Introduction

Normal development and function of the placenta is essential for a healthy pregnancy. Dysfunction of the earliest stages of trophoblast differentiation and implantation can cause common pregnancy complications, including preeclampsia, fetal growth restriction, low birth weight, early pregnancy loss, stillbirth, and preterm birth [1,2,3]. The human blastocyst implants into the uterine luminal epithelium at 6–9 days post fertilization. It consists of the inner cell mass (ICM), which gives rise to the embryo, and the trophectoderm (TE), which gives rise to the different types of trophoblast cells. In vitro, TE gives rise to trophoblast stem (TS) cells. TSC can also be derived from first-trimester villous cytotrophoblast (CTB). After implantation, cells of the polar TE invade through the endometrium and later the decidual glands, and fuse into the multinucleated primitive syncytium (PS) which then acquires lacunae [3,4,5]. The CTB cells adjacent to the ICM then form projections through the PS that merge to surround the PS as a cytotrophoblastic shell. These projecting CTB become the primitive villi containing villous cytotrophoblast (VCT) covered by a layer of fused CTB, the syncytiotrophoblast (STB). The STB is responsible for gas and nutrient exchange and secretion of human chorionic gonadotropin (hCG) and placental lactogen. These villi branch and proliferate further, then acquire a mesenchymal core (secondary villi), and next fetal capillaries (tertiary villi). Cells from the cytotrophoblastic shell further invade and become extravillous trophoblast (EVT) in anchoring villi. The EVT differentiates into two subtypes, interstitial EVT (iEVT) which invades the myometrium near the spiral arteries, and endovascular EVT (eEVT) which invades inside the spiral arteries. Remodeling of the maternal spiral arteries by EVTs is critical for optimal blood flow to the fetus [3,4,5].

The different stages of trophoblast development and subtypes of trophoblast cells are marked by the expression of different genes, but the processes, molecular pathways, and epigenetic modifications that control them are still being unraveled. This is due to ethical limitations and technical difficulties with accessing human trophoblast tissues at early developmental stages and to differences in placentation between humans and commonly used rodent models. One approach used to overcome these limitations is the in vitro differentiation of human embryonic stem cells (hESCs) into trophoblast lineages [6, 7]. Human ESCs are derived from the ICM of blastula-stage preimplantation embryos. Under appropriate in vitro culture conditions, they can be sustained in a pluripotent state and undergo prolonged self-renewal, but they are primed to differentiate into all three germ layers and into extraembryonic tissues. After the first report in 2002 [6], several studies have demonstrated that hESCs can differentiate into trophoblast cells when exposed to bone morphogenetic protein 4 (BMP4) [6, 8,9,10,11,12,13,5d and 5e). This was most prominent in denser sheet-like areas with cells that appeared multinuclear, supporting more effective differentiation towards syncytiotrophoblast in those media (Fig. 5e).

Fig. 5
figure 5

Transcript levels for CGB, HLA-G, and GCM1, Immunofluorescence (IF) staining for \(\alpha\)-hCG, \(\beta\)-hCG, and HLA-G and Western blot quantification of protein levels for \(\alpha\)-hCG, \(\beta\)-hCG, HLA-G, and GCM1. a, f, h relative transcript levels of CGB (a), HLA-G (f), and GCM1 (h) assayed by qRT-PCR and quantified using ΔΔCt normalized to ACTA in differentiation media (D) compared to the corresponding pluripotency media (P). D5, D3, D5, and D7 on the X-axis represent days in culture in each differentiation medium and matching pluripotency medium. Relative transcript levels and number of replicates are on the Y-axis. Data are presented as mean \(\pm\) standard error of the mean (SEM); ****p < 0.0001. b, c, g, i Western blot quantification of protein levels of \(\alpha\)-hCG (b), \(\beta\)-hCG (c), HLA-G (g), and GCM1 (i) and corresponding western blot image (d). Lysates were collected on day 7 from cells cultured in differentiation media and on day 5 from cells cultured in matching pluripotent media. Each graph shows the relative protein levels assayed by densitometry normalized to ACTA (Y-axis) in the respective media (X-axis). Data are obtained from two independent rounds of differentiation (n = 6). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. e, j Images of cultured cells stained on day 3 (D3) and day 7 (D7) for \(\alpha\)-hCG, \(\beta\)-hCG (left and middle images) and co-stained with nuclear marker DAPI (right images) (e) and for HLA-G (j) in differentiation media. IF staining was performed in triplicate from two independent rounds of differentiation

Full size image

We also examined the expression of HLA-G, a marker that correlates with the degree of CTB invasiveness and is not expressed in the non-invasive STB of the chorionic villi [36]. Colonies cultured in basal-BAP had higher levels of HLA-G transcripts (Fig. 5f). Western analysis of lysates showed higher levels of HLA-G protein in basal-BAP (p < 0.05) and E7-BAP (p < 0.05) compared to corresponding pluripotency media, but no HLA-G was detected in any other differentiation media (Fig. 5d, g). Immunostaining confirmed HLA-G expression on day 7 in colonies grown in basal-BAP, but not in E7-BAP (Fig. 5j). Thus, combined hCG and HLA-G expression studies suggest that E7-BAP cultures yielded more differentiation towards hCG-expressing STB, while cells grown in basal-BAP medium achieved both STB and EVT lineage differentiation.

GCM1 plays a role in the regulation of syncytialization of cytotrophoblast [37]. Transcript abundance of GCM1 increased starting day 3 and reached a highly stable level by day 5 (p < 0.0001) that was maintained on day 7 in colonies cultured in basal-BAP, but it was not altered in other media. Although a small increase was seen on day 7 in E7-BAP, it was not significant (Fig. 5h). Western analysis of day 7 lysates detected low expression of GCM1 protein in all culture conditions with no significant difference between different culture media (Fig. 5d, i).

The Abundance of C19MC miRNAs and ELF5 Promoter Methylation

Previous studies have shown that primary first-trimester trophoblast and choriocarcinoma cell lines (JEG-3, JAR) have a high abundance of C19MC miRNAs and that H9 hESCs have moderate levels of C19MC miRNAs which further decreased upon differentiation into trophoblast [20]. All four C19MC miRNAs that were examined (has-miR-517-5p, has-miR-517b-3p, has-miR-525-3p, has-miR-526b-3p) were expressed in hESCs grown in the three-pluripotency media (Fig. 6a–d). Transcript levels of all four C19MC miRNAs were lower by day 3 of differentiation in all media compared to their pluripotent media, and remained relatively low thereafter, except for levels of miR-525-3p in hESC grown in basal-BAP, which were not significantly lower until day 5 (p < 0.05). Levels of miR-517-5p in E7-BAP and basal-BAP and of miR-517b-3p in E7-BAP appeared slightly higher than in the other differentiation media at each time point, but this difference was not significant (p > 0.05).

Fig. 6
figure 6

Abundance of selected miRNAs from the C19MC complex and promoter CpG methylation and transcript levels of ELF5. a–d Graphs of relative transcript levels of four C19MC miRNAs (a miR517b-3p; b miR517b-5p; c miR525-3p; d miR526b-3p) assayed by qRT-PCR and quantified using ΔΔCt normalized to miR103a in corresponding pluripotency media in hESCs grown in differentiation media (D) on days 3, 5, and 7 of differentiation, compared to hESC cultured in matching pluripotent media (P) collected on day 5. Days in culture in each differentiation medium and matching pluripotency medium are shown on the X-axis. Relative miRNA levels and number of replicates for each gene are on the Y-axis. Data are presented as mean \(\pm\) standard error of the mean (SEM) and each experiment was performed in triplicate (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. e, f ELF5 promoter methylation analysis by bisulfite sequencing of individual clones in cells cultured in pluripotency media (e) and differentiation media (f) on day 7. Each row represents an individual sequenced clone; methylated and unmethylated CpG dinucleotides are represented as open and filled circles, respectively. Percentages show the proportion of methylated to non-methylated CpG sites for each set of cloned PCR products. g Relative transcript levels of ELF5 assayed by qRT-PCR and quantified using ΔΔCt normalized to ACTA in differentiation media (D) compared to the corresponding pluripotency media (P). D5, D3, D5, and D7 on the X-axis represent days in culture in each differentiation medium and matching pluripotency medium. Relative transcript levels and number of replicates are on the Y-axis. Data are presented as mean \(\pm\) standard error of the mean (SEM); ***p < 0.001, ****p < 0.0001

Full size image

In early human placenta, the ELF5 promoter is hypomethylated and ELF5 is highly expressed. It has been shown that the ELF5 promoter in hESC is hypermethylated and ELF5 gene expression is downregulated [20, 26]. We examined methylation of the ELF5 promoter in hESC grown in pluripotency media and quantified how it changes upon differentiation in the four differentiation media. As expected, we found that the ELF5 promoter is highly methylated in pluripotent hESCs in all examined pluripotency media (between 83.8 and 91.9%) (Fig. 6e). After 7 days of differentiation, methylation of the ELF5 promoter was reduced to 64.6% in mTeSR1-BMP4, to 38.3% in mTeSR1-BAP, to 37.3% in basal-BAP, and was the lowest in E7-BAP at 28.2% (Fig. 6f). Transcript levels assessed by qRT-PCR showed high persistent levels starting on day 3 in cells grown in basal-BAP compared to transcript level in corresponding pluripotency media (p < 0.0001). A smaller increase that did not reach significance was observed in cells grown in E7-BAP on day 3 (Fig. 6g).

Discussion

In this study, we evaluated three published protocols and culture conditions along with a protocol developed in our laboratory. We examined morphological changes over seven days of in vitro differentiation. We also studied various trophoblast differentiation markers by quantitative RT-PCR of expressed transcripts and by western analysis and immunofluorescence staining of expressed proteins. Finally, we assayed changes in methylation at the ELF5 promoter and in the expression of the C19MC miRNA cluster. Our results indicate that, consistent with other data [6, 8,9,10,11, 18, 19, 38], all differentiation media can drive trophoblast lineage differentiation from the periphery towards the center of hESC colonies, but to variable degrees and at a variable pace (differences in silencing of pluripotency markers and the expression levels of certain other markers are summarized in Supplementary Table S4). For our primary analysis, we compared the marker expression analysis for the new E7-BAP differentiation protocol to its corresponding pluripotency medium, E8. To verify that the additional step required for culturing hESC in the E8 pluripotency medium (see Fig. 1a) does not influence the results, we also compared qRT-PCR marker expression data in E7-BAP to the mTeSR1 pluripotency medium and found equivalent results (Supplementary Figs. S4 and S5).

Human ESC cultured in a standard pluripotency medium have an epithelial morphology, high levels of DNA methylation, and a transcriptome resembling primed post-implantation epiblast. Therefore, they have been referred to as epiblast-type or primed cells [39]. Different reported protocols for differentiation of primed human embryonic stem cells along trophoblast lineages have shown variable results, which has led some to debate the ability of hESC to differentiate into trophoblast in these culture systems [17, 40]. After the initial report by Xu et al. that BMP4 drives trophoblast differentiation [6], approaches of BMP4-driven differentiation of hESC in defined media had variable results, with some yielding primarily mesoderm [17] or endoderm [41]. Further studies have shown that more effective BMP4-driven differentiation towards trophoblast requires suppression of FGF2 and activin/nodal signaling, but there remains variation in the details between these protocols that affect the fidelity and efficiency of differentiation towards trophoblast. This can be influenced by ambient O2 conditions, the source and concentration of BMP4 used, starting sizes of colonies, and initial colony density at the time of BMP4 addition [41]. In some reports, FGF2 was present [6, 12, 16, 42] and in other reports, it was excluded at the time BMP4 was added [8, 43,44,45]. For example, Wei and colleagues used a novel E6 medium, which lacks TGF-β and bFGF, to differentiate induced pluripotent stem cells (iPSCs) into trophoblast-like cells. They observed that the iPSC-derived trophoblast-like cells had similar gene expression profiles and displayed syncytialization and invasiveness in Matrigel [45].

We obtained effective downregulation of pluripotency factors in all differentiation media, except for a slight upregulation of NANOG in basal-BAP, in which cells also proliferated much faster than in other media. The mTeSR1 medium containing only BMP4 resulted in slower differentiation that started later and was associated with less effective suppression of mesodermal fate, as evidenced by EOMES expression on day 3, and with less effective syncytialization. This is consistent with other observations of upregulation of mesoderm and endoderm markers in this culture medium [16, 18, 19]. Colonies cultured in mTeSR1 with BMP4 or BAP for 7 days did not produce CGB transcripts and β-hCG was not seen by western analysis. These colonies also had variable and lower evidence of β-hCG-positive cells by immunofluorescence. In contrast, the fastest upregulation of CGB transcripts (by day 3) was in cells cultured in basal-BAP, followed by cells cultured in E7-BAP, both of which showed sheets of multinucleated cells with a syncytial appearance.

Basal-BAP was the only medium that also led to strong and early upregulation of HLA-G transcripts by day 3 and significantly elevated HLA-G protein levels by western analysis on day 7. Culturing in E7-BAP resulted in lower but still significant upregulation of HLA-G transcripts from day 5 onwards and HLA-G protein was identified by western analysis on day 7. However, the only cultures that contained cells with HLA-G by immunofluorescence labeling were basal-BAP cultures on day 7. This could suggest that only basal-BAP drives efficient differentiation into EVT lineages, and that cells that acquire EVT characteristics are rarer with differentiation in E7-BAP. This is consistent with other studies wherein FGF2 inhibition together with in vitro differentiation with BMP4 or BAP primarily drives STB differentiation and not EVT lineages [10, 11, 20, 43, 46]. Alternatively, in E7-BAP EVT lineages may appear later and possibly in a more controlled fashion, but evaluating this will require future studies with longer cultures in E7-BAP. While the strong expression of HLA-G in basal-BAP, along with the highest and earliest expression of several other markers of trophoblast differentiation (GATA3, TFAP2C, KRT7, CDX2, and GCM1) compared to other cultures at first may appear promising, this medium has several disadvantages that make it less optimal for experiments requiring a purer trophoblast population. Morphologically, the cultures are more mixed and disorganized, proliferate more rapidly, and re-express NANOG transcripts on day 7, suggesting that cells have maintained higher proliferative capacity. One limitation of our study was that we did not test for the absence or downregulation of HLA-A, HLA-B, and HLA-C expression, which are important to evaluate the efficiency and purity of trophoblast lineage differentiation [47]. Another limitation of our study is that although in our IF experiments we used validated antibodies and all experiments were performed on cells from two independent rounds of differentiation with appropriate control pluripotency cultures, we did not include a specific positive control for each marker with nuclear staining and a negative control to detect any background staining.

To further evaluate how well the differentiated cells adopted trophoblast identity, we examined two additional markers, expression of the C19MC miRNA cluster and methylation of the ELF5 promoter [20]. High expression of C19MC miRNAs is a hallmark of first-trimester trophoblast, increases as gestation progresses, and is also implicated in the regulation of EVT invasion [20, 48, 49]. We found that all four miRNAs of the C19MC cluster we tested were downregulated by day 3 in all differentiation media, which has been observed before in hESC differentiation towards trophoblast lineages [20]. This result may reflect the incomplete differentiation on day 7 and we speculate that longer cultures may increase the C19MC miRNA expression and more advanced first-trimester trophoblast-like identity, possibly more in the E7-BAP media. This can be tested in future longer-term experiments. ELF5 promoter methylation is considered an epigenetic modification that keeps extraembryonic and embryonic lineages separate; in the human placenta, the promoter is hypomethylated and the gene is expressed, although promoter methylation increases from the first to the third trimester of pregnancy [26]. The ELF5 promoter is highly methylated in non-trophoblast placental villous mesenchymal cells and is hypomethylated to different degrees in EVTs and vCTBs, but these lower methylation levels are not associated with strong expression [20]. Earlier studies of in vitro differentiation showed that ELF5 is hypermethylated and repressed in hESC and remains hypermethylated in derived trophoblast cell lines, suggesting that these cells did not exhibit the epigenetic signature of trophoblast lineages [20, 26]. We confirmed that the ELF5 promoter was highly methylated in hESCs cultured in pluripotency media, consistent with older previous studies [20, 26, 27]. However, unlike what has been reported in some previous studies [26], but consistent with other data [12], it became hypomethylated to varying degrees in differentiation media, with E7-BAP resulting in the lowest methylation levels (Fig. 6f). This supports that these cells began to acquire epigenetic characteristics of trophoblast lineages.

In conclusion, we confirmed the ability of BAP to generate trophoblast-like cells in four examined media including a newly developed defined medium, E7-BAP, which lacks FGF2. This combination adequately suppressed mesendodermal factors and emphasizes the importance of FGF2 inhibition for preventing BMP4-mediated differentiation into mesoderm and endoderm lineages. We also found that compared to other media, differentiation in E7-BAP resulted in more hypomethylation of the ELF5 promoter, supporting an epigenetic state closer to that seen in the first-trimester trophoblast.