Inhibition of protein kinase D by CID755673 promotes maintenance of the pluripotency of embryonic stem cells

Since Evans and Kaufman isolated mouse embryonic stem cells (mESCs) in 1981 (Evans and Kaufman, 1981; Martin, 1981), enormous progress has been made in our understanding of many aspects of mESCs. Initially, only ESCs derived from the mouse strain 129 could be maintained using various empirical combinations of feeder cells in serum-containing medium (Evans and Kaufman, 1981; Martin, 1981). In 1988, researchers discovered that leukemia inhibitory factor (LIF) could replace feeder cells to support mESC identity (Smith et al., 1988; Williams et al., 1988). Later, it was demonstrated that bone morphogenetic protein 4 (BMP4) could replace serum to maintain the mESC self-renewal state in the serum-free medium N2B27 (Ying et al., 2003). However, the use of animal-derived cytokines is not conducive to the future application of stem cells. In 2008, Ying et al. (2008) discovered that the application of two small molecules, PD0325901 (PD) and CHIR99021 (CHIR) (also named 2i), inhibitors of mitogen-activated protein kinase kinase (MEK) and glycogen synthase kinase 3 (Gsk3), respectively, could maintain the stem cell phenotype better than the previously used culture conditions in N2B27 medium (Ying et al., 2008); they also derived rat ESCs based on 2i treatment conditions (Buehr et al., 2008; Li et al., 2008). Although human ESCs (hESCs) were established in 1998, they required distinct culture conditions for the maintenance of their pluripotent state (Thomson et al., 1998). Interestingly, hESCs are similar to mouse postimplantation epiblast-derived stem cells (EpiSCs) in their growth requirements, morphology, clonogenicity and gene expression patterns (Brons et al., 2007; Tesar et al., 2007). However, to date, only mouse and rat ESCs have been shown to have the ability to contribute to the formation of germline-competent chimeras (Buehr et al., 2008; Evans and Kaufman, 1981; Li et al., 2008; Martin, 1981). These stem cells are thus considered to be in a naïve pluripotent state, whereas mouse EpiSCs and hESCs are in a ‘primed’ pluripotent state (Huang et al., 2015). Researchers have been working hard to develop a suitable medium for the growth of ESCs for species other than mice and rats.

In addition to different culture conditions, multiple signaling pathways and genes that regulate ESC pluripotency have been identified. For example, activation of the LIF/STAT3 and Wnt/β-catenin signaling pathways, and inhibition of the FGF/MEK/ERK signaling pathway can promote the maintenance of naïve pluripotency, whereas hESCs and mouse EpiSCs require the FGF/MEK/ERK signaling pathway to maintain the primed pluripotent state. In addition, numerous targets of these signaling pathways have also been identified, such as Tfcp2l1, Gbx2, Pim1, Sp5, Myc, Klf4, Esrrb, Klf2 and Nanog (Martello et al., 2013, 2012; Tai and Ying, 2013; Ye et al., 2013, 2016), and overexpression of some of these genes, such as Nanog and Tfcp2l1, can promote mouse and human ESC self-renewal (Chambers et al., 2003; Sun et al., 2018; Vallier et al., 2009). Moreover, naïve and primed pluripotent states share common features, as all of them have a common subset of transcription factors (Oct4, Sox and Nanog) that make up the core pluripotency circuitry (Huang et al., 2015). Together, these data indicate that naïve and prime ESCs have a conserved mechanism that controls pluripotency.

In this study, we show that the small molecule CID755673 (CID), a selective inhibitor of protein kinase D (PKD) enzymes, can promote the short-term self-renewal of mESCs and hESCs. Furthermore, we discovered that CID can support the long-term undifferentiated state of mESCs in combination with another small molecule, PD, a specific inhibitor of MEK. Additionally, we observed that CID promotes mouse and human ESC self-renewal, in part through the PI3K/AKT signaling pathway. Thus, we describe a novel and conserved mechanism that promotes the maintenance of ESC self-renewal and pluripotency.

To identify novel compounds that could promote mESC self-renewal, we screened 48 small molecules, most of which have not been tested before. mESCs of the 46C line were cultured in serum-containing media and treated with individual compounds (Fig. S1). After 8 days, both LIF- and CID-treated mESCs showed undifferentiated morphology (Fig. 1A), whereas the cells treated with the remaining compounds differentiated (Fig. S1). The addition of LIF or CID maintained high alkaline phosphatase (AP) activity in mESCs (Fig. 1A). Simultaneously, immunofluorescence staining results showed positive expression of the pluripotency marker Oct4 in CID-treated cells (Fig. 1B). The qRT-PCR and western blot results also showed that CID-treated cells expressed higher levels of the pluripotency genes Oct4, Nanog and Tfcp2l1, while exhibiting low levels of the differentiation-associated genes Gata4 and Foxa2 compared with that observed in the untreated groups (Fig. 1C,D). Next, to test whether CID maintains mESC self-renewal in a dose-dependent manner, different concentrations of CID were supplemented in serum-containing medium. After 8 days, we observed that in concentrations above 5 µM CID, AP-positive stem cell colonies could develop (Fig. 1E), whereas 20 µM CID caused growth arrest and cell death (Fig. 1E). Thus, 10 µM CID was used in subsequent experiments. To examine whether CID is sufficient to support long-term self-renewal, CID-treated mESCs were passaged. As shown in Fig. 1F, CID-treated mESCs gradually lost AP activity and became flat, indicating that CID treatment alone allows for the short-term self-renewal of mESCs.

Next, we replicated these experiments in another mESC line derived from C57BL/6 mice. Unlike 46C mESCs isolated from 129 strain mice, C57BL/6 mESCs could not be maintained by LIF alone under feeder cell-free conditions (Ye et al., 2012). To determine whether the addition of CID can enhance the self-renewal ability of C57BL/6 mESCs, we designed two groups of experiments. After two passages, the no treatment control mESCs seeded on feeder cells died or differentiated, whereas the addition of LIF or CID could maintain the undifferentiated state of mESCs, as determined by their compact colony morphology and AP activity (Fig. S2A). Under feeder cell-free conditions, mESCs treated with LIF or CID lost AP activity (Fig. S2A). However, CID was capable of working synergistically with LIF to promote the maintenance of the stemness of C57BL/6 ESCs (Fig. S2A). Overall, these data suggest that CID has the ability to promote mESC self-renewal in vitro.

CHIR and PD are two important compounds with self-renewal-promoting effects (Ying et al., 2008). To investigate whether they can contribute to the function of CID, 46C mESCs were cultured in CID/serum-containing media in the presence of CHIR or PD. As shown in Fig. 2A, the dual administration of CID and PD (named PC), but not CID and CHIR, allowed for the long-term maintenance of undifferentiated mESCs in serum-containing or serum-free medium, although the colonies were not compact, similar to those formed under 2i treatment conditions (Fig. 2A, Fig. S3). mESCs cultured in PC could be routinely passaged by single-cell dissociation and remained positive for AP staining (Fig. 2B). Additionally, immunofluorescence staining and qRT-PCR results revealed that they expressed higher levels of the pluripotency markers Oct4, Sox2, Nanog and Tfcp2l1 in the presence of PC than in the presence of medium without treatment (Fig. 2C,D). To examine whether the function of CID depends on the presence of STAT3, we first assessed the total and phosphorylated protein levels of STAT3 in the presence or absence of CID, and the results showed no difference (Fig. S4A). Subsequently, we decreased STAT3 protein in 46C mESCs with STAT3 short hairpin RNA (shRNA) lentivirus (Fig. S4B). As shown in Fig. S4C, downregulation of STAT3 induced differentiation in mESCs in the presence of LIF, and these cells retained positive AP activity when cultured in PC-containing medium, although they grow slightly slower than the control cells (Fig. S4C). Collectively, these results suggest that PC is able to promote the maintenance of the undifferentiated state in a STAT3-independent manner.

To determine the differentiation capacity of mESCs, these cells were maintained in PC/serum-containing medium for ten passages and then grown in suspension to form embryoid bodies (EBs). PC-treated mESCs retained the ability to differentiate into Tuj1⁺ neurons, myosin⁺ myocardial cells and Gata4⁺ primitive endoderm cells in vitro (Fig. 2E). In addition, we evaluated the ability of PC-treated ESCs to be used to generate chimeric mice. C57BL/6 mESCs were cultured on feeder cells in serum-containing or serum-free medium supplemented with PC for ten passages. These mESCs exhibited high Oct4 expression levels (Fig. S2B) and were injected into 60 3.5 days post coitum (dpc) blastocysts. Next, these injected blastocysts were put into four pseudopregnant female mice. Finally, four live pups were born, one of which was a chimera male pup that showed germline transmission (Fig. 2F). Taken together, these data indicate that the mESCs maintained in PC-containing medium retain pluripotency.

CID is a selective inhibitor of PKD (Sharlow et al., 2008). Therefore, we sought to investigate whether the suppression of PKD gene expression can mimic the effect of CID in promoting mESC self-renewal. The PKD family comprises three members: PKD1, PKD2 and PKD3 (Valverde et al., 1994). We designed shRNAs to inactive mouse Pkd1, Pkd2 and Pkd3 expression in 46C mESCs (Fig. 3A-C). Compared with the scramble control cells, stable knockdown of PKD transcript levels was observed following drug selection in 46C mESCs infected with lentiviruses encoding shRNAs specific for PKD genes (Fig. 3A-C). All PKD shRNA-infected cells remained undifferentiated in the LIF/serum-containing medium but differentiated after ceasing LIF treatment (Fig. 3D). To rule out the possibility that residual expression of the PKD gene was the cause of the phenotype, we designed a gene-targeting vector to knockout (KO) PKD gene expression in 46C mESCs through CRISPR/Cas9-mediated DNA double-strand breaks. After gene transfection and selection, the disruption of PKD1 alleles was confirmed by genomic DNA sequencing and western blotting (Fig. 3E,F). Unfortunately, the PKD1 gene KO mESCs differentiated when grown in serum-containing medium (Fig. 3G). As PKD1, PKD2 and PKD3 share high sequence homology (Hayashi et al., 1999; Johannes et al., 1994; Sturany et al., 2001; Valverde et al., 1994), they may have cross-compensatory functionality in mESCs. Moreover, CID inhibits all three PKD isoforms with a similar potency (Sharlow et al., 2008). These results prompted us to deplete the PKD3 gene in PKD1 KO mESCs (Fig. 3E,F). Similar to PKD1 KO mESCs, PKD1/PKD3 double KO cells became flat in serum-containing medium without LIF (Fig. 3G). Finally, we established a PKD1, PKD2 and PKD3 triple gene KO cell line by disrupting the PKD2 gene in PKD1/PKD3 KO mESCs (Fig. 3E,F). As expected, depletion of the three PKD members induced the generation of many undifferentiated colonies in the absence of LIF (Fig. 3H). In line with this phenomenon, the triple KO mESCs formed much more compact colonies than the wild-type control cells in the presence of PD (Fig. 3H). Taken together, these data suggest that inhibition of PKD family members can mimic the self-renewal-promoting effect of CID.

To observe the location of PKD proteins in the undifferentiated and differentiated cells, we generated four mESC lines that overexpressed empty vector or FLAG-tagged Pkd1, Pkd2 and Pkd3 using PiggyBac vector (PB, PB-PKD1, PB-PKD2 and PB-PKD3), in which the expression of individual PKD genes was efficiently enhanced (Fig. 4A). These cells were cultured in the presence or absence of LIF for 48 h, after which immunofluorescence staining was performed with an antibody against FLAG. As shown in Fig. 4B, the PKD genes were primarily located in the cytoplasm regardless of the self-renewal or differentiation conditions (Fig. 4B). To assay their expression levels in mESCs and during EB differentiation. mESCs were cultured in suspension without LIF and aggregated to form EBs, in which differentiation proceeds into the three germ layers. qRT-PCR results showed that the expression of PKD genes was low in undifferentiated mESCs maintained in LIF/serum-containing medium, but increased when mESCs were transferred to culture medium for differentiation (Fig. 4C). In contrast, the expression pattern of Nanog, a pluripotency marker, was opposite to that of PKD (Fig. 4C). Furthermore, we noticed that the expression levels of many protein kinase C (PKC) family genes, which are closely associated with PKD activity (Brändlin et al., 2002; Li et al., 2004; Xu et al., 2015; Zugaza et al., 1996), also showed a similar expression pattern with PKD (Fig. 4D). Altogether, these findings suggest that upregulation of PKD gene expression might trigger the initiation of mESC differentiation. To further investigate whether upregulation of PKD genes leads to mESC differentiation, mESCs overexpressing PKD genes were passaged for an extended period of time in LIF/serum-containing medium. We noticed that many PB-PKD1 colonies became flat before passaging, whereas the empty vector PB, PB-PKD2 and PB-PKD3 mESCs grew normally (Fig. S5A). However, after three passages, all of these cells retained their AP⁺ activity (Fig. S5A), and the undifferentiated phenotype was not due to the loss of exogenous PKD gene expression (Fig. S5B). These results suggest that PKD can induce a differentiation signal, but this signal is still lower than the self-renewal signal maintained by LIF.

To further explore the mechanism by which CID promotes mESC self-renewal, we analyzed PC-treated mESCs by RNA-Seq to assess the gene expression pattern. Compared with the PD treatment, cells treated with PC were induced for many genes with different expression levels (Fig. 5A). Among these genes, 613 genes were upregulated and 980 were downregulated by 50% or more. The gene ontogeny and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis results showed that many genes are associated with cell adhesion, development processes and metabolism (Fig. S6). We further evaluated the KEGG results and identified 14 candidates, including Otx1, Wnt11, Wnt16, Dusp9, Pik3r3, Id2, Apc2, Wnt6, Myc, Wnt5a, Wnt8b, Inhbc, Lefty2 and Hesx1, which were enriched in signaling pathways regulating the pluripotency of stem cells (Fig. 5B). We focused on genes that were upregulated by PC and validated their expression using qRT-PCR, with the results showing that only the expression patterns of Dusp9, Pik3r3 and Id2 were similar to those observed in the RNA-Seq results (Fig. 5C). Finally, we chose Pik3r3 and Id2 for further validation but not Dusp9 because it can steadily attenuate MEK/ERK activity and PD is already a MEK-specific inhibitor (Li et al., 2012; Ying et al., 2008). To confirm the regulatory relationship among PKD members, Pik3r3 and Id2, we detected the expression levels of Pik3r3 and Id2 in mESCs overexpressing Pkd1, Pkd2 and Pkd3. As expected, all three PKD genes were able to repress the expression of Pik3r3 and Id2 when upregulated (Fig. 5D), implying that Pik3r3 and Id2 might be able to mediate the self-renewal-promoting function of CID. We first tested the role of Id2, as it has been previously reported to promote mESC self-renewal (Guo et al., 2017; Ying et al., 2003) . FLAG-tagged Id2 was inserted into the PB vector and then transfected into 46C mESCs (PB-Id2) (Fig. S7A). Unfortunately, both PB and PB-Id2 mESCs differentiated when transferred into PD-containing medium (Fig. S7B). In addition, knockdown of Id2 did not impair the function of PC (Fig. S7C,D). We then turned our attention to the second candidate, Pik3r3, which could increase the phosphorylation of AKT at site 473 (pAKT) when overexpressed (Fig. 5E), whereas Pik3r3 knockdown reduced AKT activity (Fig. 5F,G). These results are consistent with those observed in tumors (Cao et al., 2015; Yu et al., 2015). Therefore, we investigated whether CID can affect the activation of the PI3K/AKT pathways. The levels of total and phosphorylated AKT were determined in 46C mESCs by western blot analysis. The results showed that the addition of PC increased the phosphorylation of AKT (Fig. 5H). To investigate whether inhibition of the PI3K/AKT signaling pathway can impair the ability of CID to promote mESC self-renewal, we used a specific small molecule inhibitor of the PI3K/AKT signaling pathway, LY294002 (LY294) (Paling et al., 2004). Western blotting results showed that LY294 could significantly repress the level of pAKT in mESCs (Fig. 5H). Moreover, after two passages, we observed that LY294 impaired the self-renewal-promoting effect of CID (Fig. 5I). Although the addition of LY294 did not significantly reduce the number of AP-positive stem cell colonies maintained by PC, some colonies differentiated at the edges (Fig. 5I). These findings collectively suggest that the PI3K/AKT signaling pathway can partially mediate the ability of CID to promote mESC self-renewal.

As hESCs also share the same common features as mESCs (Huang et al., 2015), we examined whether the mechanism by which CID promotes mESC self-renewal is conserved in hESC maintenance. HES2 hESCs were cultured in basal medium in the absence or presence of CID. Before passaging, CID was able to maintain hESCs in an undifferentiated state (Fig. 6A), with these cells expressing high levels of the pluripotent markers, OCT4, NANOG and PRDM14 (Fig. 6B), whereas the untreated control cells began to differentiate (Fig. 6A). Notably, hESCs cultured in CID-containing medium could not be split for two more passages and lost AP activity (Fig. 6A). Interestingly, we observed that another PKD inhibitor, CRT0066101 (CRT), had a more powerful effect on the maintenance of hESCs (Fig. 6A). hESCs cultured in CRT-containing medium expressed higher levels of OCT4, NANOG and PRDM14 than hESCs treated with CID (Fig. 6B), and retained AP⁺ activity, even after four passages (Fig. 6A). Moreover, after two passages, the CRT-treated hESCs retained the ability to differentiate into three germ layers (Fig. S8). Therefore, inhibition of PKD supports short-term maintenance of the undifferentiation of hESCs. Next, we tested the function of PKD members ID2 and PI3K/AKT signaling on hESC maintenance. First, we efficiently knocked down human PKD1, PKD2 and PKD3 gene expression in hESCs (Fig. S9A-C), and observed that these cells could not be maintained in an undifferentiated state after the cessation of treatment with activin A and bFGF, two important cytokines for hESC maintenance (Beattie et al., 2005; Vallier et al., 2005) (Fig. S9D). Subsequently, we established PKD1, PKD2 and PKD3 triple-knockdown hESCs and cultured them in basal medium (Fig. 6C). After 6 days, many undifferentiated clones were still observed in the basal medium but they differentiated after 15 days (Fig. 6D), indicating that inhibition of PKD can phenocopy the function of CID in sustaining hESCs in an undifferentiated state for a short period of time. Meanwhile, we found that the expression levels of PKD1 and PKD2, but not PKD3, gradually increased upon hESC differentiation (Fig. S10A,B). However, enforced expression of PKD1, PKD2 or PKD2 had no obvious negative effects on the maintenance of the hESC identity (Fig. S9E). Second, we overexpressed the ID2 gene in hESCs using the PiggyBac system (PB-ID2), and, as observed in mESCs, PB and PB-ID2 hESCs differentiated (Fig. S11A,B). Finally, to examine the effect of the PI3K/AKT signaling pathway on the function of CID and CRT in hESCs, we treated cells with LY294 and found that it greatly inhibited the self-renewal-promoting effects of CID and CRT (Fig. 6E,F). Collectively, these data indicate that inhibition of PKD can also promote hESC self-renewal and relies on the PI3K/AKT signaling pathway.

The results of our study demonstrate that the small molecule CID can promote mESC and hESC self-renewal. CID primarily functions by inhibiting three PKD members. Therefore, the combined inhibition of the activity of PKD1, PKD2 and PKD3 can largely recapitulate the self-renewal-promoting effects of CID. Moreover, the combination of PD and CID is able to maintain the self-renewal and pluripotency of mESCs. Mechanistically, CID induces the activation of AKT. Therefore, suppression of AKT activity partially impairs the function of CID in mESCs and hESCs. Collectively, these results reveal a conserved and novel mechanism mediated by PKD inhibition in sustaining mESC and hESC stemness.

The PKD family comprises three closely related Ser/Thr kinases (PKD1, PKD2 and PKD3) that have two diacylglycerol (DAG)-binding C1 domains, which are homologous to the DAG-binding domain of PKC members. Because of this feature, PKD genes were first classified as members of the PKC family (Johannes et al., 1994). A previous study demonstrated that the inhibition of PKC signaling is sufficient to maintain mESC self-renewal and pluripotency (Dutta et al., 2011). The PKC family consists of several serine/threonine protein kinases that are divided into three subfamilies (Dutta et al., 2011; Spitaler and Cantrell, 2004): (1) classical PKCs (isoforms α, β1, β2 and γ; calcium and phospholipid dependent); (2) novel PKCs (isoforms δ, ε, η and θ; calcium independent and phospholipid dependent); and (3) atypical PKCs (isoforms ξ and ι/λ). However, only depletion of PKCξ is able to reproduce the PKC inhibitor-mediated effect of promoting mESC self-renewal (Dutta et al., 2011), indicating that the function of PKC family members is not complementary in the negative regulation of stem cell pluripotency. Although PKD inhibition can also promote mESC self-renewal (Fig. 1A-E, Fig. 2A-D), unlike PKC, only the simultaneous inhibition of PKD1, PKD2 and PKD3 can simulate the function of CID (Fig. 3E-H, Fig. 6C-F), indicating that PKD isoforms are functionally redundant in regulating mESC and hESC pluripotency. However, it is worth noting that the expression of PKD might be closely associated with the expression of PKC genes (Fig. 4D). The similarities and differences between PKD and PKC functions are understandable, as later studies revealed that PKD has mixed features of different subclasses of the PKC family such that it does not belong to any one of them. For example, PKD lacks an autoinhibitory pseudosubstrate region, a characteristic feature of PKC, and possesses a pleckstrin homology domain, which is absent in PKC. In addition, the catalytic domain is different from that of PKCs and shows greater homology with that of calcium/calmodulin-dependent kinases (CaMKs) (Rozengurt et al., 1995; Van Lint et al., 1995). Thus, it will be interesting to investigate whether inhibition of CaMK impacts the maintenance of ESCs in the future.

Another significant difference between PKC and PKD is that PKC inhibition mediates mESC self-renewal and is associated with the inactivation of NF-κB transcription but not the activation of the PI3K/AKT pathway (Dutta et al., 2011). As a result, PI3K inhibition does not prevent the PKC inhibitor-mediated maintenance of mESC self-renewal (Paling et al., 2004). In contrast, AKT phosphorylation is associated with the activity of the PKD inhibitor CID, which mediates the maintenance of mESC self-renewal (Fig. 5H,I). Therefore, blocking PI3K/AKT signaling partially diminishes the self-renewal-promoting effects of PKD inhibition (Fig. 5I, Fig 6E,F). These results are unsurprising, because the PI3K/AKT signaling pathway has been implicated in regulating mESC and ESC pluripotency. In mESCs, PI3K/AKT signaling regulates many self-renewal associated genes, such as Nanog, Ptpn6 and Zscan4c (Storm et al., 2007, 2009). PI3K/AKT signaling is thus required for optimal maintenance of mESCs in their undifferentiated state (Paling et al., 2004; Watanabe et al., 2006), whereas treatment with the PI3K inhibitor leads to a differentiated morphology in mESCs (Paling et al., 2004). Likewise, in hESCs, the levels of PI3K/AKT pathway components decreased upon differentiation (Armstrong et al., 2006). Disruption of this pathway eventually results in differentiation in the middle of hESC colonies (Armstrong et al., 2006; Singh et al., 2012). In contrast, activation of the PI3K/AKT pathway can enhance the survival activity of hESCs (Pyle et al., 2006). Additionally, a stable hESC line expressing a myristoylated and constitutively active version of AKT is capable of maintaining the levels of pluripotency markers for a time (Singh et al., 2012). Nonetheless, future studies are needed to elucidate additional key components downstream of PKD inhibitors, because the inhibition of AKT is not sufficient to completely remove the ability of CID to promote ESC self-renewal (Fig. 5I, Fig. 6E,F).

In summary, our data show that inhibition of PKD in mESCs and hESCs promotes the maintenance of an undifferentiated state in these cells. Particularly with regard to hESC maintenance, which is poorly understood compared with mESC self-renewal, a better understanding of the mechanisms associated with PKD, and its interplay with other signaling pathways and genes, will improve the culture and application of hESCs in the future. Furthermore, given the functional and molecular similarities and differences between mESCs and hESCs, treatment with a PKD inhibitor may allow for the isolation and maintenance of pluripotent ESCs from other species.

The 46C mESCs, kindly provided by Qi-Long Ying (University of Southern California, Los Angeles, CA, USA), were cultured in 0.1% gelatin-coated dishes at 37°C in an atmosphere with 5% carbon dioxide. Cells were routinely maintained in Dulbecco's modified eagle medium (DMEM) (TransGen Biotech) supplemented with 10% fetal bovine serum (ExCell Bio, FND500), 1× MEM non-essential amino acids (Invitrogen, 11140-050), 0.1 mM β-mercaptoethanol (Sigma-Aldrich, M3148) and 1000 U/ml LIF (Millipore, LIF1010). HES2 hESC lines were kindly provided by Qi-Long Ying. The basal medium used to culture hESCs was N2B27 plus 10% KnockOut serum replacement (KSR) (Life Technologies, 10828). N2B27 consisted of the following ingredients: DMEM/F12 (Life Technologies, 11330-032) and Neurobasal medium (Life Technologies, 21103-049) mixed at a 1:1 ratio; 1×N2 (Life Technologies, 17502-048); 1×B27 (17504-044, Life Technologies); 2 mM GlutaMAX (Life Technologies, 35050-061); 1×non-essential amino acids solution (Life Technologies, 11140-050); and 0.1 mM β-mercaptoethanol (Gibco, 21985). HES2 hESCs were cultured on plates precoated with Matrigel (BD Biosciences) without feeder cells in basal medium supplemented with 10 ng/ml FGF2 and 10 ng/ml activin A. Y27632 (1 µM; Sigma-Aldrich, Y0503) was added when hESCs were passaged.

The coding regions of the mouse genes Id2, Pik3r3, Pkd1, Pkd2 and Pkd3 were inserted into PiggyBac transposon vectors. shRNA constructs were designed to target the gene-specific regions of mouse or human PKD1, PKD2, PKD3 and ID2, and were then cloned into pLKO.1-TRC (Addgene, #10878). The target sequences are listed in Table S1.

Cells were fixed in 4% paraformaldehyde for 2 min at room temperature, washed in PBS and incubated in AP staining reagent (Beyotime Biotechnology, C3206) for 30 min at room temperature in the darkroom. Then, after two washes with PBS, the cells were visualized using a Leica DMI8 microscope.

The plasmid pX330-U6-Chimeric_BB-CBh-hSpCas9 (Addgene, #42230) was used to perform genome editing in 46C mESCs. A total of 2 µg of pX330-U6-Chimeric_BB-CBh-hSpCas9 plasmid was transfected into mESCs with Lipofectamine 3000 (Life Technologies, L3000-015) according to the manufacturer's protocol. After 2 days, puromycin was added to the supernatant at a final concentration of 2 μg/ml. Finally, we picked single colonies and extracted the genomic DNA. The disruption of the PKD1, PKD2 and PKD3 alleles was confirmed by DNA sequencing and western blot. The gDNA sequences are listed in Table S2.

To demonstrate the pluripotency of mESCs maintained in PC-containing medium, spontaneous differentiation was performed by EB formation for 8 days and further plating of EBs for an additional 7 days on gelatin-coated plates in mESC media without LIF. For hESC differentiation, hESCs were differentiated into hepatocytes-like cells (Hay et al., 2008), cardiomyocytes-like cells (Weng et al., 2014) and neural cells (Schwartz et al., 2015), as previously reported.

Cells were lysed in ice-cold RIPA cell buffer (Beyotime Biotechnology, P0013B) supplemented with protease inhibitor cocktail (TransGen Biotech, DI111). The proteins were separated on 8% PAGE gels and electrotransferred onto polyvinylidene fluoride membranes that were placed in 5% milk for blocking. Antibody probing was performed with specific primary antibodies (4°C overnight) and horseradish peroxidase-conjugated secondary antibodies (25°C for 1 h). The primary antibodies used in this study targeted FLAG (GNI, SG4110-26, 1:2000), AKT (Cell Signaling Technology, 4691P, 1:1000), AKT (Beyotime Biotechnology, AA326, 1:500), phospho-AKT (Cell Signaling Technology, 4060P, 1:1000), PKD1 (Abcam, ab51246, 1:1000), PKD2 (Abcam, ab51250, 1:1000), PKD3 (Cell Signaling Technology, 5655S, 1:1000), β-actin (Proteintech, 60008-1-Ig, 1:2000), Stat3 (Cell Signaling Technology, 12640, 1:1000), phospho-Stat3 (Tyr705) (Cell Signaling Technology, 9145, 1:1000) and GAPDH (Boster Bio, BA2913, 1:2000).

Total RNA was extracted using the TRIzol Up Plus RNA Kit (TransGen Biotech, ER501-01). cDNA was synthesized from 1 µg of total RNA using the Hifair III 1st Strand cDNA Synthesis SuperMix for qPCR (Yeasen, 11141ES60) according to the manufacturer's instructions. qRT-PCR was carried out with the Hieff qPCR SYBR Green Master Mix (Yeasen, 11201ES08) using a PikoReal Real-Time PCR machine (Thermo Scientific). The relative expression level was determined by the 2-ΔCq method and normalized to RPL19 or β-actin expression. The primers used are listed in Table S3.

Cells were fixed in 4% paraformaldehyde for 20 min and incubated at 37°C in blocking buffer, containing 5% bovine serum albumin and 0.2% Triton X-100. The cells were then incubated in the presence of primary antibodies at 4°C overnight. Subsequently, after being washed three times with PBS, the cells were incubated with secondary antibody for 1 h at 37°C. The nuclei were stained with Hoechst 33342 (Invitrogen, H3570, 1:10,000) in blocking buffer for 30 s followed by extensive washing with PBS. The cells were then photographed with a Leica DMI8 microscope. The primary antibodies used for immunofluorescence staining included antibodies against Oct4 (Santa Cruz Biotechnology, SC-5279, 1:200), FLAG (GNI, SG4110-26, 1:2000), GATA4 (Cell Signaling Technology, 36966S, 1:100), AFP (Proteintech, 14550-1-AP, 1:100), myosin (Abcam, ab50967, 1:100) and Tuj1 (R&D Systems, MAB1195, 1:100).

Briefly, C57BL/6 mESCs were trypsinized into single cells, after which 12-15 of the single cells were injected into 3.5 dpc blastocysts collected from B6(Cg)-Tyrc-2J/J mice (The Jackson Laboratory, 000058), which harbor a mutation in the tyrosinase and have a white coat (Townsend et al., 1981). The injected blastocysts were incubated at 37°C for 30 min in M16 medium to allow for their recovery. Sixty embryos were then transferred into the uterine horns of four pseudopregnant female mice. Following implantation and gestation, the offspring generated that were chimeric for the donor ESCs and the recipient blastocyst were easily identified by coat color. To test germline transmission, the male chimera was mated to female B6(Cg)-Tyrc-2J/J mice.

All data are reported as the mean±s.d. Student's t-test was used to determine the significance of differences in comparisons. Values of P<0.05 were considered statistically significant.

We thank Qi-Long Ying (University of Southern California, Los Angeles, CA, USA) for providing the 46C mESCs and HES2 hESC; and GemPharmatech for supporting the microinjection.