Genome-wide screening identifies Polycomb repressive complex 1.3 as an essential regulator of human naïve pluripotent cell reprogramming

Abstract

Uncovering the mechanisms that establish naïve pluripotency in humans is crucial for the future applications of pluripotent stem cells including the production of human blastoids. However, the regulatory pathways that control the establishment of naïve pluripotency by reprogramming are largely unknown. Here, we use genome-wide screening to identify essential regulators as well as major impediments of human primed to naïve pluripotent stem cell reprogramming. We discover that factors essential for cell state change do not typically undergo changes at the level of gene expression but rather are repurposed with new functions. Mechanistically, we establish that the variant Polycomb complex PRC1.3 and PRDM14 jointly repress developmental and gene regulatory factors to ensure naïve cell reprogramming. In addition, small-molecule inhibitors of reprogramming impediments improve naïve cell reprogramming beyond current methods. Collectively, this work defines the principles controlling the establishment of human naïve pluripotency and also provides new insights into mechanisms that destabilize and reconfigure cell identity during cell state transitions.

Fig. 1.. CRISPR-Cas9 knockout screen defines new regulators of naïve PSC reprogramming.

(A) Screening strategy to identify regulators of naïve PSC reprogramming. Cas9-expressing primed PSCs were transduced with a gRNA library targeting the exons of 18,365 genes. The library also contained 1004 gRNAs targeting negative control regions. Puromycin-selected transduced cells were reprogrammed to naïve pluripotency under 5i/L/A conditions for 10 days. The cell population was cell-sorted using a panel of cell surface markers into nascent naïve cells (successfully reprogrammed) and refractory cells (not reprogrammed). High-throughput sequencing counted gRNAs in the sorted cell populations. (B) Scatterplot shows the counts of each gRNA in the nascent naïve and refractory cells. Each gene is targeted by an average of six gRNAs. Red, gRNAs that target impediment genes; blue, gRNAs that target essential genes; purple, negative control gRNAs; gray, all other sgRNAs. (C) Ranked differential enrichment (DE) plots by comparing nascent naïve with refractory populations. Numbers of identified impediment and essential genes are shown (applying a cutoff of P < 0.02, permutation test) together with the percent out of all genes targeted. (D) MA plot shows that the majority of essential and impediment genes are expressed at similar levels in nascent naïve and refractory cells. Transcriptional data are shown for each gene (represented by a dot), comparing nascent naïve and refractory cell populations; dark gray, differentially expressed; light gray, all other genes. Essential and impediment genes are colored in blue and red, respectively.

Fig. 2.. Analysis of essential genes identifies new roles for PRC1.3 and SAGA complexes in naïve PSC reprogramming.

(A and B) Venn diagrams show the overlap between genes that are essential for naïve PSC reprogramming and genes that are essential for (A) fibroblast to primed iPSC reprogramming (34) and (B) primed PSC proliferation. (C) Charts show the adjusted P value for essential genes in the biological processes gene ontology (GO) category (Fisher’s exact test). (D) Ranked DE plot by comparing nascent naïve with refractory populations. Essential genes are highlighted. The top 15 ranked essential genes are shown; SAGA components, purple; PRC1.3 components, green. (E and F) Charts show the P values of (E) PRC1 members and (F) SAGA complex members, as a measure of their depletion in the nascent naïve cell population following the CRISPR-Cas9 screen. The red line indicates P = 0.02 as a cutoff (permutation test). Schematics of the complexes are also shown; bold lines indicate components classified as an essential gene.

Fig. 3.. Defining PRC1.3 as an essential regulator of naïve PSC reprogramming.

(A) Parental WT and PCGF3 KO PSCs at days 2 and 10 of naïve PSC reprogramming (5i/L/A). Images are representative of three independent experiments. Scale bars, 100 μm. (B) Immunofluorescent images of WT and PCGF3 KO cells at day 10 of naïve PSC reprogramming (5i/L/A). OCT4 and NANOG are expressed in naïve and primed PSCs; KLF17 is naïve specific. Scale bar, 400 μm. (C) Classification of cell colonies from (B). Number of counted colonies (n). The proportion of colony type (naïve, primed, or differentiated) is dependent on whether the culture is WT or PCGF3 KO (chi-square test; 2 df; P < 0.0001 for WT versus each KO). (D) Flow cytometry of parental WT and PCGF3 KO PSCs at day 14 of naïve PSC reprogramming (5i/L/A). Reprogrammed naïve cells in bottom right quadrant. Data are representative of three independent experiments. (E) RNA-seq principal components analysis (PCA) for parental WT and PCGF3 KO PSCs at days 0 and 10 of naïve PSC reprogramming (5i/L/A). Samples from bulk cell populations. (F) MA plot of RNA-seq compares WT and PCGF3 KO cell populations at day 10 of naïve PSC reprogramming (5i/L/A). Differentially expressed genes (DEG), blue (FDR < 0.05; Wald test with Benjamini-Hochberg correction). Genes higher in WT include naïve PSC and pan-pluripotency markers. Genes higher in KO include neural markers. Data show average of two (WT) or three (KO) biological replicates. (G) Expression of pan-pluripotency gene OCT4 and naïve PSC markers KLF4 and DPPA3. Error bars, SEM; circles show biological replicates (n = 2 for WT day 10; n = 3 for all other samples). (H) Flow cytometry of parental WT, PCGF3 heterozygous (Het), PCGF3 KO, and PCGF3 KO expressing PCGF3 transgene (Rescue) PSCs at day 10 of naïve PSC reprogramming (CR conditions). Data are representative of two KO lines and two independent experiments.

Fig. 4.. PRC1.3 represses target genes to ensure naïve PSC reprogramming.

(A) Heatmaps of normalized PCGF3 and RING1B ChIP-seq read counts within a 20-kb transcriptional start site (TSS)–centered window in naïve and primed PSCs. Regions were subsetted into naïve-specific PRC1.3 sites (n = 204), shared PRC1.3 sites (n = 19), and primed-specific PRC1.3 sites (n = 205). Genes bound by PRC1.3 only in naïve PSCs are enriched for transcriptional regulators, and several example categories and genes are shown. (B) ChIP-seq genome browser tracks over the LETM1 locus exemplifies a change in PRC1.3 (PCGF3 and RING1B) occupancy between naïve and primed PSCs. (C) Violin plot shows the fold change (FC) in the expression of PRC1.3 target genes (n = 191) in naïve compared to primed PSCs (average values of three biological replicates for each cell type). Central line, median. (D) Violin plots show the FC in the expression of PRC1.3 target genes (n = 191) between WT and PCGF3 KO cells at day 0 compared to day 10 of naïve reprogramming (5i/L/A conditions; n = 2 for WT day 10; n = 3 for all other samples). P values were calculated using a Kruskal-Wallis test with Dunn’s multiple comparison test. Central line, median. (E) Charts show log₂ RPKM (Reads Per Kilobase of transcript per Million mapped reads) for two PRC1.3 target genes in WT and PCGF3 KO cell populations at days 0 and 10 of naïve PSC reprogramming. Error bars, SEM; circles show each biological replicate (n = 2 for WT day 10; n = 3 for all other samples). (F) Violin plots show the FC in the expression of PRC1.3 target genes (n = 191) in cell-sorted populations at day 10 of naïve reprogramming [5i/L/A conditions; data from (32)]. Average of three biological replicates for each sample. P values were calculated using a Kruskal-Wallis test with Dunn’s multiple comparison test. Central line, median. n.s., not significant.

Fig. 5.. FBRS and AUTS2 paralogs switch during human development.

(A) Expression of PRC1.3 components at day 0 (primed PSCs), day 10 (nascent naïve PSCs), and day 34 (established naïve PSC lines). Data from (32). (B) FC in abundance of PCGF3-interacting proteins when comparing naïve and primed PSCs. PRC1.3 components are labeled. Adjusted P < 0.1, red dots; adjusted P > 0.1, gray dots (Limma test with Benjamini-Hochberg correction). Biological replicates: n = 4 primed; n = 2 naïve. (C) Relative abundance of PRC1.3 components in naïve, 48-hour 5i/L/A, and primed PSCs, measured using qPLEX-RIME analysis of PCGF3-interacting proteins. Mean with SD; biological replicates: n = 4 primed; n = 4 48-hour 5i/L/A; n = 2 naïve. (D) Expression of PRC1.3 proteins in naïve and primed PSCs. Naïve PSCs (WA09) in t2i/L+PKCi and primed PSCs (WA09) in KSR-containing medium. (E) Expression levels of FBRS and AUTS2 in single cells; <2% cells have RPKM > 1 for both genes. Naïve PSCs, blue; primed PSCs, orange. Data from (66). (F) Expression of AUTS2 and FBRS in primed PSCs, nascent naïve PSCs (day 10, reprog.), refractory cells (day 10, refract.), and established naïve PSC lines. Individual replicates shown (n = 3). Data from (32). (G) Expression of FBRS and AUTS2 over a naïve (day 0) to primed (day 10) capacitation time course. Naïve marker KLF17 and primed marker DUSP6 are also shown. Data from (67). (H) FBRS and AUTS2 expression in pluripotent cells over three stages of human development. Circle color, median expression level of single cells; circle size, percentage of single cells with an RPKM value > 0. ICM, inner cell mass (n = 10); preimpl. epiblast, preimplantation epiblast cells (n = 61); postimpl. epiblast, postimplantation epiblast cells (n = 64). Data and cell annotations from (68). (I) Expression of FBRS and AUTS2 in single cells during human development; <8% cells have RPKM > 1 for both genes. Data and cell annotations from (68).

Fig. 6.. PRC1.3 and PRDM14 associate on chromatin to repress target genes and ensure naïve PSC reprogramming.

(A) qPLEX-RIME data show FC in abundance of PCGF3-interacting proteins. All proteins in dataset, gray dots; factors essential for naïve reprogramming, blue circles. (B) Motif enrichment for PRC1.3 target sites in naïve PSCs. Top three motifs shown, ranked by adjusted P value (Fisher’s exact test with Bonferroni correction). (C) Normalized ChIP-seq read counts within a 20-kb peak-centered window in naïve PSCs. Peaks correspond to PRC1.3-bound sites in naïve PSCs (n = 378). PRDM14 heatmaps show ChIP-seq signal under normal [no indole-3-acetic acid (IAA)] and PRDM14-depleted (+IAA) conditions. (D) ChIP-seq (top 6 tracks) and RNA-seq (lower 10 tracks) at an example PRC1.3 and PRDM14 target gene. SCD5 expression is low in WT and PCGF3 KO primed PSCs (day 0) and remains low in WT cells after 10 days of 5i/L/A reprogramming. In contrast, SCD5 is derepressed in PCGF3 KO PSCs at day 10 of reprogramming, as well as in 4i-cultured PSCs following IAA-induced degradation of PRDM14-AID. Treating parental (Par.) cells that lack the AID tag with IAA has no effect. PRDM14 ChIP-seq and RNA-seq data from (47). (E) ChIP-seq normalized read counts without IAA and following IAA-induced PRDM14 degradation. Blue circles, PRC1.3 target sites (n = 378); green circles, RING1B-only sites with matched distribution of RING1B ChIP-seq levels (n = 397). Data from (47). (F) Gene set enrichment analysis (GSEA) of naïve PRC1.3 target genes ranked by FC in transcription following IAA-induced degradation of PRDM14 in naïve PSCs. Genes derepressed in the absence of PRDM14 are enriched in PRC1.3 targets (FDR < 0.001; Kolmogorov-Smirnov test). (G) Flow cytometry of PRDM14-AID-VENUS PSCs at day 10 of naïve PSC reprogramming (CR conditions) in the absence (top) or presence (bottom) of IAA-induced PRDM14 degradation. Successfully reprogrammed naïve cells in the bottom right quadrant. Data are representative of three independent experiments.

Fig. 7.. Inhibiting HDAC2 improves naïve PSC reprogramming efficiency using multiple methods.

(A) Ranked DE plot by comparing nascent naïve with refractory populations. Impediment genes are highlighted. The top 15 ranked impediment genes are shown. (B) Chart shows the FC in gRNA counts and associated P values for all HDAC genes. (C) Chart shows the expression levels of each HDAC gene in naïve PSCs. Mean with SD, n = 3 biological replicates. Data are from (32). TPM, Transcripts Per Million. (D) Chart shows the P values of HDAC2-containing complex members as a measure of their depletion in the refractory cell population following the CRISPR-Cas9 screen. The red line indicates P = 0.02 as a significance cutoff. (E) Flow cytometry contour plots of WT PSCs at day 10 of naïve PSC reprogramming (CR conditions). HDAC2 inhibitors BRD4884 and BRD6688 and pan-HDAC inhibitor VPA were added at 25 μM. VPA added at 1 mM is also shown, which is the concentration used in the CR protocol and serves as a positive control. Successfully reprogrammed naïve cells appear in the bottom right quadrant. Data are representative of three independent experiments. (F) Chart shows the summary results from (E). Data show mean with SD for three independent experiments. Note that there was no significant difference in the reprogramming efficiency between the HDAC2 inhibitor–treated cells and the cells treated with VPA at 1 mM (two-tailed t test). (G) Chart shows the percentage of nascent naïve cells as measured by flow cytometry after 10 days of reprogramming under 5i/L/A conditions. HDAC2 inhibitors BRD4884 and BRD6688 and pan-HDAC inhibitor VPA were supplemented to the media at 5 μM for the first 3 days. Data show mean with SD for two independent experiments. DMSO, dimethyl sulfoxide. (H and I) Phase-contrast images of naïve PSCs following reprogramming in (H) CR medium or (I) 5i/L/A medium, supplemented with HDAC2 inhibitors BRD4884 or BRD6688. Scale bars, 100 μm.