Germline competency of human embryonic stem cells depends on eomesodermin^†

Human fetal samples

Human prenatal testes and ovaries were acquired following elected termination and pathological evaluation after University of California, Los Angeles (UCLA)-Institutional Review Board (IRB) review, which deemed the project exempt under 45 CRF 46.102(f). All prenatal gonads were obtained from the University of Washington Birth Defects Research Laboratory (BDRL), under the regulatory oversight of the University of Washington IRB-approved Human Subjects protocol combined with a Certificate of Confidentiality from the Federal Government. BDRL collected the fetal testes and ovaries and shipped them overnight in HBSS with ice pack for immediate processing in Los Angeles. All consented material was donated anonymously and carried no personal identifiers. Developmental age was documented by BDRL as days post fertilization using a combination of prenatal intakes, foot length, Streeter Stages, and crown-rump length. All prenatal gonads documented with birth defect or chromosomal abnormality, were excluded from this study.

hESC culture

All primed hESC lines were cultured on mitomycin C-inactivated mouse embryonic fibroblasts (MEFs) in hESC media, which is composed of 20% knockout serum replacement (KSR) (GIBCO, 10828-028), 100 μM L-Glutamine (GIBCO, 25030-081), 1× MEM Non-Essential Amino Acids (NEAA) (GIBCO, 11140-050), 55 μM 2-Mercaptoethanol (GIBCO, 21985-023), 10 ng/mL recombinant human FGF basic (R&D systems, 233-FB), 1× Penicillin-Streptomycin (GIBCO, 15140-122), and 50 ng/mL primocin (InvivoGen, ant-pm-2) in DMEM/F12 media (GIBCO, 11330-032). All hESC lines were split every 7 days with Collagenase type IV (GIBCO, 17104-019). 4i hESCs were maintained as described before [19]. 4i cells were grown on irradiated MEFs (GlobalStem) in knockout DMEM containing 20% KSR, 2 mM L-glutamine, 0.1 mM NEAA, 0.1 mM 2-mercaptoethanol (all GIBCO), 20 ng/mL human LIF (Stem Cell Institute [SCI]), 8 ng/mL bFGF (SCI), 1 ng/mL TGF-β1 (Peprotech), 3 μM CHIR99021 (Miltenyi Biotec), 1 μM PD0325901 (Miltenyi Biotec), 5 μM SB203580 (TOCRIS bioscience), and 5 μM SP600125 (TOCRIS bioscience). 4i hESCs were split every 3 to 5 days using TrypLE Express (GIBCO). An amount of 10 μM of ROCK inhibitor (Y-27632, TOCRIS bioscience) was used for the first 24 h after passage. All hESC lines used in this study are registered with the National Institute of Health Human Embryonic Stem Cell Registry and are available for research use with NIH funds. Specifically, the following hESC lines were used in this study: UCLA1 (46XX), UCLA2 (46XY), UCLA3 (46XX), UCLA4 (46XX), UCLA5 (46XX), UCLA6 (46XY), UCLA7 (47XX+13), UCLA8 (46XX), UCLA9 (46XX), UCLA10 (46XY), UCLA11 (46XY), UCLA12 (46XX), UCLA13 (46XY), UCLA14 (46XX), UCLA15 (46XX), UCLA16 (46XX), UCLA17 (46XX), UCLA18 (46XX). The derivation and basic characterization (Karyotype and teratoma analysis) of UCLA1–6 were previously reported [22]. UCLA8–10, UCLA14, and UCLA16–18 were reported [23].

Derivation and characterization of ESC lines from human embryos

The following UCLA hESC lines UCLA7 (47XX+13), UCLA11 (46XY), UCLA12 (46XX), UCLA13 (46XY), UCLA15 (46XX) were derived from human embryos according to the methods described [22]. All hESC derivations were performed after human subjects’ approval from the UCLA-IRB and following Embryonic Stem Cell Research Oversight committee approval. No NIH funds were used for the derivation or initial characterization (karyotype and teratoma) of hESC lines. Teratoma analysis was performed after Institutional approval by the UCLA Office of Animal Research Oversight. Teratomas were created by injecting of collagenase digested clumps of hESCs into the testicles of male scid/beige C.B.17-Prkdc(scid)Lyst(bg) mice. Prior to injection, hESCs were resuspended in ice-cold matrigel (Corning, 354277), with 3× wells (from a 6-well plate) of colonies injected per testis. Karyotype analysis was conducted by Cell Line Genetics.

Induction of human PGCLCs though incipient mesoderm-like cells from primed hESCs

Human PGCLCs were induced from primed hESCs as described in [19] with some modifications. Day 7 hESCs were dissociated into single cells with 0.05% Trypsin-EDTA (Gibco, 25300-054) and plated onto Human Plasma Fibronectin (Invitrogen, 33016-015)-coated 12-well plate at the density of 200 000 cells/well in 2 mL/well of incipient mesoderm-like cell (iMeLC) media, which is composed of 15% KSR, 1× NEAA, 0.1 mM 2-Mercaptoethanol, 1× Penicillin-Streptomycin-Glutamine (Gibco, 10378-016), 1 mM sodium pyruvate (Gibco, 11360-070), 50 ng/mL Activin A (Peprotech, AF-120-14E), 3 μM CHIR99021 (Stemgent, 04-0004), 10 μM of ROCKi (Y27632, Stemgent, 04-0012-10), and 50 ng/mL primocin in Glasgow MEM (GMEM) (Gibco, 11710-035). iMeLCs were dissociated into single cells with 0.05% Trypsin-EDTA after 24 h of incubation unless otherwise mentioned and plated into ultra-low cell attachment U-bottom 96-well plates (Corning, 7007) at the density of 3000 cells/well in 200 μL/well of PGCLC media, which is composed of 15% KSR, 1× NEAA, 0.1 mM 2-Mercaptoethanol, 1× Penicillin-Streptomycin-Glutamine (Gibco, 10378-016), 1 mM sodium pyruvate (Gibco, 11360-070), 10 ng/mL human LIF (Millipore, LIF1005), 200 ng/mL human BMP4 (R&D systems, 314-BP), 50 ng/mL human EGF (R&D systems, 236-EG), 10 μM of ROCKi (Y27632, Stemgent, 04-0012-10), and 50 ng/mL primocin in GMEM (Gibco, 11710-035). An amount of 100 ng/mL stem cell factor (SCF; PEPROTECH, 250-03), 10 μM SB431542 (Stemgent, 04-0010-10), and 500 ng/μL Dickkopf Wnt Signaling Pathway Inhibitor 1 (DKK1) (R&D systmes, 5439-DK) was added in PGCLC media for some experiments.

Induction of human PGCLCs from 4i hESCs

4i WIS2 cells were dissociated with TrypLE and plated to ultra-low cell attachment U-bottom 96-well plates (Corning, 7007) at the density of 2000–4000 cells/well in 200 μL PGCLC media, which is composed of 15% KSR, 0.1 mM NEAA, 0.1 mM 2-mercaptoethanol, 100 U/mL Penicillin–0.1 mg/mL Streptomycin, 2 mM L-Glutamine, 1 mM Sodium pyruvate, 500 ng/mL BMP4 (R&D Systems) or BMP2 (SCI), 1 μg/mL human LIF (SCI), 100 ng/mL SCF (R&D Systems), 50 ng/mL EGF (R&D Systems), and 10 μM ROCK inhibitor in GMEM (Gibco, 11710-035).

Flow cytometry and fluorescence activated cell sorting

Human prenatal gonads or day 4 aggregates were dissociated with 0.25% trypsin (Gibco, 25200-056) for 5 min or 0.05% Trypsin-EDTA (Gibco, 25300-054) for 10 min at 37°C. The dissociated cells were stained with conjugated antibodies, washed with fluorescence activated cell sorting (FACS) buffer (1% BSA in PBS), and resuspended in FACS buffer accompanying with 7-AAD (BD Pharmingen, 559925). The single cell suspension was analyzed or sorted. The conjugated antibodies used in this study are ITGA6 conjugated with BV421 (BioLegend, 313624), EPCAM conjugated with 488 (BioLegend, 324210), EPCAM conjugated with APC (BioLegend, 324208), tissue nonspecific alkaline phosphatase (TNAP) conjugated with PE (BD Pharmingen, 561433), and cKIT conjugated with APC (BD Pharmingen, 550412).

Real-time quantitative polymerase chain reaction

Sorted cells or cell pellets were lysed in 350 μL of RLT buffer (QIAGEN) and RNA was extracted using RNeasy micro kit (QIAGEN, 74004). Complementary DNA was synthesized using SuperScript II Reverse Transcriptase (Invitrogen, 18064-014). Real-time quantitative polymerase chain reaction (PCR) was performed using TaqMan Universal PCR Master Mix (Applied Biosystems, 4304437) and the expression level of genes-of-interest were normalized to the expression of housekeeping gene GAPDH. The Taqman probes used in this study include GAPDH (Applied Biosystems, Hs99999905_m1), NANOS3 (Applied Biosystems, Hs00928455_s1), PRDM1 (Applied Biosystems, hs01068508_m1), TFAP2C (Applied Biosystems, Hs00231476_m1), SOX17 (Applied Biosystems, Hs00751752_s1), cKIT (Applied Biosystems, hs00174029_m1), DAZL (Applied Biosystems, hs00154706_m1), DDX4 (Applied Biosystems, Hs00251859_m1), SOX9 (Applied Biosystems, Hs01001343_g1), AMH (Applied Biosystems, Hs01006984_g1), DND1 (Applied Biosystems, Hs00832091_s1), T (Applied Biosystems, Hs00610080_m1), CER1 (Applied Biosystems, Hs00193796_m1), MIXL1 (Applied Biosystems, Hs00430824_g1), WNT3 (Applied Biosystems, Hs00902257_m1), EOMES (Applied Biosystems, Hs00172872_m1), GSC (Applied Biosystems, Hs00906630_g1), FOXA2 (Applied Biosystems, Hs00232764_m1), POU3F1 (Applied Biosystems, Hs00538614_s1), TCF15 (Applied Biosystems, Hs00231821_m1). For human gonad samples, two technical replicates of real-time quantitative PCR were performed. For hESCs, iMeLCs, and PGCLCs, two to three independent experiments were performed.

Generation of EOMES mutant hESC lines

A pair of guide RNA (gRNA) targeting EOMES was designed using crispr.mit.edu, and the corresponding gDNA sequence was cloned into px330 vector [24]. An amount of 4 μg of gRNA pair or 2 μg of pMax-GFP was electroporated into 800 000 UCLA1 cells using P3 Primary Cell 4D-Nucleofector X Kit according to the manufacturer's instructions (Lonza, V4XP-3024). Twenty-four hours after nucleofection, cells were dissociated with Accutase (ThermoFisher Scientific, A1110501) and seeded in low density. A total of 96 individual colonies were picked after 9 days and expanded. Lines were screened for the presence of shorter bands due to deletion. To determine the precise mutations, PCR product from the targeted allele was cloned using Topo-TA cloning (Thermo-Fisher) and analyzed by Sanger sequencing. Two mutant lines were chosen and subcloned before experiments. The gRNA sequences used to target EOMES are 5’- GCGGTGTACAGCCGTAACAT and 5’-GTTATCTACACCGAAAGTGC. Karyotyping by Cell Line Genetics was performed before experimentation. GFP-labeled control and EOMES mutant hESCs were made by lentivirus-based transfection of UbiC-GFP-IRES-Puromycin and maintained as stable cell lines with puromycin (1 μg/mL) selection.

Immunofluorescence

Immunostaining paraffin sections or aggregates in whole mount were described previously [25, 26]. For cells cultured on chamber slides, samples were fixed in 4% paraformaldehyde in PBS for 10 min and washed with PBS containing 0.1% Tween 20 and permeabilized with PBS containing Triton X for 10 min. Samples were blocked with 10% donkey serum for 30 min before antibody incubation. The primary antibodies used for immunofluorescence in this study include rabbit-anti-EPCAM (Abcam, ab71916, 1:50), goat-anti-VASA (R&D Systems, AF2030, 1:20), rabbit-anti-cKIT (DAKO, A4502, 1:100), goat-anti-OCT4 (Santa Cruz Biotechnology, sc-8628x, 1:100), rabbit-anti-PRDM1 (Cell Signaling Technology, 9115, 1:100), mouse-anti-PRDM1 (R&D Systems, MAB36081SP, 1:100), rabbit-anti-TFAP2C (Santa Cruz Biotechnology, sc-8977, 1:100), mouse-anti-TFAP2C (Santa Cruz Biotechnology, sc12762, 1:100), rat-anti-ITGA6 (Santa Cruz Biotechnology, sc-80554, 1:100), goat-anti-T (R&D Systems, AF2085, 1:100), goat-anti-SOX17 (Neuromics, GT15094, 1:100), rabbit-anti-EOMES (Abcam, ab23345, 1:100), rabbit-anti- β-CATENIN (Cell Signaling Technology, 9582, 1:100), rabbit-anti-pSMAD2/3 (Cell Signaling Technology, 8828, 1:100). The secondary antibodies used in this study are all from Jackson ImmunoResearch Laboratories including donkey-anti-rabbit-488, donkey-anti-mouse-488, donkey-anti-goat-488, donkey-anti-rat-488, donkey-anti-rabbit-594, donkey-anti-mouse-594, and donkey-anti-goat-594. DAPI is counterstained to indicate nuclei.

RNA-sequencing

Sorted cells or cell pellets were lysed in 350 μL of RLT buffer (QIAGEN), and total RNA was extracted with RNeasy micro kit (QIAGEN, 74004). Total RNA was reverse transcribed and cDNA was amplified using Ovation RNA-Seq System V2 (Nugen, 7102-32) according to manufacturer's instructions. Amplified cDNA was fragmented into ∼200 bp by Covaris S220 Focused-ultrasonicators. RNA-sequencing (RNA-seq) libraries were generated using Ovation Rapid Library Systems (Nugen, 0319-32 for index 1-8 and 0320-32 for index 9–16) and quantified by KAPA library quantification kit (Illumina, KK4824). Libraries were subjected to single-end 50 bp sequencing on HiSeq 2000 or HiSeq 2500 sequencer with 4–6 indexed libraries per lane.

RNAseq analysis

ITGA6 and EPCAM can be used to isolate human PGCs from embryonic ovaries but not embryonic testes

Recently, in vitro human PGCLCs were isolated using INTEGRINα6 (ITGA6) and EPCAM following hiPSC differentiation [19]; however, whether ITGA6/EPCAM can be used to isolate in vivo PGCs from human embryos has not been shown. To address this, we examined cells from a pair of human embryonic ovaries at 72 day postfertilization by staining with four antibodies that detect ITGA6, EPCAM, TNAP, and cKIT. cKIT and TNAP were used as a positive control, given that these surface proteins can be used to sort human PGCs from the prenatal embryonic ovary and testis [25, 29]. The labeled cells were divided in two and sorted by FACS gating on either ITGA6/EPCAM or TNAP/cKIT (Figure 1A). We discovered that the percentage of ITGA6/EPCAM double positive cells and TNAP/cKIT double positive cells in the embryonic ovary was comparable (Figure 1A). Furthermore, the ITGA6/EPCAM double positive population was also double positive for TNAP/cKIT. Conversely, the TNAP/cKIT double positive cells were also double positive for ITGA6/EPCAM (Figure 1A). Using real-time PCR, we show that both populations expressed PGC genes at equivalent levels (Figure 1B). We repeated this experiment with a pair of human embryonic ovaries at 94 day, and found a similar result (Figure 1A and B). Therefore, these observations indicate that ITGA6/EPCAM can be used to isolate TNAP/cKIT positive germ cells from the human embryonic ovary and vice versa.

Next, we performed the same analysis using embryonic testes. Unlike the ovary, we discovered that most ITGA6/EPCAM double positive cells are negative for TNAP/cKIT (Figure 1C), and the ITGA6/EPCAM double positive cells have reduced levels of PGC genes relative to the TNAP/cKIT double positive population (Figure 1D). Given that ITGA6 and EPCAM are epithelial markers [30, 31], and PGCs in the embryonic testis are known to be enclosed within epithelial cords [25], we reasoned that ITGA6 and EPCAM double positive cells are a mixture of both germ cells and somatic cells in prenatal testis. To test this, we performed immunofluorescence and real-time PCR and show that ITGA6 and EPCAM are expressed by both PGCs and epithelial Sertoli cells (Figure S1A and B), and the ITGA6/EPCAM double positive population is also enriched in Sertoli cell genes SOX9 and AMH (Figure 1D). Therefore, using prenatal tissues containing PGCs, TNAP/cKIT can be used to sort PGCs from both the embryonic ovary and testis, whereas ITGA6/EPCAM can only be used to sort PGCs from the embryonic ovary.

Next, we performed RNA-seq on sorted ITGA6/EPCAM female PGCs at 89 and 103 day (Figure 1E and F), and compared this to a third sample involving a pair of ovaries collected at 89 d, which were pooled and then stained and sorted separately for ITGA6/EPCAM or TNAP/cKIT (Figure 1E–G, marked by asterisk). We also included a testis at 59 day, which was sorted only using TNAP/cKIT to specifically isolate the male PGCs. Using unsupervised hierarchical clustering (UHC), the sorted PGC samples clustered together forming a distinct group relative to an undifferentiated female hESC line (UCLA1), and a male hESC line (UCLA2) (Figure 1E). Evaluation of candidate genes revealed that all putative PGC samples expressed the early and late stage PGC genes and have acquired a naïve pluripotent signature (Figure S1C). Moreover, the ITGA6/EPCAM and TNAP/cKIT PGCs sorted from the same pair of 89 day embryonic ovaries clustered the closest compared to ITGA6/EPCAM populations sorted from different ovaries (Figure 1E). This result further supports the conclusion that ITGA6/EPCAM and TNAP/cKIT are expressed on the same population of PGCs. To better display the similarities and differences between samples, we performed PCA, and found that the PGCs are well separated from undifferentiated hESCs along the PC1 axis, while the 59-day male TNAP/cKIT PGCs are separated from the other gonadal PGCs along the PC2 axis (Figure 1F). This separation is anticipated to be caused by the 4-week difference in age between male and female samples in this study, as well as gender [32]. Last, we examined the differentially expressed genes (DEGs) between ITGA6/EPCAM and TNAP/cKIT positive cells sorted from the same ovary pair and found very few DEGs as the two transcriptomes are highly correlated with linear regression R squared = 0.98 (Figure 1G). Therefore, we propose that ITGA6/EPCAM and TNAP/cKIT are coexpressed on the same cell population in the embryonic ovary.

Germline competency is associated with the induction of genes required for primitive streak formation

Prior to examining PGCLC differentiation across 18 independently derived UCLA hESC lines with ITGA6/EPCAM/TNAP/cKIT, we first piloted the sorting strategy on the male hESC line UCLA2. We discovered that all ITGA6/EPCAM double positive putative PGCLCs at day 4 of differentiation were positive for TNAP; however, cKIT was not detected on the PGCLC population (Figure S2A). In mice, cKIT is critical for PGC survival, migration, and proliferation [33–35]. In humans, cKIT is present on PGCs from at least week 4 to week 20 of prenatal life postfertilization [25, 29, 32, 36]. Furthermore, analysis of cyno PGCs during gastrulation reveals that cKIT RNA is expressed by PGCs at the time of specification [3]. Thus, we reasoned that cKIT is either not translated in early stage human PGCs, or alternatively the conditions for PGCLC induction must be optimized to enable cKIT expression on the cell surface. Given that cKIT is subject to ligand induced endocytosis [37], we removed the ligand SCF from the PGCLC media, and this resulted in ITGA6/EPCAM/TNAP/cKIT positive PGCLCs (Figure S2B and D). To determine if excluding SCF affects germ cell identity, we performed real-time PCR and RNA-Seq of the putative ITGA6/EPCAM positive PGCLCs differentiated with or without SCF (Figure S2C and E). To determine the molecular similarity of PGCLCs generated in primed conditions on MEFs, to those generated from hESCs cultured in 4i on MEFs [12], we differentiated the WIS2 hESC line for 4 days according to the methods of Irie et al. [12] and sorted the TNAP/NANOS3-mCherry PGCLC population. RNA-seq analysis showed that PGCLCs generated from either the primed or 4i cultured hESCs clustered together, and were distinct from the undifferentiated hESCs (Figure S2E and F), indicating that the starting culture media (4i on MEFs versus primed media on MEFs) ultimately yielded PGCLCs with similar transcriptional identities.

Using ITGA6/EPCAM as a sorting approach to analyze PGCLCs, we next examined PGCLC competency of 18 hESC lines, all derived at UCLA from 18 single frozen embryos (Figure 2A). All hESC lines were capable of teratoma formation when injected into severe combined immunocompromized (SCID) beige mice (UCLA1–6 [22]; UCLA8–10, UCLA14, UCLA16–18 [23]; and this study UCLA7, UCLA11–13, and UCLA15: Figure S3A–C). Seventeen out of eighteen hESC lines are karyotypically normal, while UCLA7 has a duplication of chromosome 13 in 100% of cells karyotyped (Figure S3B). We included this cell line in the analysis to determine whether aneuploidy may be associated with alterations in PGCLC potential. It has been previously reported that PGCLC induction efficiency is variable between experiments [12, 19]. In order to minimize variability, we induced PGCLCs from 18 hESC lines simultaneously and repeated this experiment four times to determine the average PGCLC induction efficiency for all 18 lines. All 18 independently derived hESC lines were germline competent. However, UCLA6 consistently exhibited the highest germline potential generating on average 35% PGCLCs at day 4 of aggregate formation, whereas UCLA9 had the lowest germline potential generating on average less than 1% PGCLCs at day 4 (Figure 2A; Figure S3D). The aneuploid UCLA7 female hESC line generated a comparable percentage of PGCLCs to the other female hESC lines in the data set. Meanwhile, we found that male hESC lines on average were more competent for PGCLC induction than female (Figure 2B).

In a recent study using nine hiPSC lines, it was determined that primitive streak genes were associated with PGCLC competency [21]. To determine whether this was also the case for hESCs, we performed RNA-seq on the 18 hESC lines, and the corresponding iMeLCs in biological duplicate. We used the following parameters to define genes whose expression levels in iMeLCs are positively correlated with PGCLC induction efficiency. First, we chose genes that were expressed at low levels or not expressed in hESCs by setting the RPKM value < 2. Second, we used the RPKM value > 2 in at least one of the 18 iMeLC samples to select for genes that are upregulated in at least one of the 18 hESC lines induced to become iMeLCs. Third, we correlated gene expression in iMeLCs with PGCLC induction efficiency, and set the cutoff as >0.45. Based on these parameters, we found 78 genes upregulated from hESCs to iMeLCs that were also positively correlated with resulting PGCLC induction efficiency in the aggregates (Figure 2C). Using gene ontogeny (GO) term analysis of these 78 genes by WebGestalt [38, 39], we discovered that the top term was “formation of primary germ layer” (GO: 0001704), which included seven genes: EOMES, T, GATA6, MIXL1, GATA4, WLS, and LHX1. These genes are associated with primitive streak formation that are known to be induced by NODAL/ACTIVIN A (TGFβ) and WNT [40].

Signaling pathways that promote primitive streak formation are also required for PGCLC induction

Next, we confirmed the relationship between TGFβ and WNT signaling pathways, and the expression of T and EOMES in iMeLCs. We performed immunofluorescence and identified nuclear accumulation of phosphorylated SMAD2/3 (pSMAD2/3) (Figure 3A) and β-CATENIN (Figure 3B) in both iMeLCs and hESCs, indicating that these cells are capable of responding to both signaling pathways. We then confirmed expression of T (Figure 3C) and EOMES (Figure 3D) protein by immunofluorescence, and as predicted from the RNA-Seq, T was induced in iMeLCs (Figure 3C), and EOMES protein was expressed in both hESCs and in iMeLCs (Figure 3D).

The correlation of germline competency with TGFβ and WNT signaling, as well as T and EOMES induction promoted us to test if these molecules are critical for PGCLC formation. To block TGFβ signaling we added SB431542 [41], which inhibits the TGFβ type I receptors ALK5, ALK4, and ALK7, and phosphorylation of SMAD2 and SMAD3. This inhibitor does not block the ALKs or SMADs downstream of BMP4. To block WNT receptor binding, we added DKK1 [42] (Figure 3E). We discovered that the addition of these two molecules prevented the induction of T and EOMES in iMeLCs despite the presence of ACTIVIN A and GSK3βi in the iMeLC media (Figure 3F). In order to determine if T and EOMES expression is dependent on TGFβ or WNT, or both, we evaluated T and EOMES expression in iMeLCs in the presence of either SB431542 or DKK1. We found that T and EOMES were repressed in the presence of either SB431512 or DKK1 (Figure S4A and B), suggesting the expression of T and EOMES in iMeLCs requires both TGFβ and WNT signaling. To evaluate the effect on germline competency, we induced PGCLCs in media with or without SB431542 and DKK1 (Figure 3E). We discovered that the PGCLC population was completely abolished when SB431542 and DKK1 were included in the media, despite the presence of exogenous BMP4 and other cytokines known to induce PGCLC fate (Figure 3G). These results suggest that the ability to respond to TGFβ and WNT signaling and potentially the induction of genes associated with primitive streak formation such as T and EOMES are required for germline induction.

EOMES is required for PGCLC induction

To provide direct evidence for EOMES in PGCLC induction, we used CRISPR/Cas9 to mutate EOMES in the UCLA1 hESC line (Figure S4C). We used two independent EOMES mutant clones for analysis. By differentiating the EOMES mutant and control hESC lines in parallel, we found that the ITGA6/EPCAM double positive PGCLC population was dramatically reduced in the EOMES mutant clones (Figure 4A–B). To confirm this result, we also performed immunofluorescence on aggregates and detected PGCLCs by triple staining for TFAP2C, SOX17, and PRDM1. In wild-type control cells, we detected clusters of triple TFAP2C, SOX17, and PRDM1 human PGCLCs (Figure 4C, white dot outlined). However, in the EOMES mutant lines, we identified very few triple positive PGCLCs (Figure 4C, white arrowheads). Therefore, our results demonstrate that EOMES is required for PGCLC formation in human, most likely downstream of WNT and TGFβ.

In cyno embryos, nascent PGCs in the amnion are negative for EOMES [3]. Similarly, our RNA-seq data show that EOMES is not expressed in prenatal PGCs or PGCLCs. Therefore, we hypothesize that EOMES acts either noncell autonomously in the niche to support PGCLC formation, or alternatively EOMES acts cell autonomously in iMeLCs to prime PGC fate. To test this, we made stable GFP lines of control and EOMES mutant hESCs. We made iMeLCs and generated PGCLC aggregates containing a mix of GFP and unlabeled cells in a 1:1 ratio. The iMeLC mixtures were composed of GFP-labeled control cells with unlabeled controls (Figure 4D) or GFP-labeled EOMES mutant cells with unlabeled controls (Figure 4E). In the controls, about 45% of the total cells were positive for GFP, and about 35% of the PGCLCs were positive. Therefore, control hESCs with or without GFP both contribute to PGCLC induction (Figure 4D). In contrast, only about 8% of the PGCLCs were induced from GFP-labeled EOMES mutant cells, whereas the total cells were composed of about 48% GFP-labeled EOMES mutant cells (Figure 4E). This suggests that EOMES is required cell autonomously to prime PGC fate.