The TEA domain transcription factors TEAD1 and TEAD3 and WNT signaling determine HLA-G expression in human extravillous trophoblasts

Significance

This study significantly advances the understanding of maternal–fetal immune tolerance by identifying key regulators of HLA-G expression in extravillous trophoblasts (EVT). Using a genome-wide CRISPR-Cas9 screen, the WNT pathway was found to negatively regulate HLA-G, while the transcription factors TEAD1 and TEAD3 were shown to be essential for HLA-G transcription. These findings not only deepen our understanding of EVT immune modulation during pregnancy but also offer potential strategies for manipulating HLA-G expression in cancer therapy and transplant immunology.

Abstract

Maternal–fetal immune tolerance guarantees a successful pregnancy throughout gestation. HLA-G, a nonclassical human leukocyte antigen (HLA) molecule exclusively expressed in extravillous trophoblasts (EVT), is a crucial factor in establishing maternal–fetal immune tolerance by interacting with inhibitory receptors on various maternal immune cells residing in the uterus. While trophoblast-specific cis-regulatory elements impacting HLA-G transcription have been described, the identity of trans-acting factors controlling HLA-G expression in EVT remains poorly understood. Utilizing a genome-wide CRISPR-Cas9 knockout screen, we find that the WNT signaling pathway negatively regulates HLA-G expression in EVT. In addition, we identified two trophoblast-specific transcription factors, TEAD1 and TEAD3, required for HLA-G transcription in EVT in a Yes-associated protein-independent manner. Altogether, we systematically elucidated essential genes and pathways underlying HLA-G expression in EVT, shedding light on the mechanisms of maternal–fetal tolerance and potentially providing insights into controlling HLA-G expression beyond EVT to protect allogeneic cells from immune rejection.

A fundamental and still largely unresolved question in reproductive biology is how the semiallogeneic fetus is accommodated by the maternal immune system throughout pregnancy without facing immune rejection. An intricate coordination between invading extravillous trophoblasts (EVT) and decidual immune cells balances tolerance and immunity at the maternal–fetal interface. EVT induce immune tolerance by expressing a unique set of human leukocyte antigen (HLA) molecules, namely the classical MHC class Ia HLA-C and the nonclassical MHC class Ib HLA-E, HLA-F, and HLA-G (HLA-G), but not the classical polymorphic HLA-A and HLA-B molecules (1–3). HLA-G was first discovered in 1982 through Southern blot hybridization as an unconventional HLA class I gene on human chromosome 6 (4). The gene encoding HLA-G was later isolated from a genomic DNA library constructed from the B-LCL 721 cell line and initially named HLA-6.0 (5). Under normal physiological conditions, HLA-G expression is restricted to cells in immunologically privileged sites, including placental trophoblasts (6) and thymic epithelial cells (7). HLA-G plays a pivotal role in immune tolerance induction at the maternal–fetal interface during pregnancy by supporting EVT invasion and migration for proper placental development. Aberrant HLA-G expression has been associated with defective EVT differentiation, leading to placental insufficiencies such as preeclampsia and intrauterine growth restriction (1, 6). In addition, emerging evidence supports the importance of HLA-G in pathologic conditions, such as autoimmunity, transplant rejection, and tumor immune evasion (8–12).

HLA-G exerts its immunomodulatory function by interacting with receptors on immune and immunomodulatory cells, such as Ig-like transcript (ILT)-2 and ILT-4 expressed on multiple immune cells; killer cell Ig-like receptor 2DL4 on decidual NK (dNK) cells; CD8 on T cells and NK cells; and CD160 on endothelial cells (13, 14). Other than directly binding to cell surface receptors, HLA-G can also induce immune tolerance by being acquired by dNK in soluble form (sHLA-G) or via trogocytosis (15, 16). Previously, our group identified an EVT-specific cis-regulatory element 12 kb upstream of HLA-G that is essential for HLA-G expression named Enhancer L (17). The identity of trans-acting factors controlling the EVT-specific expression of HLA-G, however, remained to be uncovered.

In this study, we employed a pooled genome-wide CRISPR-Cas9 knockout (KO) screen to uncover and characterize the regulatory network underlying HLA-G expression in the EVT model cell line JEG-3. We found that WNT signaling negatively regulates HLA-G expression. In addition, two trophoblast-specific transcription factors (TFs), TEAD1 and TEAD3, were identified to transactivate HLA-G expression by directly acting on the HLA-G promoter. Interestingly, this transactivating capacity seems to be independent of Yes-associated protein (YAP), a transcriptional coactivator that forms complexes with TEAD TFs for regulating genes involved in cell growth and proliferation.

Results

A Genome-Wide CRISPR-Cas9 KO Screen Identifies HLA-G Modulators in JEG-3.

In order to systematically identify regulators of HLA-G expression in EVT in an unbiased way, we transduced JEG-3, a choriocarcinoma cell line widely used as an EVT model, with the Brunello human CRISPR KO pooled lentiviral library, which contains 76,441 single guide RNAs (sgRNAs) targeting 19,114 genes (18). Following lentiviral transduction and puromycin selection, JEG-3 cells were stained with a highly specific anti-HLA-G antibody and sorted into four populations using fluorescence-assisted cell sorting: the dimmest 20%, next dimmest 20%, brightest 20%, and next brightest 20% of HLA-G expression level (Fig. 1A), aiming to identify genes that promote or suppress HLA-G expression based on its cell surface levels. This screening approach was first benchmarked in either WT or β2 microglobulin (B2M) KO JEG-3 cells. B2M is an essential component of MHC class I molecules and required for HLA-G surface trafficking. As expected, HLA-G surface expression was undetectable in B2M KO cells, as determined by flow cytometry (Fig. 1B). Sorted cells transduced with the CRISPR library were then analyzed by next-generation sequencing, followed by data analysis using the MAGeCK RRA pipeline (19), to identify genes significantly affecting HLA-G expression (Fig. 1A). Four biological replicates with independent transduction were performed and showed good reproducibility (Fig. 1C). MAGeCK analysis identified 30 genes whose disruption significantly altered HLA-G cell surface levels (false discovery rate <0.1) (Dataset S1).

Fig. 1.

A genome-wide CRISPR-Cas9 KO screen identifies regulators and pathways essential for HLA-G expression in JEG-3. (A) Schematic of HLA-G regulator CRISPR-Cas9 KO screening. (B) Flow cytometric analysis of HLA-G levels in WT JEG-3 cells (blue) or B2M KO JEG-3 (orange). (C) Spearman correlation and P-value of the ratio of representation of all sgRNAs presorting. (D) Ranked dot plots of gene enrichment analyzed by dividing top by bottom sorted bins. The x axis shows the rank of each gene, the y axis shows the log2 enrichment of sgRNAs for each gene in the Top compared with the Bottom population. Several significant genes are listed. (E) Coessential modules of the most significantly enriched CRISPR-KO genes (hypergeometric P-value) labeled with the top 10 significantly associated gene ontology (GO) terms. (F) STRING protein–protein interaction (PPI) network of the top 100 genes depleted in the HLA-G screen. The minimum required interaction score was set to medium (0.4). k-means clustering was applied. Thickness denotes the STRING PPI score/confidence.

We performed a sgRNA enrichment analysis by dividing the sgRNA counts from the HLA-G bright cells by those obtained from the HLA-G dim cells (Materials and Methods). As expected, HLA-G ranked 1st among the most depleted genes. NPHP3 was identified as the most enriched gene in the screen, representing the most plausible inhibitor of HLA-G expression (Fig. 1D). NPHP3 plays a crucial role in early embryogenesis through modulating canonical and noncanonical WNT signaling (20). The screen also identified multiple HLA class I antigen-processing machinery (APM) components among the depleted genes, including transporters associated with antigen processing 1 and 2 (TAP1/2) and TAP binding protein (Fig. 1D). As our screen readout is HLA-G cell surface expression, these well-referenced APM proteins act as positive controls of a successful screen.

Gene Module Analysis Reveals Pathways Associated with HLA-G Expression.

To examine whether significantly changed sgRNAs were enriched in specific gene sets or function through similar biological pathways, we performed a coessentiality analysis, which groups genes into functionally coherent modules, utilizing genome-wide CRISPR cell essentiality screen datasets (DepMap) (21) (Dataset S2). The most significantly associated modules contained genes associated with cell adhesion, ribosome biogenesis, mitotic cell cycle, antigen processing, mRNA tail shortening, negative regulation of WNT signaling pathway, epigenetic regulation of gene expression, G protein–coupled purinergic nucleotide receptor signaling pathway, protein palmitoylation, and postembryonic digestive tract morphogenesis (Fig. 1E). Because we used HLA-G cell surface level as readout in our screen, cell adhesion, mitosis, ribosome biogenesis, and cell cycle modules enriched in the HLA-G negative cell populations were regarded as nonspecific effects, while “Antigen processing and presentation of endogenous peptide antigen” was regarded as a positive control. Of note, 5 out of 5 genes in the “Negative regulation of WNT signaling pathway” module, namely Casein Kinase 1 Alpha 1 (CSNK1A1), AXIN1, APC Regulator Of WNT Signaling Pathway (APC), Catenin Beta Interacting Protein 1 (CTNNBIP1), and Zinc And Ring Finger 3 (ZNRF3) were all significantly enriched in the HLA-G negative cell populations, suggesting that inhibition of the WNT pathway is required for proper HLA-G expression in EVT, despite the fact that WNT signaling is important during embryonic development and for adult tissue homeostasis.

STRING PPI network analysis of the top 100 negatively enriched hits in our screen further revealed protein networks involved in antigen processing and presentation of endogenous peptide antigen, MHC Class II deficiency, CCAAT-binding factor complex, rRNA modification in the nucleus and cytosol, RHO GTPases Activate WASPs and WAVEs, Cell-extracellular matrix interactions, Hippo/WNT signaling pathway, Interleukin-7 signaling, and SAGA complex as the most significant modules (Fig. 1F). Thus, with gene module analysis and STRING PPI network analysis, we uncovered WNT signaling as a potential repressive pathway negatively regulating HLA-G expression.

WNT Pathway Negatively Regulates HLA-G Expression.

Negative regulation of the WNT signaling pathway is significantly enriched by coessentiality analysis of our CRISPR screen. Thus, we next focused on individual hits within this pathway. ZNRF3, a transmembrane E3 ubiquitin ligase known to inhibit WNT signaling by degrading the WNT receptor frizzled, ranked after HLA-G as the second most enriched sgRNAs in HLA-G dim cells. Cytoplasmic APC/Axin Destruction Complex component gene AXIN1, Glycogen Synthase Kinase 3 Beta (GSK3B), and APC were also significantly enriched in HLA-G dim cells. Moreover, NOTUM, an extracellular carboxylesterase that limits WNT activity (22, 23) and has a highly EVT-specific expression pattern, according to single-cell transcriptomic data in first-trimester placenta (24) (Fig. 2A), was recently shown to promote and maintain EVT cell differentiation (25). Those data suggest that WNT inhibition is required for HLA-G expression in EVT.

Fig. 2.

WNT signaling negatively regulates HLA-G in JEG-3 cells. (A) Visualization of WNT signaling antagonist NOTUM single-cell gene expression in first-trimester placenta (24). (B) Representative flow cytometry histograms (Left) and relative HLA-G MFI (Right) in JEG-3 cells treated with WNT pathway activator CHIR99021 at different concentrations (0.5 µM, 1 µM, 5 µM) or DMSO as a control. (C) Representative flow cytometry histograms (Left) and HLA-G MFI (Right) in JEG-3 cells treated with WNT pathway inhibitor XAV-939 or DMSO as a control. (D) Relative mRNA level of HLA-G in JEG-3 treated with DMSO, CHIR99021 at different concentrations or XAV-939.

To further substantiate that the WNT pathway negatively regulates HLA-G expression, JEG-3 cells were treated with the WNT/β-catenin activator CHIR99021, which targets GSK-3α and GSK-3β. HLA-G cell surface levels were decreased after CHIR99021 treatment for 48 h in a dose-dependent manner, as determined by flow cytometry (Fig. 2B). Conversely, after treatment with a WNT/β-catenin inhibitor XAV-939 for 48 h, a moderate upregulation of HLA-G in JEG-3 cells was observed on the protein level (Fig. 2C). Interestingly, those effects were more pronounced at the transcriptional level when assessed by RT-qPCR (Fig. 2D), further suggesting that the WNT pathway regulates HLA-G at the transcriptional level.

APC KO in JEG-3 Impairs HLA-G Expression.

APC is a key component of the destruction complex for cytoplasmic β-catenin, switching off the canonical WNT pathway (26, 27). To induce constitutively elevated WNT pathway activity, we generated an APC KO JEG-3 cell line using CRISPR-Cas9 (Fig. 3A). HLA-G, but not HLA-C or HLA-E, expression was significantly downregulated upon APC KO at the mRNA (Fig. 3B) and protein (Fig. 3C) level. We then performed RNA sequencing of the APC KO JEG-3 cells. Upon inspecting a set of EVT marker genes (24), most of these EVT signature genes were downregulated in APC KO cells (SI Appendix, Fig. S1), suggesting that APC is essential for EVT identity maintenance. Of note, 3 of the 10 most significantly enriched GO terms in the down-regulated genes in APC KO cells were related to the WNT signaling pathway (Fig. 3D), confirming that components of the WNT signaling pathway were indeed affected by APC depletion. We further examined 13 WNT target genes commonly seen in human diseases, such as human colon cancer; MYC, SOX2, CCND1, and AXIN2 were all upregulated upon APC deletion in JEG-3 cells (Fig. 3E). Taken together, these findings suggest that APC is required for HLA-G expression, and WNT signaling negatively regulates HLA-G expression in EVT.

Fig. 3.

HLA-G expression is impaired by APC KO. (A) Schematic of CRISPR-Cas9 KO strategy of APC in JEG-3 cells and Sanger sequencing validation of edited region of the single clone. (B) Relative mRNA level of HLA-G, HLA-C, HLA-E in WT, and APC^−/− JEG-3 cells. (C) Western blot of APC and HLA-G in WT or APC^−/− JEG-3 cells, β-actin is a loading control. (D) The top 10 gene ontology GO terms of the downregulated genes in APC^−/− JEG-3 cells. (E) Heatmap of the WNT target genes in WT or APC^−/− JEG-3 cells plotted with Z-scores, each with three replicates.

TEAD1 and TEAD3 Are Trophoblast-specific TFs Required for HLA-G Expression.

To understand how HLA-G expression is transcriptionally regulated, we mined our data for TFs that directly regulate HLA-G transcription. Among the top 500 genes enriched in the HLA-G negative population resulting from our pooled CRISPR screen, 19 TFs were identified (Fig. 4A). First, we examined the expression of these TFs at the single-cell level in a publicly available dataset of human first-trimester placenta (24). Among the 19 TFs, ELF4, TEAD1, and TEAD3 showed a highly EVT-specific expression pattern (Fig. 4A). ELF4 was present at extremely low levels in JEG-3 cells by RNA-seq, and we thus mainly focused on the two TEA domain (TEAD) family TFs, TEAD1 (ranked #32 in the list of genes depleted in the CRISPR screen) and TEAD3 (ranked #18 in the list of genes depleted in the CRISPR screen). TEAD1 was recently identified as an essential transcription factor for trophoblast stem cell to EVT differentiation (28). Chromatin landscape profiling of EVT differentiation also highlighted a potential role for TEAD family members as transcriptional regulators of the EVT lineage (29). By analyzing the expression profile of all four members of the TEAD gene family at early pregnancy, we found that TEAD3 showed the highest expression level in all three trophoblast subsets, while TEAD1 is slightly less expressed in EVT than TEAD3 but with higher specificity (SI Appendix, Fig. S2A and Fig. 4B). To validate whether TEAD1 and TEAD3 are indeed required for HLA-G expression, we generated TEAD1 and TEAD3 KO JEG-3 cells, picked two independent single cell-derived clones for each KO, and verified respective protein loss by Western blot (Fig. 4 C and D). HLA-G expression in TEAD1 KO and TEAD3 KO cells was then assessed at both the cell surface protein and mRNA levels. Strikingly, HLA-G was downregulated by 35% in TEAD1 KO cells and 58% in TEAD3 KO cells, as determined by flow cytometry (Fig. 4E), and downregulated by 79 and 77%, respectively, at the mRNA level (Fig. 4F). Of note, restoring TEAD3 expression in TEAD1 or TEAD3-deficient cells reversed their HLA-G reduction phenotype (SI Appendix, Fig. S2B). These results revealed an indispensable role of TEAD1 and TEAD3 in maintaining HLA-G levels in EVTs.

Fig. 4.

TEAD1 and TEAD3 transactivate HLA-G. (A) Bubble plot, single-cell expression of the top 19 transcription factor genes of the CRISPR screen in first-trimester placenta (24). fFB, fetal fibroblasts; DC, dendritic cells; MO, Monocytes; HB, Hofbauer cells; dNK; VCT, villous cytotrophoblast; Epi, epithelial glandular cells; dP, decidual perivascular cells; Endo, endothelial cells; dM, decidual macrophages; SCT, syncytiotrophoblast; EVT; dS, decidual stromal cells. (B) Violin plot of TEAD1, TEAD2, TEAD3, and TEAD4 expression in EVT, SCT, and villous cytotrophoblast trophoblast subsets derived from the first-trimester pregnancy single-cell RNA-seq dataset (24) (C) Western blot of TEAD1 protein in WT or TEAD1^−/− single cell-derived clones and (D) TEAD3 protein in WT or TEAD3^−/− JEG-3 single cell-derived clones, β-actin is a loading control. (E) HLA-G MFI of flow cytometry in WT, TEAD1^−/−, TEAD3^−/− JEG-3 cells. (F) Relative HLA-G mRNA level in WT, TEAD1^−/−, TEAD3^−/− JEG-3 cells.

TEAD1 and TEAD3 Transactivate HLA-G Independently of YAP.

To understand how TEAD1 and TEAD3 affect HLA-G expression, we first checked whether TEAD1 or TEAD3 DNA binding motifs are present in the HLA-G promoter and the Enhancer L region. Motif analysis identified six TEAD3 binding sites and one TEAD1 binding site within the 801 bp promoter region upstream of the HLA-G transcription start site (TSS), but not in the 12 kb upstream Enhancer L region (Fig. 5A). Firefly luciferase activity driven by the 801 bp promoter region of HLA-G was measured in WT, TEAD1, and TEAD3 KO JEG-3 cells. Luciferase activity decreased by nearly half in TEAD1 and TEAD3 KO cells compared to WT cells, suggesting that both TEAD1 and TEAD3 bind to the HLA-G promoter region and transactivate HLA-G (Fig. 5B).

Fig. 5.

TEAD1 and TEAD3 transactivate HLA-G independently of YAP. (A) Six TEAD3 motifs (green) and one TEAD1 motif (orange) were identified within the 801 bp promoter region upstream of the HLA-G TSS. (B) Luciferase reporter assay of luciferase expression driven by 801 bp HLA-G promoter in WT, TEAD1^−/−, or TEAD3^−/− JEG-3 cells. (C) Representative flow cytometry histograms of HLA-G in WT (C) or TEAD3^−/− (D) JEG-3 cells treated with YAP inhibitors CIL-56 and Verteporfin (0.5 µM) or DMSO as a control for 24 h. (E) Working model of HLA-G gene regulatory network in EVT.

TEAD proteins are Hippo pathway components which are usually found interacting with YAP and TAZ for downstream gene transcription (30). To test whether a YAP/TEAD complex is required for HLA-G transcription, we treated JEG-3 with 0.5 µM Verteporfin or CIL-56, both of which are YAP inhibitors that specifically disrupt YAP–TEAD interactions, and evaluated HLA-G expression by flow cytometry. No obvious changes in HLA-G expression levels were observed after 24 h YAP inhibitor treatment in either WT or TEAD3^−/− JEG-3 cells (Fig. 5 C and D). HLA-G expression was not affected by the YAP inhibitors (SI Appendix, Fig. S2C) while the expression of two YAP/TEAD target genes, CTGF and CRY61, was reduced after 24 h treatment with the YAP inhibitors (SI Appendix, Fig. S2D), suggesting that TEAD1 and/or TEAD3 transactivate HLA-G independently of YAP. In light of our combined observations, we propose the following HLA-G regulatory network in EVT (Fig. 5E): WNT signaling is suppressed by APC in the cytoplasm in EVT; in the nucleus, TEAD1 and TEAD3 transactivate HLA-G by binding to its promoter independent of YAP; Translated HLA-G protein is processed and presented on the cell surface in the presence of TAP proteins.

Discussion

HLA-G is a unique HLA molecule that is highly expressed in EVT cells. HLA-G plays a key role during pregnancy as a mediator of maternal–fetal immune tolerance, binding to various inhibitory receptors expressed across immune cell types either directly or following acquisition by trogocytosis (14, 16). Unlike the polymorphic HLA class Ia genes that are broadly expressed in somatic tissues, the constitutive expression of HLA-G is restricted to trophoblasts and a subset of thymic epithelial cells, suggesting a tight transcriptional control mechanism. HLA class I gene transactivation is broadly mediated by three main promoter elements: enhancer A, ISRE, and SXY motif. However, the HLA-G promoter sequence has a divergent enhancer A and SXY module and lacks the ISRE module (31). Consequently, HLA-G is unresponsive to IFN-γ/IFN-β induction or TNF-α signaling, with the HLA-G promoter lacking binding capacity for key HLA gene regulation-associated TFs, such as ATF1/CREB1, RFX, as well as NLRC5 and CIITA, the master transcriptional regulators of HLA class I and class II gene expression, respectively (32, 33).

In fact, neither NLRC5 nor CIITA are expressed in EVT (34). Ferreira et al. previously revealed that chromatin looping mediated by members of the CEBP and GATA family of TFs between Enhancer L and the HLA-G classical promoter is crucial for placenta-specific HLA-G expression (17). Other elements upstream of the HLA-G promoter may also regulate HLA-G transcription, such as a locus control region identified 1 kb upstream of HLA-G (35).

In this study, TEAD1 and TEAD3 binding sites were identified within an 801 bp region of the HLA-G promoter. Luciferase assays demonstrated that both TEAD1 and TEAD3 play positive regulatory roles in HLA-G transcription via the HLA-G promoter (Fig. 5B). Of note, TEAD1 and TEAD3 were not redundant in EVT, as deleting either one of them alone led to a significant decrease in HLA-G expression at the mRNA and protein level (Fig. 4 E and F) and diminished HLA-G promoter activity (Fig. 5B). TEAD TFs play an important role in regulating progenitor proliferation, stem cell identity, and lineage specification (36, 37). However, their involvement in trophoblast lineage determination was understudied until recently, with TEAD4 being implicated in trophoblast progenitor renewal (28, 29, 38, 39). The expression patterns of TEAD1 and TEAD3 are highly EVT-specific (Fig. 4 A and B), with TEAD3 expression recently described as overlapping best with HLA-G expression in single-cell RNA sequencing data of first-trimester human placenta (39).

TEAD proteins are well-studied components of the Hippo signaling pathway, interacting with transcriptional cofactors such as YAP, transcriptional coactivator with PDZ-binding motif (TAZ), and VGLL family proteins. Up to 80% of TEAD-bound motifs are co-occupied with YAP/TAZ across cell types (40–42). We found that TEAD1 and TEAD3 are required for optimal HLA-G expression potentially in a YAP-independent manner. Consistent with our observations, a recent study showed that VGLL2 and TEAD1 recruit the histone acetyltransferase p300 and transactivate downstream target genes independently of YAP/TAZ (43). Curiously, we found TEAD binding sites in the HLA-G promoter, but not in Enhancer L, even though genome-wide mapping of TEAD binding sites in a multitude of cell types has shown that they occur mostly in enhancers, with only 5% of TEAD binding sites found in gene promoters (41, 44). The cofactors that partner with TEAD1 and TEAD3 to regulate HLA-G via its core promoter in a YAP-independent fashion in EVT remain unknown and warrant further exploration.

The WNT pathway is a highly conserved network of molecular interactions crucial in a plethora of developmental and homeostatic processes. WNT signaling involves stabilization of beta-catenin, which otherwise is continuously degraded by a destruction complex that includes APC (26). Growing evidence suggests a negative correlation between activity of the WNT signaling pathway and EVT lineage specification. Loss of WNT activators has been shown to promote formation of EVT and EVT progenitor cells in trophoblast organoid cultures (45). We found that in the APC deficient, hence with constitutive WNT signaling, JEG-3 cells, a subset of WNT targets initially described in human colon cancer (MYC, SOX2, CCND1, and AXIN2) were upregulated, representing an EVT-specific pattern of WNT targets regulated by APC. Yet, T-cell factor/lymphoid enhancer factor TFs, which are major endpoint mediators of the WNT signaling pathway, were not significantly enriched in our CRISPR screen. Of note, EVT identity-associated gene expression was markedly altered upon APC ablation (SI Appendix, Fig. S2). Thus, WNT signaling might indirectly regulate HLA-G expression by antagonizing EVT cell identity rather than or in addition to directly acting on the HLA-G promoter.

Here, we find that constitutive WNT pathway activity, achieved by deleting APC, results in near ablation of specifically HLA-G expression, leaving HLA-C and HLA-E unaffected (Fig. 3B). TEAD1 and TEAD3 are downstream of WNT signaling and activate HLA-G expression. Yet, TEAD1 also induces the expression of WNT-negative regulatory genes, such as DKK1, WNT5A, and WNT5B (46). A mechanism by which TEAD1 and TEAD3 induce HLA-G expression may thus be WNT signaling inhibition. Intriguingly, progesterone, which upregulates HLA-G expression in trophoblasts via a nonclassical membrane progesterone receptor (47), also induces DKK1 expression (48). Hence, progesterone’s stimulatory impact on HLA-G expression in EVT may also involve WNT signaling inhibition.

In the mouse intestine, YAP has been shown to induce an epithelial cell state marked by low Wnt signaling, a wound-healing transcriptional response, and expression of the transcription factor Klf6 (49, 50). Interestingly, our unbiased CRISPR screen also uncovered KLF6 (Fig. 4A) and inhibited WNT signaling (Fig. 1E) as being associated with HLA-G expression in EVT. In addition, STRING PPI analysis of our CRISPR screen hits (SI Appendix, Fig. S1) grouped WNT and Hippo pathways together as one module, indicating a potential convergence of these two signaling pathways on HLA-G expression. The interplay between the WNT and the Hippo pathways, the wound-healing response, progesterone, and HLA-G regulation in EVT lineage specification are fascinating avenues to be explored further.

One limitation of our study in identifying transcriptional regulators of HLA-G is that the CRISPR screen readout was HLA-G protein surface level detected by flow cytometry. Common regulators in cell surface protein processing and trafficking could thus mask the effect of TFs regulating expression of HLA-G. Future CRISPR screens using a more focused transcription factor guide RNA library and/or an endogenous HLA-G knock-in reporter gene screen could potentially increase the sensitivity toward uncovering a more comprehensive transcriptional regulatory network. Another limitation is that the JEG-3 cells used for screening are a rapidly proliferating choriocarcinoma cell line, making the assessment of WNT and Hippo pathway activities more challenging. Therefore, it is critical that our results are confirmed in primary EVT cells in future studies.

Overall, our study shed light on a decades-old question by systematically unraveling TFs and signaling pathways that specifically regulate HLA-G expression. We revealed that the EVT-restricted TFs TEAD1 and TEAD3 transactivate HLA-G by binding to its promoter and that WNT signaling inhibition is required specifically for HLA-G expression in EVT. These findings elucidate not only how HLA-G expression is regulated in EVT, but potentially also how to manipulate HLA-G expression in other tissues for treating cancer or organ transplant rejection.

Materials and Methods

Details of Cell culture and transfection; CRISPR-KO screening; Individual candidate gene validation; RNA extraction and RT-qPCR quantification; Flow cytometry; Western blot; Bulk RNA sequencing; Cell clustering and visualization of single-cell RNA sequencing data; Luciferase Reporter Gene Assays; and Quantification and statistical analysis are given in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Data, Materials, and Software Availability

RNA-seq data for Fig. 3 are available in the Gene Expression Omnibus database (Accession no. GSE273945) (51). Single-cell RNA-seq data from early human embryonic development and placenta were retrieved from E-MTAB-6701 (52) and E-MTAB-6678 (53). The antibodies, compounds, and oligonucleotides used in this study can be found in SI Appendix. All other data are included in the manuscript and/or supporting information.