We have previously used the distinctive HLA class

We have previously used the distinctive HLA class I profile of VCT and EVT to characterize BMP-treated hESC (Bernardo et al., 2011). As a further refinement, we now show that we can distinguish between products of different HLA class I loci, particularly HLA-G and HLA-A and -B. With knowledge of the HLA class I locus-specific glutathione reductase present in 2102Ep and the hESC lines, we selected mAbs that bind specifically to different HLA allotypes (Brodsky et al., 1979; Parham and Brodsky, 1981; Josephson et al., 2007; NIH, 2009). Other mAbs are available that bind various combinations of HLA-A and -B allotypes, and HLA-Bw4 and -Bw6 epitopes (Koene et al., 2006; Schumm et al., 2007; Duquesnoy et al., 2013). With HLA genomic typing of the test cells and selection of appropriate antibodies, it should generally be possible to make a comprehensive comparison to discern whether if HLA-A, -B, or -G molecules are expressed. HLA-G is never expressed together with HLA-A and -B in normal trophoblast. Flow cytometry allows analysis of the frequency of subpopulations, providing another advantage of screening the HLA class I profile of putative trophoblast cells. Indeed, no subpopulation of either HLA class I negative or HLA-G positive cells was detected in BAP-treated hESC, highlighting the power of flow cytometric analysis. This compares with the difficulties in interpreting immunofluorescence images of small cellular clusters that may not be representative of the whole population. We also show that hypomethylation of the ELF5 promoter is specific for normal VCT and EVT but not for non-trophoblast placental villous mesenchymal cells (Hemberger et al., 2010). The ELF5 promoter is hypermethylated in the following cells: 2102Ep cells, hCG-secreting hESC, and BMP4-treated hESC, TCL1, SWAN-71, and HTR-8/SVneo (Hemberger et al., 2010; Bernardo et al., 2011; Novakovic et al., 2011; Sarkar et al., 2015). Fibroblasts reprogrammed with CDX2, EOMES, and ELF5 have some characteristics of trophoblast (KRT7, GATA3, and HLA-G expression), but the ELF5 hypermethylation pattern is similar to that of the parental fibroblasts (Chen et al., 2013a). This differential methylation between hESC and trophoblast suggests that, as in mice, the epigenetic status of human ELF5 segregates the embryonic and extraembryonic lineages. We find that despite partial hypomethylation in BAP-treated hESC, ELF5 expression levels remain very low, as in mouse ESC (Cambuli et al., 2014). This is also similar to EVT, where ELF5 is hypomethylated but only expressed at low levels, indicating that either ELF5 is silenced by other mechanisms, or that the transcriptional machinery for its activation is not in place. Overall, this is in line with the commonly accepted view that promoter hypomethylation is necessary but not sufficient for gene activation (Deaton and Bird, 2011). Thus, both the methylation status of ELF5 and its expression levels are useful as trophoblast identifiers. The methylation status of other genes (e.g. the promoters of CGB are hypomethylated in trophoblast) compared with other cell types might serve as additional trophoblast markers (Novakovic et al., 2011). We have now added another marker for trophoblast, the expression of high levels of C19MC miRNAs, which is characteristic of primary trophoblast and choriocarcinoma cells (10- to 10,000-fold higher expression of these miRNAs compared with other cells including hESC and EC). Because levels of C19MC were much lower in hESC than in trophoblast cells, we would predict upregulation if trophoblast lineage differentiation occurs. However, we observed the opposite, with downregulation in BAP-treated hESC. It is essential to include primary trophoblast or choriocarcinoma cells as positive controls for the analysis of C19MC and ELF5 expression levels and appropriate negative controls, such as leukocytes. To summarize, our results show that very high expression of C19MC miRNAs is a hallmark of first-trimester trophoblast.