Dual repression of endocytic players by ESCC microRNAs and the Polycomb complex regulates mouse embryonic stem cell pluripotency

Cell fate determination in the early mammalian embryo is regulated by multiple mechanisms. Recently, genes involved in vesicular trafficking have been shown to play an important role in cell fate choice, although the regulation of their expression remains poorly understood. Here we demonstrate for the first time that multiple endocytosis associated genes (EAGs) are repressed through a novel, dual mechanism in mouse embryonic stem cells (mESCs). This involves the action of the Polycomb Repressive Complex, PRC2, as well as post-transcriptional regulation by the ESC-specific cell cycle-regulating (ESCC) family of microRNAs. This repression is relieved upon differentiation. Forced expression of EAGs in mESCs results in a decrease in pluripotency, highlighting the importance of dual repression in cell fate regulation. We propose that endocytosis is critical for cell fate choice, and dual repression may function to tightly regulate levels of endocytic genes.

Repression of endocytosis-associated genes by the PRC2 complex. SUZ12 is a component of the PRC2 complex, which has four subunits: SUZ12 (zinc finger containing protein), EED, EZH1 or EZH2 (SET domain containing protein with histone methyl transferase activity) and RBAP48 (histone binding domain containing protein). The PRC2 complex has histone methyltransferase activity and trimethylates histone H3 on lysine 27 (i.e. H3K27me3), a mark of transcriptionally silent chromatin 18 . Knockout of Suz12, Ezh2, Eed results in embryonic lethality at E7.5-8.5 with major defects in gastrulation, with mESCs failing to properly differentiate 19 . In order to determine whether the PRC2 complex was indeed involved in regulating expression of EAGs, (shown in Fig. 1b,c), the catalytic component of the PRC2 complex, Ezh2, was knocked down in two mESC lines, V6.5 and R1, using lentiviral shRNA constructs (Supp. Figure 2a,b). The expression of 45 EAGs was determined using RT-qPCR following Ezh2 knockdown in V6.5 and R1 mESCs (Fig. 2a). Expression of 25 out of these 45 genes was significantly and consistently derepressed upon Ezh2 knockdown (Fig. 2a). Similar results were also obtained upon the knockdown of Suz12, a zinc finger-containing DNA binding component of the PRC2 complex (Supp. Figure 2c,d). We further validated that the regulation of a handful of these EAGs was due to a direct association of the PRC2 complex at their promoter and not due to an indirect effect by performing chromatin immunoprecipitation (ChIP) for a few EAGs. SUZ12 showed strong binding to the Cav1, Cdh2, Tgfbr1, Tgfbr2 and Tgfbr3 promoters as determined by ChIP followed by RT-qPCR (Fig. 2b). Together, these results indicate that the expression of a number of EAGs is indeed repressed in mESCs through the direct action of the PRC2 complex.
Dual regulation of EAG expression by the ESCC miRNA family and the PRC2 complex. The ESCC family of microRNAs (miRNAs) are a family of conserved miRNAs that are highly expressed in ESCs, play a role in cell cycle regulation, and enhance the efficiency of somatic cell reprogramming 8,15,20 . miRNAs are small non-coding RNAs that post-transcriptionally regulate gene expression through complementary binding of their seed sequence with a seed match present in the target mRNA 21 . Intriguingly, of the 50 endocytic genes whose promoter regions are bound by Suz12 (Fig. 1b), we found that 21 genes also had seed matches for miR-294, a member of the ESCC family of miRNAs, either in their 3′UTR or ORF (Supp. Table 1). To validate if these seed matches were indeed functional targets of the ESCC miRNA family, the ESCC miRNA, miR-294 was exogenously introduced into MEFs, cells that are naturally devoid of these miRNAs. 16 out of 21 EAGs were found to be functional targets of miR-294 as their expression significantly decreased upon overexpression of miR-294 (Fig. 2c), indicative of a role of these miRNAs in regulating the stability of EAG transcripts. To further verify this regulation, we utilized the previously published Dgcr8 KO ESC line, which lacks all mature miRNAs 22 . Introduction of synthetic miR-294 into Dgcr8 KO mESCs resulted in the decreased expression of 11 out of 21 EAGs (Fig. 2d). Knockdown  Chen et al.), and the set of genes whose expression levels are 2-fold higher in MEFs relative to mESCs. Differentially upregulated genes are significantly over-represented in the gene targets of SUZ12 (indicated by a low p-value and high Z-score). (b) Bar chart showing the 50 genes with known association with endocytosis, whose promoter regions are bound by SUZ12 and are upregulated in MEFs relative to mESCs. The interaction score between SUZ12 and each gene, as well as the log2 fold expression change of each gene, are shown. (c) RT-qPCR analysis of SUZ12 bound endocytic genes in mESCs and MEFs. mRNA expression is normalized to Gapdh, and further normalized to expression in mESCs. Error bars represent mean ± S.D for experiments in triplicates (N = 3). *p < 0.05; **p < 0.01; ***p < 0.001 by Students T-test. of Ezh2 (Supp. Figure 2e) resulted in an increase in expression, while a combination of Ezh2 knockdown along with exogenous supply of miR-294 resulted in intermediate levels (Fig. 2d). This powerfully demonstrates the responsiveness of specific EAGs to dual regulation by the PRC2 complex as well as by miR-294.
To demonstrate binding and regulation of EAGs by the ESCC family of miRNAs, the 3′UTRs of Cav1, Cdh2 and human Tgfbr2 (which shows a high degree of conservation with the mouse Tgfbr2 3′UTR) were amplified and cloned downstream of renilla luciferase in the pSiCHECK2 vector. The Cav1 3′UTR contains one seed match (6-mer), Cdh2 ORF contains one seed match (7-mer) (Supp. Table 1), and the Tgfbr2 3′UTR contains three (7-mer) and two (6-mer) seed matches for the ESCC miRNAs. A significant downregulation of renilla luciferase activity in the miR-294 transfected samples was observed with all three constructs, further validating their regulation by the ESCC miRNAs (Fig. 2e). Introduction of miR-294 into HEK293 cells also resulted in a decrease of CAV1 and CDH2 protein levels (Supp. Figure 3a,b). Together these experiments validate a role for the ESCC microRNAs in regulating the stability of a number of EAGs in mESCs.

Exogenous expression of EAGs results in a decrease in mESC pluripotency. To interrogate
whether the repression of EAGs was essential for the maintenance of pluripotency, we chose two candidates, namely Cav1 and Cdh2 out of the 12 dual regulated genes. CAV1 is an integral member of the caveolae-mediated endocytic pathway, and has been shown to regulate specific signalling pathways that initiate at the membrane, including WNT and TGFβ [23][24][25] . CAV1 works in conjunction with its partner, CAVIN1 for normal caveolar biogenesis 26 . CDH2 is a single-pass transmembrane protein, whose compartmentalization into different endocytic compartments has been shown to play an important role during mouse cerebral cortex development 27 .
Caveolin-mediated endocytosis is an important route for the internalization of proteins and molecules within a cell 23 . Further, in mammary stem cells, the loss of Cav1 has been shown to result in an increase in the stem cell population 25 . The other dual regulated gene, Cdh2, enhances mESC differentiation upon supplementation with FGF2 28 . Moreover, CDH2-based substrates have been used for mESC and iPSC differentiation towards the neural lineage 29,30 . The dual repression of Cav1 and Cdh2 in mESCs suggests that their repression may be essential to the maintenance of the pluripotent state. To test this, we checked the expression of Cav1 and Cdh2, and found their expression to be very low at both the mRNA and protein levels in mESCs, with expression increasing upon differentiation (Fig. 3a,b). Similar to Cav1, Cavin1 that acts along with Cav1 for caveolin-mediated endocytosis, had lower expression in mESCs, with levels increasing upon differentiation (Supp. Figure 3c). Immunostaining of mESCs using antibodies specific to CAV1 and CDH2 showed very low expression, whereas MEFs showed expression of CAV1 and CDH2 (Supp. Figure 3d,e).
In order to demonstrate that the dual repression of Cav1 and Cdh2 is critical for the maintenance of pluripotency, we ectopically expressed Cav1 along with Cavin1 (Supp. Figure 3f,g), or Cdh2 (Supp. Figure 3h), in mESCs. Overexpression of Cav1 along with Cavin1, or Cdh2 caused a significant decrease in the expression of the pluripotency markers (Fig. 3c,e), along with a significant increase in the expression of differentiation markers (Fig. 3d,f). Together, our results demonstrate that the regulation of endocytosis associated genes is important for the maintenance of pluripotency.
Caveolin has been implicated in regulating the meiotic progression during oocyte development in C. elegans 31 . CAV-1 is expressed in the germline and early embryos in C. elegans 31 . We therefore investigated the role of CAV-1 in the development of C. elegans, post fertilization. To this end we carried out live imaging of control and cav-1 RNAi embryos co-expressing GFP fused to PLC1δ-PH and mCherry fused to Histone to mark membrane and DNA respectively. We monitored the first cell division in the embryonic 1-cell stage. While initiation of furrow was similar in control and cav-1 RNAi embryos, ingression of the cytokinetic furrow was markedly delayed in cav-1 RNAi embryos compared to the control (Fig. 3g), indicative of a developmental delay as early as the 1-2 cell stage. Together, these results demonstrate the evolutionary conservation of the importance of EAGs in early embryonic development. The existence of a dual repressive mechanism, where one type of repression may function as a back-up in case the other fails, further strengthens the notion that accurate control of endocytic pathways plays a key role in cell fate maintenance.

Discussion
The involvement of cellular trafficking in the regulation of the pluripotent state is a relatively new concept. While alterations in the expression of endocytic genes during cell fate switching have been observed during the process of reprogramming 8 , the exact mechanism of regulation of these genes remains unknown. Here we show for the first time that specific endocytic genes are repressed in mESCs compared to differentiated cells (Fig. 1b,c). This repression occurs at two levels-at the transcriptional level through the action of the repressive methyltransferase, PRC2, and at the posttranscriptional level by the ESCC family of miRNAs (Fig. 2). Together, these two mechanisms ensure that the expression of these genes remains repressed in the pluripotent state. While there are isolated examples in the literature describing similar mechanisms 32,33 , this is the first time that such a dual regulation has been described in ESCs with respect to cellular trafficking and the regulation of the pluripotent state.
While Cav1 is a major endocytic player that is dually repressed, a number of other molecules that are associated with endocytosis are similarly repressed. These include Cdh2 and members of the TGFβ pathway, including Tgfbr1, Tgfbr2 and Tgfbr3 (Figs 2, 3). All these EAGs have been previously shown to play an integral part during differentiation. The caveolin-mediated endocytic pathway, regulates the activity of signalling pathways such as WNT and TGFβ, both of which are known to play a central role in ESC pluripotency [23][24][25] . Endocytosis of CDH2 in an AP-2 adaptor complex dependent manner has been shown to be essential for neurite outgrowth and circuit formation 34 . Inhibition of RAB5 (early endosomal marker), or RAB11 (recycling endosome marker), showed defects in CDH2 trafficking, which resulted in neuronal migration defects during mouse cerebral cortex development 27 . The TGFβ pathway has been shown to affect the pluripotency of stem cells 8,35,36 . The signalling outcome of the TGFβ pathway is also tightly regulated in an endocytic pathway-dependent manner. Clathrin-mediated endocytosis promotes TGFβ induced SMAD activation, while lipid rafts/caveolae facilitate the degradation of TGFβ receptors and therefore turn off TGFβ signalling 24,37 .
Interestingly, overexpression of two candidate EAGs, Cav1 and Cdh2 result in a decrease in pluripotency and an upregulation of differentiation markers (Fig. 3). Knockdown of cav-1 in C. elegans also results in a delay in cytokinetic furrow progression (Fig. 3). While these genes singly do not cause a complete shift in cell fate, they do cause a decrease in pluripotency upon overexpression, suggestive that perhaps collectively these genes, along with others that are similarly regulated, may be capable of facilitating more drastic transitions in cell fate.
Apart from the dual regulation that we describe in this study, it is possible that additional modes of regulation may operate to control the levels of expression of these genes during development. Indeed previous reports indicate that CAV1 itself can also undergo ubiquitination, followed by lysosomal degradation 38 . It is thus likely that other EAGs may also be subject to additional mechanisms of regulation during development.
In conclusion, our data stresses the importance of regulation of specific endocytic pathways in cell fate maintenance. We propose a model wherein key genes (such as those that are involved in endocytosis) that can drive cell fate changes, are kept under a tight check by both transcriptional and post-transcriptional mechanisms (Fig. 4). We speculate that dual repression functions as a back-up mechanism to prevent leaky expression of genes that may play a role in facilitating a shift between cell fates.

Materials and Methods
Mouse embryonic stem cell culture. V6.5 or R1 mouse embryonic stem cells were cultured on tissue culture grade plastic plates coated with 0.2% gelatin. They were maintained in Knockout DMEM supplemented with 15% heat-inactivated fetal bovine serum, 0.1 mM beta mercaptoethanol, 2mM L-glutamine, 0.1 mM nonessential amino acids, 5000 U/ml penicillin/streptomycin, and 1000 U/ml LIF (ESC medium). Cells were passaged every 3 days using trypsin.
SiRNA preparation. Suz12 was amplified from V6.5 mESCs cDNA using Suz12 specific primers having flanking T7 sequence. T7 primer was then used to amplify the entire sequence. T7 RNA polymerase was used to perform in-vitro transcription, and the double stranded RNA generated was digested using RNAse III 39 . Non targeting siRNA was prepared using GFP as a template.
Transfection. 6   18870) were obtained from Addgene. pEGFPN1-Cav1 was a kind gift from Dr. Nagaraj Balasubramanian. For Cdh2 overexpression, mESCs were grown in E.S media without LIF supplemented with Retinoic Acid (10 −8 M) and 5 μg/ml FGF2. Immunocytochemistry and imaging. 3 × 10 4 mESCs or 2 × 10 4 MEFs were plated on gelatin coated coverslips in a 24 well plate. The following day, cells were washed with PBS, fixed with 4% paraformaldehyde at room temperature for 20 min, and permeabilized with 0.1% Triton X-100 in PBS for 5 min. After blocking in 5% FBS and 5% bovine serum albumin in PBS for 1 h, cells were incubated with primary antibodies appropriately diluted in blocking buffer overnight at 4 °C. The following day, cells were washed with PBS for 30 min at room temperature and incubated with fluorescently conjugated secondary antibodies for 1 h. Post washing, nuclei were stained with DAPI. After 2 washes, coverslips were mounted onto glass slides and analyzed using a Zeiss 510 laser-scanning confocal microscope.
RNA isolation and Real time PCR. Total RNA was isolated from mESCs or MEFs using TRIzol as per manufacturer's instructions. Complementary DNAs (cDNA) were synthesized using Superscript-III first-strand synthesis system for RT-PCR as per manufacturer's instructions. Gene-specific primers for RT-qPCR were designed using ABI Primer Express 3.0 software (sequences provided in Supp. Table 2). The quantitative RT-PCR reactions were done using ABI power SYBR Green PCR master mix and reactions were run on the ABI qPCR system, 7900 HT.
Western blotting. Total proteins were extracted from cells using RIPA buffer containing proteinase inhibitors on ice followed by centrifugation at 12,000 rpm for 20 min at 4 °C. Protein concentration was measured using Bradford's reagent. Equal quantity of total protein were subjected to SDS-PAGE under reducing conditions followed by transfer to PVDF membrane. After transfer, the membrane was blocked using 5% BSA in TBS. Post blocking, the membrane was incubated at 4 °C overnight with the appropriate primary antibody. After 3 × 10 min wash in Tris-buffered saline (1X TBS) containing 0.1% Tween-20 (TBS-T), the membranes were incubated with an HRP-conjugated secondary antibody (1:1000) for 1 hour at room temperature. NovexECL reagent was added to the membranes and images were captured post exposure using a chemi-doc system (GE Healthcare, Catalog no. AI600). Western blot intensity was documented using the ImageJ software. Ratio of Renilla luciferase activity to firefly luciferase was calculated for each experiment.

Luciferase Reporter
Regulatory TF data. Data for genome-wide transcriptional regulation in ES cells was independently obtained from Chen et al. 16 . This regulatory network was assembled by pooling together the genome-wide ChIP-seq profiles (binding sites) for 13 sequence-specific transcription factors (TFs) and 2 transcriptional regulators (P300 and SUZ12). Each pair of gene and TF in the network has also been assigned an association score between 0 and 1, reflecting the relative proximity of the TF binding site to the TSS. TF enrichment analysis. In order to identify putative transcription factors whose differential activity can account for the gene expression differences between ES and MEF cells, we integrated information about the differentially expressed genes (DEGs) with the ES cell-specific transcriptional regulatory network. DEGs were identified using previously published microarray expression profiles for mouse embryonic stem cells (mESCs) and mouse embryonic fibroblasts (MEFs) 17 . The expression profiles along with the associated annotation data, were downloaded from the GEO database (GEO accession number: GSE8024). This dataset consists of three replicates of WT ES cells and two of MEFs, profiled with the Affymetrix Mouse Genome 430 2.0 Array. Only those probes whose expression was detected in at least one of the two conditions (at least 2 out of 3 ES samples/in both the MEF samples) were retained for further analysis. Significance Analysis of Microarrays (SAM) 40 was applied to the quantile-normalized, log2-transformed expression values. A significance threshold of FDR < 0.001 and absolute log2 fold expression change threshold of 1 were used to identify differentially abundant genes between the two cell types. For every TF, the over-representation of its target set (all genes assigned nonzero TF-gene association scores) in the DEG set was estimated by one-sided Fisher's exact test. TFs with a p-value of less than 0.05 (after Bonferroni correction for multiple testing) were identified as significantly enriched, and provide hypotheses to explain the transcriptional remodelling that accompanies differentiation. This analysis was carried out separately on the up-regulated and down-regulated genes. As a 'weighted' alternative to the above approach that additionally makes use of the TF-gene association scores, we also computed enrichment z-scores for every TF, as follows. First, the sum of its association scores with genes in the DEG set was estimated. Next, subsets of equal size were randomly drawn 1000 times from the full target set of that TF (i.e. all the genes having non-zero association score with it), and the aggregated score of every such sample was recorded. This yielded a baseline distribution of aggregate scores, which was used to compute a z-score (number of standard deviations from the mean) for the DEG set. A large positive z-score was taken to indicate a concordance between the binding site information (regulatory influence of the TF) and expression changes, and thus provides an alternative measure to associate TFs with the genome-wide differential expression.