivo-derived porcine embryos (datasets deposited in the Gene Expression Omnibus database at www.ncbi.nlm.nih.gov/ geo/, record GSE92889). These pluripotent cells were isolated by immunosurgery from (i) the inner cell mass of ten embryos at embryonic Day 7/8, which represent a potential na€ ıve state of pluripotency; (ii) from the epiblast of the embryonic disc (after manual removal of the hypoblast) from 10 embryos at embryonic Day 10/11, which represent a potentially primed state of pluripotency; and (iii) from the epiblast of the elongated embryonic disc (after manual removal of the hypoblast) during initial gastrulation from five embryos at embryonic Day 12/13, which represent a mixture of pluripotent and differentiating cells. The resulting heat maps revealed that LIF and FGF piPSCs, as well as their porcine neonatal fibroblast founders, clustered together, and are clearly distinct from the embryonic cell populations (Fig. 4A). A certain similarity in gene expression patterns between the fibroblasts and their resulting piPSCs might be expected, given that the parental fibroblasts were derived from neonatal pigs, but the surprising match between these cell types clearly indicates the incomplete pluripotent nature of the piPSCs. Among the three different embryonic stages, the embryonic epiblast and the differentiating gastrulating epiblast samples clustered apart from the inner cell mass sample (Fig. 4A); principal component analysis confirmed this observation.
We further dissected the expression patterns of the piPSCs and embryonic cell types by implementing a fiveset Venn diagram (Fig. 4B). Generation of germ linetransmitting chimeric porcine embryos can be achieved using freshly isolated cells of the inner cell mass (Nagashima et al., 2004), so we conducted a detailed comparision of this particular stage with the piPSCs. Genes exclusively expressed in the Day7/8 inner cell mass, but not in the piPSCs, were of particular interest since they might represent overlooked transcription factors or receptors that are necessary for maintenance of pluripotency in the pig. We identified 279 genes exclusively expressed in the Day 7/8 inner cell mass. A comprehensive literature search of predicted gene functions resulted in 22 genes that we considered potentially important for the pluripotency of the porcine inner cell mass (Table 1), and include genes encoding transcription factors, growth factor receptors, membrane proteins involved in cellÀcell interactions, as well as proteins involved in phosphoinositide 3-kinase/ AKT, transforming growth factor beta, and Wnt signaling À Àall of which have previously been shown to play a role in pluripotency and differentiation (Watanabe et al., 2006;Watabe and Miyazono, 2009;Holland et al., 2013).
Finally, we performed a bioinformatic comparison of the pluripotency gene network in order to reveal aberrancies in the LIF and FGF piPSCs, comparing LIF versus FGF piPSC versus cells from the Day-7/8 oorcine inner cell mass. The selection of genes was based on previous publications of pluripotency networks in the mouse embryonic stem cells (Xu et al., 2014), and included the genes expressed in our lentiviral reprogramming vector (OCT4, SOX2, c-MYC, and KLF4). We created correlation networks for LIF iPSCs, FGF iPSCs, and the inner cell mass (Fig. 5). The correlation network of the two iPSC lines differ remarkably from the A: Heat map hierarchical clustering diagram showing similarity measurement based on Pearson correlation and the complete linkageneighbor method across LIF piPSCs, FGF piPSCs, parental neonatal fibroblasts, and three embryonic developmental stages (Day 7/8 inner cell mass; Day 10/11 epiblast; and Day 12/13 gastrulating epiblast). B: Multi-set Venn diagram showing the overlap and intersections of differentially expressed genes among LIF piPSC (red), FGF piPSC (blue), Day 7/8 inner cell mass (brown), Day 10/11 epiblast (yellow), and 12/13 gastrulating epiblast (green). correlation network of the inner cell mass; indeed, even though almost all key pluripotency genes are expressed in the piPSC and inner cell mass, they do not have the same correlations with other important pluripotency genes within the network. The most striking difference among these networks was that ZFP42 was not expressed and had no correlation to other pluripotency genes in the piPSC-like cells, whereas ZFP42 is expression in the inner cell mass and shows strong correlation with NANOG, OCT4 (POU5F1), and SALL4 ( Fig. 5A and B). Conversely, NROB1, which is completely absent and has no correlations to other genes in the inner cell mass, is expressed in both iPSC lines ( Fig. 5A and B): NROB1 is correlated to ESRRB in the LIF piPSCs, (Fig. 5B) whereas NROB1 is correlated to ESRRB and OCT4 (POU5F1) in the FGF iPSCs (Fig. 5A). Both porcine iPSC networks diverged from one another, except for correlations between SALL4 and JARID2 and between OCT4 (POU5F1) and SOX2,  (Fig. 5C). Another striking difference between piPSC lines and the inner cell mass was the correlation of NANOG and KLF4, two essential pluripotency markers (Radzisheuskaya and Silva, 2014). NANOG has strong correlation with ZFP42, SALL4, and OCT4 (POU5F1) in the inner cell mass, whereas no significant correlations were observed with NANOG in either piPSC lines. Similar results are present for c-MYC, which correlates with TCF3 in the inner cell mass but showed no significant correlations in the piPSC lines. Another interesting outcome of this comparison is that most of the LIF iPSC network is present in the FGF iPSC network; the exception is ESRRB, which correlates with OCT4 (POU5F1), ATOH1, and NROB1 in LIF iPSCs (Fig. 5B) versus with ATOH1 and NROB1 in FGF iPSCs (Fig. 5A). Reciprocally, FGF iPSCs had several significant correlations that are absent in LIF iPSCs and the inner cell mass. These findings together underline the fact that reprogramming was incomplete, but also show that key gene expression and correlation among these genes differ not only between the inner cell mass, but also between the derived piPSC lines.

DISCUSSION
Here, we report the use of a lentiviral doxycycline-inducible OCT4, c-MYC, SOX2, KLF4 construct utilizing porcine coding sequences for the generation of potential piPSCs. By culturing the resulting piPSC under either LIF or FGF conditions, we aimed to produce na€ ıve-like and primedlike piPSCs, respectively. Accordingly, the LIF piPSCs presented dome-shaped colonies, in accordance with earlier observations Cheng et al., 2012;Fujishiro et al., 2013), whereas the FGF piPSCs presented the expected flat colony morphology, as also noted earlier (West et al., 2010;Ezashi et al., 2012).
Our transcriptome analyses of these lines showed some differences between LIF and FGF piPSCs. LIF piPSCs exhibited a 2-fold and a 1.22-fold increase in LIN28 and NROB1 expression, respectivelyÀ Àboth markers associated with na€ ıve pluripotency in mouse embryonic stem cells (Hanna et al., 2010;Guo et al., 2016)À Àpossibly indicating that LIF piPSCs are superior to the FGF piPSCs in regards to pluripotency. LIN28 encodes a highly conserved, RNAbinding protein that down-regulates let-7 microRNA (Heo et al., 2008). let-7 plays a role in cell-cycle regulation Melton et al., 2010), and its down-regulation may give the LIF piPSCs a small but significant advantage in self-renewal capacity. LIN28 may also up-regulate OCT4 translation, which would also give LIF piPSCs an advantage (Qiu et al., 2010). Indeed, more LIF piPSCs survived after injection into blastocysts cultured with doxycycline. Interestingly, the elevated KLF4 expression in both the LIF and FGF piPSC lines is negligible when normalized to the endogenous expression of KLF4 in porcine fetal fibroblasts (Fig. 1B). KLF4 is associated with reverting the primed state of pluripotency into the na€ ıve state via overexpression in murine stem cells (Guo et al., 2009), so the absence of significantly higher expression of this factor in our piPSCs could either be due to incomplete reprogramming or the possibility that KLF4 is not a key transcription factor involved in the generation and maintenance of pluripotency in piPSCs.
The general expression profile of cell surface markers in piPSCs is controversial, partially due to the great heterogeneity among various lines as well as the imprecise definition of cell surface markers associated with bona fide piPSCs. Expression of SSEA3 and SSEA4 as well as TRA1-81 and TRA1-60 was previously observed (Wu et al., 2009), whereas others report only surface expression of SSEA4 (Esteban et al., 2009;West et al., 2011). In our study, LIF piPSCs expressed both cell surface proteins SSEA3 and SSEA4 while FGF piPSCs only expressed SSEA3. Interestingly, the only chimeras derived from our piPSCs originated from lines expressing SSEA4. LIF piPSCs had a greater propensity to contribute to chimeric embryos in vitro and to form teratomas when cultured with doxycycline. Differential staining of the injected embryos to assess the potential of the LIF piPSCs to integrate into the blastocyst  revealed a preference for integration into the trophectodermÀ Àan observation confirmed by ultrastructural observation of tight junctions between injected piPSCs and cells of the trophectoderm as early as 2 hr post-injection. Amplification of the genomic Venus sequence at Day 32 of pregnancy only from embryonic membranes confirmed the extraembryonic integration preference. Several groups have reported similar results, with iPSCs tending to differentiate towards the trophectoderm lineage (West et al., 2010;Ezashi et al., 2011;Fujishiro et al., 2013). In our study, the LIF piPSCs carried a supernumerary X chromosome, which may have biased them towards a trophectoderm fate (Viuff et al., 2002); since the majority of the literature on piPSCs did not include karyotyping, we cannot place our observations in perspective.
Until now, only piPSCs with reprogramming genes driven by constitutive promoters have been reported to contribute to chimeric embryonic development (West et al., 2010;Fujishiro et al., 2013). West et al. (2010) even reported germ line transmission from chimeras generated from mesenchymal stem cell-derived piPSCs, although the two survivors, out of 43 piglets, resulting from this germ line transmission exhibited many congenital defects, possibly due to epigenetic deviations (West et al., 2011); these results have not been reproduced to date. Fujishiro et al. (2013) demonstrated embryonic piPSC contribution until Day 63 of development, but could not detect chimerism in later-stage embryos, potentially due to the sustained transgene expression. Based on these previous experiments, we chose to evaluate reprogramming using a doxycycline-regulated promoter, which provides a means to control the expression of transgenes. Both our time-lapse chimeric analyses and the teratoma experiments revealed that the functionality of the LIF piPSCs was compromised when the transgenes were not expressed. Thus, we decided to administer doxycycline to the sows from 2 days before until 4 days after embryo transfer to improve the chimeric efficiency in the blastocysts transferred to the recipient sows. This would maintain transgene expression until embryonic Day 11 and 12, when gastrulation is initiated (Oestrup et al., 2009). Unfortunately, genomic integration of the Venus transgene was still not detected in any of the resultant embryos.
The fact that few of the piPSCs generated to date are capable of generating healthy, chimeric offspringÀ Àexcept of the mesenchymal stem cell-derived iPSCs (West et al., 2010)À Àled us to identify genes expressed in the inner cell mass of porcine embryos, but not in LIF and FGF piPSCs. This approach was expected to help identify crucial signaling pathways not properly activated in the piPSC. Global analysis of transcriptomes of porcine neonatal fibroblast or LIF and FGF piPSCs, compared to various pluripotent or differentiating cell populations originating from porcine embryos, showed that LIF and FGF piPSCs were highly similar to one another and to the parental neonatal fibroblasts. Moreover, these three cell populations clustered separately from the embryonic pluripotent cell types. The two later embryonic stages, Day 10/11 epiblast and Day 12/13 gastrulating epiblast, clustered together and remained distinct from the Day 7/8 inner cell mass. This cladogram can be explained by the more-pluripotent state of the inner cell mass, versus the initial differentiation that occurs in the later stages of development. This approach also allowed us to identify 279 genes expressed exclusively in the early pluripotent inner cell mass but not in our piPSCs.
Twenty-two genes may be significant for piPSCs to achieve true pluripotency. Three of these genes were particularly notable: (i) CSF2RA encodes the alpha subunit of the receptor for colony stimulating factor 2 (CSF2), suggesting that this signaling factor contributes to maintenance of pluripotency. (ii) Two members of the Gap junction beta family (GJB1 and GJB5) were also exclusively expressed in the inner cell mass. Gap junction proteins mediate intercellular metabolic and electrical communication, and have been shown to regulate cell-fate choices within early embryos and embryonic stem cell (Wong et al., 2008). These channel protein could participate in the mesenchymal-to-epithelial transition during piPSC reprogramming and during subsequent maintenance of pluripotency Hoeffding and Hyttel, 2014). Finally, (iii) KLF17 was also exclusively expressed in the inner cell mass. KLF17 was recently identified in pluripotent human embryonic cells (Guo et al., 2016), and belongs to the Kr€ uppel-like factor family. A related factor, KLF4, is widely accepted as a pivotal transcription factor in human and mouse embryonic stem cells and iPSCs, and is one of the four factors needed to reprogram fibroblasts to iPSCs (Takahashi and Yamanaka, 2006). KLF17, however, can inhibit the epithelial-to-mesenchymal transition (Dong et al., 2014), and thus may be involved in cell fates related to pluripotency.
Our bioinformatics comparison of the co-expression networks of LIF piPSCs, FGF piPSCs, and inner cell mass revealed similar relationships among key pluripotency genes in LIF piPSC and FGF piPSC. Yet both cell lines differed from the inner cell mass in their expression of genes crucial for pluripotency in mouse iPSCs, such as Zfp42, Oct4 (Pou5f1), and Nanog. These findings clearly indicate that the pluripotency gene network is different between porcine versus murine piPSC or that activation of the necessary genes to establish and sustain pluripotency were not sufficiently activated in our piPSCs; the latter is more likely. Perhaps inclusion of one of the candidate genes into the reprogramming cocktail will provide the impetus needed to achieve full reprorgamming to a pluripotent state.
2.5 Â 10 6 HEK-293T cells were plated into 10-cm dishes. The following day, these HEK-293T cells were transfected with 20 mg FUdeltaGW-pOSMK or reverse tetracycline-controlled transactivator (rtTA) in combination with packaging plasmids (5 mg pMDG and 15 mg pBRD8.91) (a kind gift from D. Trono, Lausanne) using the calcium-phosphate method (see Nielsen et al., 2009); the medium was changed the following day. After another 24 hr, the medium was harvested, centrifuged for 10 min at 48C at 2500g, and filtered through a 0.45-mm polyvinylidene fluoride membrane filter (Millipore). Filtered medium was then ultracentrifuged in a Beckman ultracentrifuge using an SW32 rotor to reduce the volume 700À1,000-fold. The titers of the viral preparations (multiplicity of infectivity [MOI]) were assessed using quantitative PCR, according to Nielsen et al. (2009). Briefly, cells were transduced with dilution series of the concentrated vector stocks, and then cells were harvested and lysed. Lysates were then subjected to quantitative PCR anslysis to determine the amount of integrated viral DNA. The titer was calculated by comparing the viral DNA content to a vector with known titer (as assessed by Venus expression in transduced cells).

Karyotyping
Karyotyping was performed at passage 13 by Cell Guidance Systems (Babraham Research Campus, Cambridge, UK).

Alkaline Phosphatase Staining
iPSCs were fixed for 30 min at room temperature with 4% paraformaldehyde. FastRed (SigmaÀAldrich) dissolved in MilliQ water (1 mg/mL) and Napthol phosphate (40 mL/mL) (SigmaÀAldrich) were mixed, and the fixed Venus piPSCs were incubated in this mixture for 15À30 min. Cells were subsequently washed twice with distilled water, and then analyzed or stored at 48C.

Immunofluorescence Analysis
Venus piPSCs were fixed for 15 min at room temperature, either on glass slides or directly on the culture vessel, using 4% paraformaldehyde. The cells were subsequently permeabilized with 0.1% Triton X-100 in phosphatebuffered saline (PBS), and blocked for 1 hr at room temperature with either normal donkey or goat serum. Incubation was then performed overnight at 48C using the following primary antibodies: rabbit anti-NANOG (1:400 dilution of 500-P236) (Preprotech); SSEA3 (1:100 dilution of 330302) (Biolegend); SSEA4 (1:100 dilution of 330402) (Biolegend); rabbit anti-alpha-fetoprotein (1:500 dilution of A0008) (DAKO); mouse anti-b-III tubulin (1:500 dilution of T8660) (Sigma); mouse anti-smooth muscle actin (1:500 dilution of M0851) (DAKO); SOX17 (1:400 diluton of sc-17355) (Santa Cruz Biotechnologies). Negative controls included non-primary controls for all antibodies, or isotype controls using mouse IgG. The cells were subsequently washed twice for 10 min in PBS, and then incubated for 30 min at room temperature with fluorophore-conjugated secondary antibodies in 0.25% bovine serum albumin (BSA)/PBS. Samples were washed twice for 5 min in PBS, incubated for 1 min with Hoechst at 0.1 mg/mL in PBS, and washed a final time for 5 min in PBS. Cells grown on slides or teratoma sections were then mounted in DAKO Fluorescent Mounting Medium.

Spontaneous Differentiation via Embryoid Bodies
Venus piPSCs were dissociated with Dispase (SigmaÀ Aldrich) at passage 10, transferred to low-binding plates (Z721050-7EA, Lot 3110; SigmaÀAldrich), and cultured in piPSC medium containing 2 mg/mL doxycycline (SigmaÀ Aldrich) in the absence of LIF and 2i. Doxycycline was withdrawn after 4 days. and after 7 days, The embryoid bodies were plated on 0.1% gelatin-coated wells after 7 days, and grown in complete DMEM AQ (Sigma) supplemented with 1% pen/strep and 10% FBS (Hyclone). After an additional 3 weeks of differentiation, the cells were tested for characteristic lineage-specific protein expression (see Immunofluorescence Analysis) of alpha-fetoprotein (endoderm), beta three tubulin (ectoderm), and smooth muscle actin (mesoderm).

RNA Extraction and cDNA Synthesis
Venus piPSCs and neonatal fibroblast samples were obtained in quadruplet, derived from one founder piPSC line for the LIF and FGF conditions. These founder piPSC lines were comprised of a mixed population of piPSCs from the lentiviral reprogramming. The cells were harvested at passage 10, and lysed in 350 mL RLT buffer (Qiagen, Hilden, Germany) containing b-mercaptoethanol (SigmaÀ Aldrich). Total RNA was extracted using the RNAeasy Micro Kit (Qiagen), according to the manufacturer's instructions. First-strand cDNA synthesis was performed using a RevertAid TM First Strand cDNA Synthesis Kit (Fermentas), according to the manufacturer's instructions. One microgram of total RNA was used as a template for cDNA synthesis. These cDNA templates were used for subsequent quantitative PCR analysis, at a 1:10 dilution.

RNA Extraction, cDNA Synthesis, and Quantitative Real-Time PCR
Quantitative PCR was performed using SYBR Green I Master Kit (F-Hoffman La Roche Holdning AG, Basel, Switzerland), according to the manufacturer's instructions, using a Light Cycler 480 System (Roche). All samples were run in triplicate on 96-well optical PCR plates (Roche Diagnostics). Specific qPCR primers were designed for OCT4, SOX2, c-MYC, LIN28, NANOG, FGFR2 IIIC, FGFR1 IIIC, LIFR, STAT3, NROB1, KLF4, and GAPDH (Glyceraldehyde 3-phosphate dehydrogenase) ( Table 2). After an initial denaturation for 10 min at 958C, amplification was conducted for 40 cycles of 15 sec at 958C, 15 sec at 608C, and 30 sec at 728C. The DDC T method (Livak and Schmittgen, 2001) was used to analyse the qPCR data. mRNA abundances were normalized to the reference gene GAPDH. The specificity of each assay was assessed by the melting-curve analysis, where a single peak was observed. Average of each sample group AE standard error of the mean was plotted (see Fig. 1).

Parthenogenetic Activation for In Vitro Embryo Production
Ovaries from gilts (aged 5À7 months) were collected, and the contents of follicles with a diameter of between two and 9 mm were aspirated into a 50-mL tube using a 16-guage needle and a vacuum pump (pmax À2.4 bar). Follicle fluid was allowed to sediment for 10 min, and then the supernatant was removed and the pellet was washed twice with TCM199 (SigmaÀAldrich) with 50 mg/mL gentamycin (SigmaÀAldrich), 1 mM L-glutamine (SigmaÀAldrich), 0.5 mg/mL polyvinylalcohol (SigmaÀAldrich), and 20 U/mL heparin (DAK). The washed pellet was dissolved in TCM199, divided into three 9-cm Petri dishes, and examined for cumulus-oocyte complexes with a homogenous ooplasm and more than three layers of cumulus cells. The cumulusoocyte complexes were washed once in IVM medium (TCM 199, 1 mg/mL polyvinyl alcohol, 300 mg/mL sodium hydrogen carbonate, 10 mg/mL gentamycin, 0.55 mg/mL D-glucose, 0.01 mg/mL sodium pyruvate, 0.07 mg/mL cysteine, 40 ie/mL Suigonan [Intervet], and 50 ng/mL Epidermal growth factor [SigmaÀAldrich]) (Abeydeera et al., 2000), and matured in vitro for 44 hr at 38.58C and 5% CO 2 .

Production of Chimeric Embryos for In Vitro Experiments
Chimeras were made with LIF piPSCs and FGF piPSCs as well as with parental porcine neonatal fibroblasts. piPSCs with low passage number (<10) were incubated for 5 min with TrypLE 1X (Gibco), and then neutralised with piPSC medium. Cells were spun down (200 g for 5 min), resuspended in KSR medium, transferred to 0.1 M gelatincoated 3-cm Petri dishes, and incubated for 30 min to remove the mouse embryonic fibroblasts by differential plating. The parental neonatal fibroblasts were harvested by incubation with Trypsin-EDTA (SigmaÀAldrich) for 5 min, and then neutralised with complete DMEM AQ. Cells were pelleted (200 g) and resuspended in complete DMEM AQ. Thirty microliters of the piPSCs or parental neonatal fibroblasts suspensions were added to a 50-mL drop of piPSC medium, without LIF or FGF, on a micromanipulation dish covered with mineral oil.
Five days after parthenogenic embryo activation, a total of 15 Venus piPSCs were injected into the blastocoel cavity of each embryo (n ¼ 10 total per drop) using the Zeiss Axiovert 200 and TransferMan NK2 micromanipulator setup from Eppendorf. A holding pipette with 120-mm outer diameter/25-mm inner diameter, and an injection pipette with 20-mm outer diameter/17-mm inner diameter was used (both from The Pipette Company). Following injection, each embryo was washed twice in PZM with 10% FBS, and cultured in either PZM with 10% FBS or PZM with 10% FBS and 2 mg/mL doxycycline (SigmaÀAldrich) for 2 days in either an incubator or a NIKON Biostation IM at 5% CO 2 , 5% O 2 and 90% N 2 at 38.58C. The Nikon Biostation IM allowed for fluorescence time-lapse studies of the potential chimeric embryos. Images were captured every 30 min for 48 hr by phase contrast and fluorescence (EX: 465-495; DM505; BA515-555], 10Â, 20Â, and z-stack function). Images were saved and analysed for florescence and embryo development. After 48 hr, embryos were fixed for 20 min at room temperature in 4% paraformaldehyde, and stored in PBS with 1% BSA at 48C.
Porcine Specific Primer Used for PCR on Genomic DNA

Statistical Analysis of Biostation Results
The number of hatched embryos with fluorescence (a marker of the injected cells) and number of hatched embryos (a metric of embryo survival) were counted for the following groups: parental neonatal fibroblasts in PZM (n ¼ 43), LIF piPSCs in PZM (n ¼ 45), LIF piPSCs in PZM with doxycycline (n ¼ 50), FGF piPSCs in PZM (n ¼ 45), and FGF piPSCs with doxycycline (n ¼ 43). The results were analysed in SAS interprize Guide 5.1 using proc freq for tables and proc logistic for logistic regression. Proc logistic was chosen as the data is categorical and has more than one explaining variable (cell type, AE doxycycline, passage number). Differences between cell lines were described using odds ratio, which can be interpreted as difference in odds for finding a hatched fluorescent embryo between two groups.

Differential Staining of Chimeric Embryos
Staining was performed as previously described . In brief, the embryos were permeabilized in PBS with 1% Triton X-100 (Merck, Kenilworth, NJ). The zona pellucida was removed in PBS with 2N HCl, then the embryos were cultured in 1 mM Triz-HCl (SigmaÀ Aldrich) and blocked in PBS containing 2% BSA (Sigma-Aldrich). Trophectoderm was stained for with anti-CDX2 ready-to-use antibody (undiluted ab86949) (Abcam) and Alexa Fluor 594-conjugated goat-antimouse secondary antibody (Invitrogen). Embryos were then counterstained with 10 mg/mL Hoechst (SigmaÀ Aldrich). Finally, embryos were mounted on a coverslip using 0.1% low melting-point agarose (Fisher Bioreagents), mounted on a slide using DAKO Fluorescent Mounting Medium (DAKO), and stored overnight at 48C. The mounted embryos were analyzed at the Core Facility for Integrated Microscopy at University of Copenhagen using the Zeiss CellObserver Spinning Disk microscope with the iXon3 EMCCD camera from Andor and Zeiss Zen Blue 2012 software.

Analyses of Chimeric Embryos by Transmission Electron Microscopy
Chimeric embryos were collected at either 2 or 48 hr post-injection of Venus piPSCs. Embryos were fixed at 48C for 1 hr in 3% glutaraldehyde in 0.1 M Na-phosphate buffer (pH 7.2), and then washed in the same buffer, post-fixed in 1% OsO 4 in 0.1 M sodium-phosphate buffer, dehydrated, embedded in Epon, and serially sectioned into semi-thin sections (2 mm). The sections were stained with basic toluidine blue, and evaluated by brightfield light microscopy. Selected semi-thin sections were re-embedded as described earlier (Hyttel and Madsen, 1987), and processed for ultra-thin sectioning (50À70 nm). The ultra-thin sections were treated with uranyl acetate and lead citrate, and examined on a Philips CM100 transmission electron microscope (Darmstadt, The Netherlands).

Teratoma Formation
General animal welfare guidance was followed, and the teratoma experiment was approved by the veterinary authority at Department of Experimental Medicine under the protocol number P 13À210. Female NOD/SCID mice (NOD/MrkBomTac-Prkdc<scid> female, Lot 20130624-EBU020601C-HC-M) were purchased from Taconic (Silkeborg, Denmark), and housed in groups of no more than eight at the Department for Experimental Medicine (Panum Institute, Copenhagen, Denmark).
LIF piPSCs were harvested using 1Â TrypLE (Gibco), counted, spun down, resuspended in PBS with 1% BSA (SigmaÀAldrich), and aliquotted at 1.5 Â 10 6 cells per 250 mL. The cells were injected subcutaneously into the left flank of the mice using a 1-mL insulin syringe and 29-gauge needle (Terumo Corporation, Shibuya, Tokyo). The mice were monitored for up to 3 months for tumor formation.
Two teratoma experiments were conducted: For the first experiment, six mice were injected with LIF piPSCs and four with PBS as a control. The mice received no doxycycline throughout the duration of the experiment. The mice were monitored for 4 weeks without tumor development. For the second experiment, six mice were injected with LIF piPSCs, six mice were injected with FGF piPSCs, and six mice were injected with parental neonatal fibroblasts. For the first 7 weeks, the mice received 2 mg/mL doxycycline (SigmaÀ Aldrich) dissolved in drinking water. Tumors were detectable in the mice injected with LIF piPSC after 4 weeks, and after seven weeks the largest tumor had reached a diameter of close to 10 mm and doxycycline administration was seponated. Three days later, the tumors began to decrease in size; 1 week after doxycycline withdrawal, the tumor had decreased to half the size, at which point the mice were euthanized and the tumors fixed overnight in 4% paraformaldehyde, and embedded in OCT compound (Sakura FineTek). The tumors were sectioned (5 mm) on a cryostat, and sections were subjected to hematoxylin and eosin staining or immunohistochemistry with antibodies for the lineage-specific markers alpha fetoprotein (endoderm), beta-3 tubulin (ectoderm), and smooth muscle antigen (mesoderm) (see Immunofluorescence Analysis).

Production of Chimeric Embryos for In Vivo Experiments
Four Danish Landrace sows were inseminated 4 and 5 days after weaning, and slaughtered 4 days after the last insemination. The uteri were removed and transported in styrophor boxes maintained at 388C. The uteri were flushed with PBS (Calbiochem) with 5% FBS (SigmaÀAldrich), and the medium was screened under a Leica M80 stereomicroscope. Embryos were collected and transferred to PZM with 10% FBS. A total of 48 blastocysts were injected with LIF piPSCs below passage 10. After injection, embryos were screened for fluorescence, and 44 embryos with fluorescent cells were cultured overnight at 5% CO 2 , 5% O 2 , and 90% N 2 at 38.58C. The following day, the 44 Molecular Reproduction & Development embryos were transferred to 14-mL Falcon tubes and transported to Aarhus University, where they were transferred to a surrogate sow. Embryos were transported in a G95 portable incubator from K-system capable of maintaining 5% CO 2 , 5% O 2 , and 90% N 2 at 38.58C for 4 hr.
Transfer to the surrogate sow occurred 4.5 days postheat, and was performed via flank incision. The embryos were injected into the lumen of the uterine horn near the isthmus in both horns, as described previously (Schmidt et al., 2011). The surrogate sow was slaughtered on Day 32 of the pregnancy. The uterus was immediately removed, and the embryos and embryonic membranes were isolated. A total of 16 embryos were harvested, and samples for fluorescence-activated cell sorting, immunohistochemistry, and transcriptome analyses were collected.
Additional transfers were made in which 110 blastocysts injected with LIF piPSCs (below passage 10) were cultured with 2 mg/mL doxycycline (SigmaÀAldrich), and transferred to three sows fed 14 mg doxycycline daily (ScanVet Animal Health A/S) from 2 days before to 4 days after the transfer. Embryos, but not embryonic membranes, were collected at Day 32 for immunohistochemistry, transcriptome, and genome analyses.

RNA-Seq mRNA Preparation and Bioinformatics Analysis
mRNA preparation, cDNA synthesis, genome alignment, and transcript quantification for LIF piPSCs, FGF piPSCs, and parental neonatal fibroblasts were performed by Beijing Genomics Institute (BGI), Shenzhen, Guangdong province, China as well as BGI-Europe and Aarhus University, Denmark. The mRNA preparation and cDNA synthesis for the embryonic samplesÀ ÀDay 7/8 inner cell mass (n ¼ 10), Day 10/11 epiblast (n ¼ 10), and Day 12/13 gastrulating epiblast (n ¼ 5)À Àwere performed by the Department of Molecular Biology, Faculty of Science, Nijmegen Centre for Molecular Life Sciences, NCMLS, Radboud University, Nijmegen, The Netherlands. The inner cell mass of Day 7/8 embryos were isolated as described in Telugu et al. (2011); the epiblast and gastrulating epiblast were manually isolated. Datasets can be access through the Gene Expression Omnibus (GEO) database (www.ncbi.nlm.nih.gov/geo/), using Series record GSE92889.
Genome alignment and transcript quantification were conducted at the Department of Veterinary Clinical and Animal Sciences, Faculty of Health and Medical Sciences, University of Copenhagen, Denmark. Further bioinformatics analysis was performed using both data sets after log transformation and data cross-normalization. The mRNA preparation and cDNA synthesis as well as up-and downstream bioinformatics and analysis are described in detail in the Supplementary Information.