Abstract
Mouse embryonic stem cells (mESCs) cycle in and out of a transient 2-cell (2C)-like totipotent state, driven by a complex genetic circuit involves both the coding and repetitive sections of the genome. While a vast array of regulators, including the multi-functional protein Rif1, has been reported to influence the switch of fate potential, how they act in concert to achieve this cellular plasticity remains elusive. Here, by modularizing the known totipotency regulatory factors, we identify an unprecedented functional connection between Rif1 and the non-canonical polycomb repressive complex PRC1.6. Downregulation of the expression of either Rif1 or PRC1.6 subunits imposes similar impacts on the transcriptome of mESCs. The LacO-LacI induced ectopic colocalization assay detects a specific interaction between Rif1 and Pcgf6, bolstering the intactness of the PRC1.6 complex. Chromatin immunoprecipitation followed by sequencing (ChIP-seq) analysis further reveals that Rif1 is required for the accurate targeting of Pcgf6 to a group of genomic loci encompassing many genes involved in the regulation of the 2C-like state. Depletion of Rif1 or Pcgf6 not only activates 2C genes such as Zscan4 and Zfp352, but also derepresses a group of the endogenous retroviral element MERVL, a key marker for totipotency. Collectively, our findings discover that Rif1 can serve as a novel auxiliary component in the PRC1.6 complex to restrain the genetic circuit underlying totipotent fate potential, shedding new mechanistic insights into its function in regulating the cellular plasticity of embryonic stem cells.
Background
As metaphorized in Waddington’s epigenetic landscape, the fate potential of mammalian early embryonic cells is increasingly straitened as development proceeds, from totipotency to pluripotency, all the way to terminal differentiation (Eckersley-Maslin et al. 2018; Xu et al. 2021). The molecular networks underlying this narrowing of differentiation potency have been under intensive investigations, leading to the ground-breaking rejuvenation of differentiated cells back to the pluripotent state (Takahashi and Yamanaka 2006). Sitting atop the Waddington’s landscape, cells of totipotency have received increasing attention in recent years because of their ability to produce a complete organism from a single cell (Ishiuchi and Torres-Padilla 2013; Le et al. 2020; Zhou and Dean 2015). One milestone discovery in this field is that the mouse embryonic stem cells (mESCs) in culture are not uniform, with a small population shifting in and out of a 2-cell (2C)-like totipotent state, as reflected by the transient expression of the 2C stage-specific endogenous retroviral element MERVL (Macfarlan et al. 2011; Macfarlan et al. 2012). Subsequent studies have further uncovered a set of characteristic features related to this totipotent state. At the epigenetic level, the 2C-like cells display attenuated higher-order chromatin organization, reduced global DNA methylation, and increased chromatin accessibility. Their histones are more mobile and harbor higher levels of active histone modifications, and their chromocenters are less organized relative to the pluripotent mESCs (Genet and Torres-Padilla 2020). These distinct epigenetic features coincide with a unique transcriptional program reminiscent of that of the 2C stage embryo, with downregulated pluripotent factors such as OCT4, SOX2, and NANOG, and upregulated 2C stage-specific genes (2C genes) including Zfp352, Eif1a, and the Eif1a-like cluster, the Zscan4 cluster, as well as the repetitive elements major satellites and the aforementioned endogenous retrovirus MERVL (Eckersley-Maslin et al. 2018; Genet and Torres-Padilla 2020; Lu and Zhang 2015; Xu et al. 2021; Zhou and Dean 2015).
To date, many regulators have been identified to promote or repress the totipotent state. The transcription factor DUX, which can directly activate MERVL and Zscan4, is both necessary and sufficient to induce the 2C-like totipotency (De Iaco et al. 2017; Eckersley-Maslin et al. 2016; Hendrickson et al. 2017; Whiddon et al. 2017; Yang et al. 2020). Other factors upstream of DUX, such as Dppa2, Dppa4, and NELFA, can also activate the 2C program (De Iaco et al. 2019; Eckersley-Maslin et al. 2019; Hu et al. 2020; Yan et al. 2019). On the other hand, genome-wide knockdown screens have identified a vast array of repressors whose downregulation promotes the emergence of the 2C-like cells. These repressors include histone chaperone CAF1 complex, acetyltransferase Tip60/Ep400 complex, H3K9 methyltransferase Setdb1 and its binding proteins Trim28 and Atf7ip, DNA methyltransferase Dnmt1, chromosomal protein Smchd1, transcription factor Myc, RNA N(6)-methyladenosine (m6A) modification reader Ythdc1, RNA binding protein Lin28, and components in the post-translational modification SUMOylation pathway (Cossec et al. 2018; Fu et al. 2019b; Huang et al. 2021; Ishiuchi et al. 2015; Liu et al. 2021; Rodriguez-Terrones et al. 2018; Sun et al. 2022; Theurillat et al. 2020; Wu et al. 2020; Yan et al. 2019; Yang et al. 2015).
Several members of the polycomb-group proteins have been linked to the regulation of cell fate potential. The polycomb repressive system is conserved in the five major animal lineages, and in mammals, it has diversified into canonical (cPRC1) and non-canonical complexes (ncPRC1) to regulate a plethora of cellular processes (Schuettengruber et al. 2017). The ncPRC1 can be further divided into six subcomplexes PRC1.1–1.6. Each contains a distinct Pcgf subunit (Gao et al. 2012). The PRC1.6 complex, consisting of multiple subunits including Pcgf6, RNF2, RYBP, L3mbtl2, Mga, Max, and E2F6, is known to inhibit meiotic entry of embryonic cells by targeting meiosis and germline genes (Dahlet et al. 2021; Endoh et al. 2017; Liu et al. 2020; Maeda et al. 2013; Mochizuki et al. 2021; Suzuki et al. 2016; Uranishi et al. 2021). Depletion of PRC1.6 subunits such as Pcgf6 or L3mbtl2 results in multiple defects in embryonic development, including failure of gastrulation, abnormal axis development, and embryonic lethality (Endoh et al. 2017; Liu et al. 2020; Qin et al. 2012). Besides, Pcgf6 is reported to exert essential functions in maintaining pluripotency of embryonic stem cells (Zhao et al. 2017). Of note, knockdown many subunits in the PRC1.6 complex, such as Pcgf6, RNF2, RYBP, Mga, Max, and L3mbtl2, can significantly increase the proportion of 2C-like cells in mESCs, indicating that the PRC1.6 complex is also indispensable for the control of totipotent state (Cossec et al. 2018; Li et al. 2017; Rodriguez-Terrones et al. 2018).
First characterized as a Rap1-interacting factor participating in transcriptional silencing and regulation of telomeres (Hardy et al. 1992), Rif1 has multiple functions ranging from control of replication timing (Cornacchia et al. 2012; Foti et al. 2016; Gnan et al. 2021; Klein et al. 2021; Yamazaki et al. 2013; Yamazaki et al. 2012), promotion of non-homologous end joining (NHEJ) during the repair of DNA double-strand breaks (DSBs) (Chapman et al. 2013; Gupta et al. 2018; Mirman et al. 2018; Noordermeer et al. 2018), to decatenation of DNA bridges in mitosis (Bhowmick et al. 2019; Hengeveld et al. 2015; Zaaijer et al. 2016). We have identified Rif1 as a repressor of the expression of MERVL in a previous shRNA candidate screen (Li et al. 2017). Rif1 can bind many endogenous retroviruses and silence their transcription by recruiting a panel of epigenetic regulators. Depletion of Rif1 activates MERVL and many genes specifically expressed at the 2C embryo stage (Li et al. 2017), suggesting that it acts as a barrier during the transition from pluripotency to totipotency. However, it is still not clear how Rif1 functions in concert with other totipotency factors to regulate the cellular plasticity of embryonic stem cells.
In this report, by comparing the transcriptomic dynamics after downregulation of the known totipotency regulators, we reveal a novel link between Rif1 and the PRC1.6 complex. Rif1 interacts with Pcgf6, stabilizing the PRC1.6 complex and targeting it to a group of genomic loci involved in the regulation of the 2C-like state. Our finding sheds new light on the complex genetic circuit underlying the potency switch that may translate into improved reprogramming of somatic cells for better therapeutic uses.
Results
Transcriptomic correlation analysis reveals functional modules regulating mESCs fate potential
Many proteins have been identified to regulate the transitions between totipotency and pluripotency (Fig. 1A). To situate Rif1 in the regulatory circuit governing the metastable fate potential of mESCs, we first assembled the 21 repressors of the 2C-like totipotent state into a functional protein-protein interaction (PPI) network using STRING. Proteins belonging to the same complex (PRC1.6 complex) or involved in similar biological pathways such as SUMOylation and methylation were clustered, forming distinct functional modules. However, Rif1 was rather peripheral in the PPI network, with only a potential connection to the histone chaperone Chaf1a (Fig. (Fig.1B).1B). We then collected and compared the transcriptomic data of mESCs after depletion of each of the known repressors. Although downregulating the expression of these repressors could all facilitate the pluripotency-to-totipotency transition, principal component analyses based on the expression of either genes or repetitive elements both revealed significant heterogeneity among these totipotency-like cells (Fig. (Fig.1C-D,1C-D, Supplementary Table 1). Of note, the global transcriptomic changes brought about by the depletion of Rif1 showed a certain degree of similarity to that caused by the knockdown of RNF2, the catalytic subunit of the PRC1 complex harboring the monoubiquitination activity toward lysine 119 of histone H2A (H2AK119ub). We further performed correlation analyses on the transcriptomic data, and we observed strong correlations among the components of the SUMOylation pathway and detected two additional smaller clusters. One contained Chaf1a, Chaf1b, and Senp6, and the other encompassed Rif1, Pcgf6, and RNF2 (Fig. (Fig.1E).1E). Since the repetitive portion of the genome is contributing significantly to the genetic wiring of stem cell potency (Fu et al. 2019a; Schlesinger and Goff 2015), we also performed correlation analysis using the expression data of repetitive elements (Fig. (Fig.1F).1F). The results were more or less consistent with that generated with coding genes. Proteins involved in the SUMOylation pathway again showed the strongest correlations, and the histone chaperone subunits Chaf1a and Chaf1b were clustered to each other. Although Rif1 in this analysis was not included in any of the clusters, it still manifested a marked correlation with Pcgf6, the characteristic subunit of the PRC1.6 complex.
Transcriptomic correlation analysis reveals a functional link of Rif1 to RNF2 and Pcgf6. A Known regulators controlling the pluripotency-to-totipotency transition of mESCs. B STRING visualization of the protein-protein interaction network of the previously reported 21 repressors of totipotency. C Principal component analysis (PCA) of gene expressions of the mESCs with depletion of the indicated genes. D PCA based on the expressions of repetitive elements. E Correlation matrix showing the unbiased and pairwise comparisons of the global changes in gene expression upon depletion of the indicated genes in mESCs. F Correlation matrix generated with the transcriptional changes of repetitive elements upon depletion of the indicated genes in mESCs. Color bars represent Pearson correlation coefficient. G Venn diagram illustrating the differentially expressed genes (adjusted p-value < 0.05 and |Log2 Fold Changes (FC)| > 1) in the mESCs depleted of Rif1, Pcgf6, or RNF2. H Venn diagram of the differentially expressed repetitive elements (adjusted p-value < 0.05 and |Log2 FC| > 1)
To directly evaluate the relationship between Rif1 and the PRC1.6 complex, we performed pairwise correlation analyses using the transcriptomic data from mESCs downregulated of Rif1 or core components of the PRC1.6 complex. Knockdown of either RNF2 or Pcgf6 in mESCs resulted in transcriptional changes of both coding genes and repetitive elements highly correlative to that caused by depletion of Rif1 (Fig. S1A-D). We further compared the differentially expressed genes and repetitive elements among these groups (Fig. (Fig.1G-H).1G-H). Of 1138 differentially expressed genes in the mESCs knocked down of Pcgf6, 753 showed consistent changes upon Rif1 depletion. Similarly, 719 out of 1148 differentially expressed genes caused by RNF2 knockdown displayed concomitant changes in the Rif1 depleted mESCs (Fig. (Fig.1G).1G). The overlaps of the differentially expressed repetitive elements among these groups were even more dramatic. More than 90% of the Pcgf6- or RNF2-regulated repetitive elements, 143/150 or 133/143, respectively, were also responsive to the depletion of Rif1 (Fig. (Fig.11H).
Rif1 interacts specifically with Pcgf6
The analyses of transcriptomic dynamics pointed to a functional link between Rif1 and PRC1.6. The PRC1 complex is heterogeneous, containing several subtypes according to the distinct molecular compositions (Gao et al. 2012; Schuettengruber et al. 2017)(Fig. 2A). To interrogate whether the connection of Rif1 to the PRC1 is specific to the PRC1.6 subcomplex, we collected and analyzed the transcriptomic data from the mESCs with single knockout (KO) of Rif1, Pcgf1, or Pcgf6, double KO of Pcgf2/4 or Pcgf3/5, and triple KO of Pcgf1/2/4 (Scelfo et al. 2019). Both the transcriptional changes of genes and repetitive elements demonstrated a specific correlation between Rif1 and Pcgf6 (Fig. (Fig.2B-C).2B-C). We further performed gene ontology (GO) enrichment analysis on these differentially expressed genes induced by the depletion of Rif1 or the different Pcgf proteins. The results revealed that the depletion of Rif1 or Pcgf6 influenced a similar set of genes enriched in cell fate commitment, cell junction assembly, development of nervous system, as well as multiple meiotic processes, which were markedly different from that influenced by the other Pcgf proteins (Fig. (Fig.2D,2D, Supplementary Table 2).
Similar transcriptomic changes caused by depletion of Rif1 or Pcgf6. A Schematic illustration of the compositions of the canonical PRC1 (cPRC1) and non-canonical PRC1 (ncPRC1) complexes. B-C Correlation analysis of the differentially expressed genes (B) or repetitive elements (C) in the mESCs depleted of the indicated Pcgf proteins or Rif1. Color bar represents Pearson correlation coefficient. D Gene Ontology (GO) analysis of differentially expressed genes (|FC| > 1.5, p < 0.05) among the mESCs depleted of Rif1 or the indicated Pcgf proteins. Bubble color indicates the adjusted p value, and bubble size represents the number of genes in each category
To explore if there is a physical interaction between Rif1 and Pcgf6, we utilized the LacO-LacI induced ectopic colocalization system, in which the bait protein is fused to DsRed-tagged LacI and the prey protein tagged with GFP (Gui et al. 2020). The binding of LacI to the LacO array integrated into the genome of U2OS cells concentrates the bait protein, which further recruits the prey protein to manifest an ectopic colocalization of red and green fluorescent signals (Fig. 3A). We used the Pcgf proteins as the baits, and Rif1 tagged with GFP as the prey. When co-transfected into the U2OS cells harboring the LacO array, all the bait proteins formed bright nuclear puncta. The puncta of Pcgf6 effectively recruited the prey GFP-Rif1. On the contrary, the puncta formed by other Pcgf proteins failed to do so (Fig. (Fig.3B).3B). We quantified the colocalization by calculating the relative enrichment of GFP-Rif1 in the region where the bait protein formed bright puncta (Fig. (Fig.3C).3C). The results showed that only Pcgf6 could effectively enrich GFP-Rif1 (Fig. (Fig.3D).3D). We further validated this specific interaction by co-immunoprecipitation in HEK293T cells ectopically expressing HA-tagged Rif1 and Flag-tagged Pcgf proteins (Fig. (Fig.3E).3E). HA-Rif1 was detected only in the Flag-Pcgf6 immunoprecipitants, indicating a specific interaction between Rif1 and Pcgf6.
Specific interaction between Rif1 and Pcgf6. A Schematic illustrating the LacO-LacI system used for detection of protein-protein interactions. B The LacO-LacI induced ectopic colocalization experiments reveal a specific interaction between Rif1 and Pcgf6. GFP-Rif1 is shown in green, LacI-DsRed fused with the indicated bait proteins are shown in red. DNA is stained with DAPI (blue). Bar: 10 μm. C The method to calculate the relative enrichment of GFP-Rif1. D The relative enrichment of GFP-Rif1 in the indicated experimental groups. ns: not significant, **** p < 0.0001 by t-test, Error bars represent SD, n = 12–45 cells per group. E Coimmunoprecipitation of HA-Rif1 with different Flag-tagged Pcgf proteins ectopically expressed in HEK293T cells
To further map the regions on Rif1 that mediate the interaction with Pcgf6, we fused the full-length and different truncated forms of Rif1 to GFP and performed the LacO-LacI induced colocalization assay with DsRed-LacI-Pcgf6 (Fig. 4A). While the middle region of Rif1 containing the intrinsically disordered polypeptide (IDP) domain could not be recruited to the Pcgf6 puncta, both the N-terminal and the C-terminal regions of Rif1 were contributing to the interaction with Pcgf6 (Fig. (Fig.44B-C).
Interactions between Rif1 and other components in the PRC1.6 complex. A Schematic of the full-length and truncations of Rif1 proteins. B The LacO-LacI induced colocalization experiments reveal that both the N- and C-terminal regions of Rif1 can interact with Pcgf6. The full-length and truncations of Rif1 are shown in green, LacI-DsRed fused with Pcgf6 is shown in red, and DNA is stained with DAPI (blue). Bar: 10 μm. C The relative enrichment of GFP signals in the indicated experimental groups. **** p < 0.0001 by t-test, Error bars represent SD, n = 10–45 cells per group. D The LacO-LacI induced ectopic colocalization experiments reveal that Rif1 can interact with other components in the PRC1.6 complex. GFP-Rif1 is shown in green, LacI-DsRed fused with the indicated bait proteins are shown in red. DNA is stained with DAPI (blue). Bar: 10 μm. E The relative enrichment of GFP-Rif1 in the indicated experimental groups. **** p < 0.0001 by t-test, Error bars represent SD, n = 13–45 cells per group. F Coimmunoprecipitation of Pcgf6 and RNF2 with endogenous HA-Rif1. G-H Coimmunoprecipitations of Rif1 with endogenously HA-tagged Pcgf6 or Mga in mESCs
Pcgf6 mediates the physical interaction between Rif1 and the PRC1.6 complex
The PRC1.6 complex comprises many subunits, including Pcgf6, RNF2, Max, E2F6, RYBP, and L3mbtl2. To investigate whether Rif1 also interacts with other components of the PRC1.6 complex, we used different subunits of the PRC1.6 complex as the baits and GFP-tagged Rif1 as the prey in the LacO-LacI induced colocalization assay (Fig.(Fig.4D).4D). While the Pcgf6 puncta recruited the most significant amount of GFP-Rif1, all the other PRC1.6 components were able to enrich GFP-Rif1 to varying degrees compared to the control (Fig. (Fig.4E),4E), suggesting that Rif1 can interact with the whole PRC1.6 complex. We next examined the interactions between endogenous Rif1 and the components of the PRC1.6 using different knock-in (KI) mESCs. The endogenously HA-tagged Rif1 successfully immunoprecipitated Pcgf6 as well as RNF2 (Fig. (Fig.4F).4F). Reciprocally, the endogenously HA-tagged Pcgf6 co-precipitated Rif1 (Fig. (Fig.4G).4G). Another component of the PRC1.6 complex Mga also showed a detectable interaction with endogenous Rif1 (Fig. (Fig.44H).
Given that Pcgf6 manifested the strongest interaction with Rif1, we speculated that the interaction between Rif1 and the PRC1.6 complex was mainly mediated by Pcgf6. To test this, we first used different shRNAs to knockdown the expression of components in the PRC1.6 complex, and evaluated the recruitment of GFP-Rif1 by the Pcgf6 puncta in the LacO-LacI induced colocalization assay (Fig. 5A-B). The quantification results of GFP-Rif1 enrichment showed that knockdown of other subunits in the PRC1.6 complex did not attenuate the interaction between Pcgf6 and Rif1 (Fig. (Fig.5C).5C). On the contrary, knockdown of Pcgf6 significantly reduced the GFP-Rif1 recruitment by the RNF2 puncta (Fig. (Fig.5D-F).5D-F). Taken together, these results suggested that the interaction between PRC1.6 complex and Rif1 was mediated by Pcgf6.
The interaction between Rif1 and PRC1.6 is mediated by Pcgf6. A The LacO-LacI induced colocalization experiments reveal that knockdown of different components of the PRC1.6 complex has no effect on the interaction between Rif1 and Pcgf6. GFP-Rif1 is shown in green, LacI-DsRed fused with Pcgf6 is shown in red, and DNA is stained with DAPI (blue). Bar: 10 μm. B RT-qPCR analysis of the indicated transcripts after the transfection of the corresponding shRNAs in the U2OS-LacO cell line. **** p < 0.0001 by t-test, Error bars represent SD, n = 3. C The relative enrichment of GFP-Rif1 in the indicated experimental groups. ns: not significant by t-test, Error bars represent SD, n = 10–23 cells per group. D The LacO-LacI induced colocalization experiments reveal that the interaction between Rif1 and RNF2 is weakened after the knockdown of Pcgf6. GFP-Rif1 is shown in green, LacI-DsRed fused with RNF2 is shown in red, and DNA is stained with DAPI (blue). Bar: 10 μm. E RT-qPCR analysis of Pcgf6 after the transfection of shRNAs targeting NT or Pcgf6 in the U2OS-LacO cell line. **** p < 0.0001 by t-test, Error bars represent SD, n = 3. F The relative enrichment of GFP-Rif1 in the indicated experimental groups. **** p < 0.0001 by t-test, Error bars represent SD, n = 29 in shNT group, n = 14 in shPcgf6 group. G Western blot showing the protein level of the indicated members of the PRC1.6 complex in the presence or absence of Rif1. H Coimmunoprecipitation of RNF2 and RYBP with endogenously tagged HA-Pcgf6 in the mESCs treated with shRNA targeting NT or Rif1. I Gel filtration of nuclear extracts from the Rif1 WT or Rif1 CKO mESCs. The corresponding elution volumes for each analyzed fraction are labeled. The arrow indicates the approximate elution volume of a 440 kDa protein complex
Depletion of Rif1 destabilizes the PRC1.6 complex
We next investigated if Rif1 was an integral part of the PRC1.6 complex. We induced conditional knockout (CKO) of endogenous Rif1 by treatment of the mESCs with 4-hydroxytamoxifen (4-OHT) as previously described (Li et al. 2017), and examined the protein levels of several components of the PRC1.6 complex. Rif1 protein became undetectable after 4-OHT treatment, whereas the expression of Pcgf6, RNF2, or BYBP was largely unaltered (Fig. (Fig.5G).5G). We next performed immunoprecipitation to investigate the interactions between Pcgf6 and other components in the PRC1.6 complex upon knockdown of Rif1. The endogenously HA-tagged Pcgf6 could co-immunoprecipitate RNF2 and RYBP in the control mESCs. In the Rif1 knockdown mESCs, the interaction between Pcgf6 and RNF2 was compromised (Fig. (Fig.5H).5H). We further performed gel filtration experiments in the presence or absence of Rif1 to probe the intactness of the PRC1.6 complex. In the control nuclear extracts, Rif1 was co-eluted with RNF2 and Pcgf6, whereas in the Rif1-depleted nuclear extracts, Rif1 became undetectable and concomitantly, the elution volumes of RNF2 and Pcgf6 were slightly increased, suggesting a decrease in the size of the PRC1.6 complex (Fig. (Fig.5I).5I). Taken together, these results suggested that Rif1 was a novel auxiliary component of the PRC1.6 complex.
Rif1 modulates the genomic distribution of the PRC1.6 complex
Rif1 is a multi-functional protein that harbors DNA binding activity. To explore its potential function in the PRC1.6 complex, we first analyzed the distribution of Rif1 across the genome using the ChIP-seq data generated previously (Li et al. 2017). Motif analysis revealed that the Rif1-bound peaks also enriched for the consensus binding sequences of the PRC1.6 components, Max and E2F6 (Fig. 6A), suggesting the co-occupancy of Rif1 and PRC1.6 complex. We further performed ChIP-seq experiments for Pcgf6, RNF2, and H2AK119ub, and examined their genomic distributions. The called peaks of Rif1, Pcgf6, RNF2, and H2AK119ub manifested modest correlations (Fig. (Fig.6B,6B, Supplementary Table 3). Approximately 1/3 of the Rif1 peaks (n = 12,729) were also occupied by Pcgf6 (n = 4357), and the overlaps between Rif1 and RNF2 or H2AK119ub were slightly less than that of Pcgf6 (Fig. (Fig.6C).6C). To investigate if Rif1 depletion compromises the genomic targeting of the PRC1.6 complex, we analyzed the distributions of Pcgf6, RNF2, and H2AK119ub on the Rif1-bound regions in control and Rif1-downregulated mESCs by ChIP-seq. We observed that the binding of Pcgf6 as well as RNF2 to the Rif1-bound regions were markedly reduced after the downregulation of Rif1 (Fig. (Fig.6D,6D, Supplementary Table 3), suggesting that a subset of the genomic distribution of the PRC1.6 complex was dependent on Rif1.
Rif1 modulates the genomic distribution of the PRC1.6 complex and regulates a group of 2C genes. A Motif analysis of Rif1-binding sites in mESCs reveals significant enrichment in C-Myc, Max, E2F4, and E2F6 binding motifs. B Correlation matrix showing the unbiased and pairwise comparisons of genomic distributions of the indicated proteins or histone modifications in mESCs. Color bar represents Pearson correlation coefficient. C Venn diagram comparing the Rif1 binding sites (12,729 peaks) with that of Pcgf6 (22,911 peaks), RNF2 (19,408 peaks), and H2AK119ub (8928 peaks). D Top: heatmaps showing the densities of Rif1, Pcgf6, RNF2, and H2AK119ub ChIP-seq reads on the Rif1-bound regions (±1 kb) in control or Rif1 downregulated mESCs. The genomic regions are centered according to the Rif1 ChIP-seq signals in the WT mESCs. Bottom: the averaged ChIP-seq signals of Pcgf6, RNF2, or H2AK119ub on Rif1-bound regions in the presence (blue lines) or absence (green lines) of Rif1. E Selected Gene Ontology (GO) analysis results of the genes in the Rif1 and Pcgf6 co-occupied genomic regions which showed decreased Pcgf6 binding upon Rif1 knockdown. F Venn diagrams comparing the Rif1-bound genes that showed decreased Pcgf6 binding upon Rif1 knockdown with the 2C genes. Hypergeometric p value < 0.01. G Gene set enrichment analysis (GSEA) illustrating the upregulation of 2C genes in the mESCs depleted of Rif1 or Pcgf6. The normalized enrichment scores (NES) > 1 and p < 0.05 is considered significantly enriched. H Heatmap of RNA-seq data showing the expression of the 2C genes in Rif1 CKO and Pcgf6 KD mESCs. The number of 2C genes in each cluster is shown. I Averaged Rif1 and Pcgf6 ChIP-seq profiles in the corresponding clusters. J Integrative Genomics Viewer (IGV) showing ChIP-seq intensity of indicated proteins and the RNA-seq signals on Zfp352
Since Pcgf6 is the characteristic subunit of PRC1.6 and the interaction of the PRC1.6 complex with Rif1 is mainly mediated by Pcgf6, we focused our subsequent analyses on Rif1 and Pcgf6 coregulated genomic regions. We annotated the Rif1 and Pcgf6 co-occupied genomic regions that showed decreased Pcgf6 binding upon Rif1 knockdown and identified 3633 corresponding genes. GO analysis showed that these genes were enriched in biological processes including synapse organization, cell junction assembly, and cell fate commitment (Fig. (Fig.6E,6E, Supplementary Table 2). A 2C gene list consisting of 869 genes was generated previously according to their upregulation at the 2C and zygotic genome activation (ZGA) stage in mouse embryonic development (Li et al. 2017). We found that the 2C genes were significantly enriched in the Rif1-bound genes that showed decreased Pcgf6 binding upon Rif1 knockdown (Fig. (Fig.6F),6F), suggesting that Rif1 and Pcgf6 functioned in concert to regulate the genetic circuit of totipotency.
Rif1 and Pcgf6 co-regulate the expression of many 2C genes and MERVL elements
To investigate the cooperation of Rif1 and Pcgf6 in regulating the mESCs fate potential, we performed RNA-seq analysis with mESCs downregulated of Rif1 or Pcgf6. The gene set enrichment analysis (GSEA) revealed that the 2C genes were significantly upregulated in mESCs with reduced expression of either Rif1 or Pcgf6 (Fig. (Fig.6G,6G, Supplementary Table 4). We further examined the expression of all the 2C genes using the RNA-seq data, and identified four different clusters (Fig. (Fig.6H).6H). Genes in cluster1 showed upregulated expression only in mESCs depleted of Rif1, genes in cluster3 were only activated by the downregulation of Pcgf6, and genes in cluster4 were not upregulated in either condition. We uncovered 195 genes in cluster2 whose expression was consistently upregulated in mESCs downregulated of Rif1 or Pcgf6. Interestingly, genes in cluster2 displayed higher Rif1 and Pcgf6 binding, and the amount of Pcgf6 on these genes was markedly decreased in mESCs depleted of Rif1 (Fig. (Fig.6I).6I). We noticed that several 2C stage marker genes, such as Zfp352 and Zscan4 were in cluster2. We visualized the ChIP-seq and RNA-seq signals at the Zfp352 locus, and the results showed that depletion of Rif1 almost eliminated the Pcgf6 peaks, which was accompanied by the significant upregulation of transcription (Fig. (Fig.66J).
The transcriptional activation of repetitive element MERVL is a marker for the 2C-like totipotent state, and the 2C::tdTomato reporter can reflect the transcriptional activity of the long terminal repeat (LTR) controlling the MERVL expression. We infected the 2C::tdTomato mESCs with lentiviruses carrying shRNAs targeting Rif1 or different Pcgf members. Consistent with our previous study (Li et al. 2017), the knockdown of Rif1 substantially expanded the number of tdTomato positive cells (Fig. 7A). Knockdown of Pcgf6 but not the other Pcgf members also significantly increased the tdTomato positive cells (Fig. (Fig.7B-C),7B-C), although the increase was less profound when compared with that induced by Rif1 knockdown. To further examine the cooperation between Rif1 and Pcgf6 in regulating MERVL, we simultaneously knocked down Rif1 and Pcgf6 using shRNAs, and compared the effect with that from individual knockdowns (Fig. (Fig.7D).7D). While knockdown of Rif1 induced more tdTomato positive cells than that of Pcgf6, the double knockdown showed no additive effect (Fig. (Fig.7E),7E), suggesting that Pcgf6 was in the same regulatory pathway as Rif1. Additionally, we analyzed the expression of MERVL using our ChIP-seq and RNA-seq data, and unambiguously identified 66 Rif1-bound MERVL loci that became transcriptionally activated after Rif1 depletion. 48 of them could also be activated by Pcgf6 knockdown (Fig. (Fig.7F,7F, Supplementary Table 4), indicating that these MERVL loci were co-regulated by Rif1 and Pcgf6. Taken together, these results suggested that Rif1 and the PRC1.6 complex could form a functional module to control the transition from pluripotency to totipotency by restraining a group of 2C genes and MERVL elements.
Rif1 and Pcgf6 coregulates a group of MERVL. A Top: schematic of the 2C::tdTomato reporter. Bottom: fluorescent images showing the fraction of tdTomato-positive cells in control or mESCs treated with the indicated shRNA. Control, non-transfected; shNT, non-target shRNA transfected. Bar: 50 μm. B RT-qPCR analysis of the indicated transcripts after the transfection of the corresponding shRNAs in the 2C::tdTomato reporter cell line. **** p < 0.0001 by t-test, Error bars represent SD, n = 3. C The percentages of tdTomato-positive cells in the indicated experimental groups. ns: not significant, **** p < 0.0001 by t-test, Error bars represent SD, n = 16–26 fields per group. In each field, the number of total cells is approximately 150–350. D Fluorescent images showing the fraction of tdTomato-positive cells in control or mESCs treated with the indicated shRNA. shNT, non-target shRNA transfected. Bar: 25 μm. E The percentages of tdTomato-positive cells in the indicated experimental groups. ns: not significant, **** p < 0.0001 by t-test, Error bars represent SD, n = 20–26 fields per group. In each field, the number of total cells is approximately 150–350. F Heatmap of RNA-seq data showing the expression levels of the Rif1-bound MERVL loci that are derepressed upon Rif1 depletion (n = 66) and their corresponding expression levels in mESCs after Pcgf6 knockdown. A fraction of these MERVL loci (n = 48) becomes activated in Pcgf6 KD mESCs. The color bar denotes the Log2(CPM + 1). G Schematic model of Rif1 as a new auxiliary component of the PRC1.6 complex