Rbm47 is abundantly expressed in pluripotent stem cells (PSCs) and localizes to both nucleus and cytoplasm
We found that Rbm47 transcript and protein were abundantly expressed in mESCs as compared with differentiated cells such as mouse embryonic fibroblasts (MEF) (Fig. 1B and 1C). To determine the cellular localization of RBM47 protein, we fractionated mouse ESCs into cytoplasmic and nuclear extracts and analyzed them by immunoblotting (Fig. 1D). RBM47 was found to be expressed as both, a nuclear and a cytoplasmic RBP with significant enrichment in the nucleus of mESCs, indicating its discrete roles in nuclear and cytoplasmic RNA metabolism. Similar data were obtained with human iPSC lines and human dermal fibroblasts suggesting closely related functions in human counterparts (Supplementary fig. S1 D, E, and F).
Rbm47 expression profile in specific lineages derived in vitro from mESCs
Early mouse development involves the specification of three lineages at the blastocyst stage. As shown in Fig. 1A, the early blastocyst stage (E3.25) has a fully specified outer layer of cells called trophectoderm (TE), while the inner cell mass (ICM) retains pluripotency. In the late blastocyst stage (E4.5), ICM undergoes lineage specification into the pluripotent epiblast (Epi) and primitive endoderm (PrE). As the blastocyst enters the subsequent stages of development, TE gives rise to the extraembryonic ectoderm (ExE), and the PrE gives rise to visceral endoderm surrounding both the Epi and ExE. Epi gives rise to tri-lineages of the embryo proper. Fortunately, the self-renewing ability of these cells enabled us to derive and maintain stem cell lines from each of these lineages and utilize them to study blastocyst development in vitro . ESCs can be derived from both the ICM and Epi and cultured indefinitely in vitro under defined growth conditions. Currently, protocols are available to convert mESCs into naive endoderm (nEnd) or extraembryonic endoderm (EXEn; similar to PrE) and epiblast-like cells (EpiLCs, similar to Epi) in addition to the traditional germ layer differentiation of embryo proper.
To explore the function of Rbm47 during early embryonic development using mESCs, we subjected mESCs to various differentiation strategies and analyzed Rbm47 expression. First, we differentiated mESCs as embryoid bodies (EBs), which undergo spontaneous differentiation into the three germ layers and mimic post-implantation embryos in vitro. RT-qPCR profiling revealed a 2–3 fold upregulation of Rbm47 expression in differentiating EBs compared with mESCs (Fig. 1E). In contrast, RBM47 protein levels gradually reduced as EB differentiation progressed with only in day-3 EBs protein level correlated with mRNA level. (Fig. 1F). Expression of Nanog, a core pluripotency marker, significantly reduced both at mRNA and protein levels, served as an indicator for EB differentiation (Fig. 1E and 1F).
Next, to test whether Rbm47 expression is subjected to any lineage-specific regulation, we directed mESCs into epiblast-like cells (EpiLCs) and extra-embryonic endoderm (cXEN) cells, that mimic the successive cell fates of inner cell mass in the post-implantation embryos in vitro. Furthermore, mESCs were specified into derivatives of definitive endoderm (DE), mesoderm (MES), and neuroectoderm (NE), which mimic the primitive lineages of the embryo proper. To ensure proper differentiation, all the lineage derivatives were profiled for expression of specific-lineage markers besides monitoring Rbm47 expression (Fig. 1G, 1H and Supplementary fig. S2). We observed a differential expression of Rbm47 in EpiLCs and cXEN cells, with mRNA being significantly downregulated in EpiLCs, but upregulated nearly 1.6-fold in cXEN cells (Fig. 1G). RBM47 protein level correlated substantially with mRNA level in EpiLC but not in cXEN as protein level was reduced compared to mESCs (Fig. 1H). We speculated that Rbm47 expression might undergo post-transcriptional regulation in cXEN cells. We assessed the half-life of Rbm47 mRNA by treating mESCs and cXEN cells with alpha-amanitin, followed by the determination of mRNA abundance at different time points. We observed only a slight increase in the half-life of Rbm47 mRNA (mESC t1/2= 10.01 h and cXEN t1/2= 10.83 h) in cXEN cells (Supplementary fig. S3). To gain insights into Rbm47 expression in the lineages of the embryo proper, we measured its mRNA and protein expression in DE, MES, and NE derived from mESC differentiation (Fig. 1G and 1H). We found that Rbm47 expression is indeed subjected to lineage-specific regulation, with its expression remained unchanged in DE but compromised largely in MES and NE compared to mESCs.
Taken together these results suggest that RBM47 is abundantly expressed in mESCs and its expression undergoes lineage-specific regulation. As described in previous studies, Rbm47 expression was strongly detected in the endoderm (primitive gut, liver, and pancreas) and ExEn appendages (yolk sac) in E8.5 mouse embryos, but poorly detected in developing tissues of other lineages [19, 35], demonstrating that our expression analysis utilizing mESCs differentiation has fairly recapitulated the in vivo expression patterns.
Rbm47 depletion doesn’t affect in vitro mESC maintenance
We used the RNAi approach for loss-of-function studies to investigate the significance of Rbm47 expression in mESCs. We transduced ESCs with lentivirus that expresses shRNAs targeting either Rbm47 mRNA or lacZ mRNA (non-targeting control) and established stable cell lines (Rbm47 depleted mESCs designated as shRbm47#1 and shRbm47#3; control ESCs as shlacZ in the figures). There was an efficient knockdown of Rbm47 at mRNA (~ 80%) as well as protein levels (~ 50%) (Figs. 2A and 2B). Rbm47-depleted mESCs displayed no apparent change in the undifferentiated state as they expressed similar levels of pluripotency markers as control mESCs as demonstrated by RT-qPCR, western blotting, and immunocytochemistry (Figs. 2E, 2F, and 2G). Furthermore, shRbm47 mESCs could be propagated continuously (for at least 15 passages) in the ESC medium supplemented with leukaemia inhibitory factor (LIF) and retained the ability to form colonies. Rbm47 knockdown mESCs had a cell profile that was indistinguishable to control mESCs, as demonstrated by cell cycle analysis (Fig. 2H). In terms of morphology and alkaline phosphatase (ALP) activity, shRbm47#3 mESCs were comparable to control mESCs, whereas shRbm47#1 mESCs showed a slightly diffused morphology with a reduced intensity of ALP staining (Fig. 2C and 2D). However, these cells display a similar pluripotency marker profile as control mESCs. Overall, our data suggest that Rbm47 is not necessary to maintain the pluripotent state of mESCs.
Downregulation of Rbm47 increases primitive endoderm (PrE)-like cells in mESC culture
Since Rbm47 depletion did not affect the pluripotency and self-renewal of mESCs, we next considered analyzing the expression of differentiation markers specific to PrE, mesendoderm, TE, and neuroectoderm. Previously, it was established that mESC cultures display heterogeneity and comprise a subpopulation of lineage-committed cells [36, 37]. RT-qPCR revealed a significant upregulation of PrE markers Gata6, Gata4, Sox17, Dab2, Pdgfra, and Foxa2 in shRbm47 mESCs as compared with control mESCs (Fig. 3A and 3B). In agreement to RT-qPCR data, immunostaining with GATA4 antibody revealed that Rbm47 knockdown mESCs had a greater fraction of PrE-like cells (shRbm47#1 mESCs: 6.2% \(\pm\) 1%; shRbm47#3 mESCs: 6% \(\pm\) 0.4%) than control mESCs (shlacZ mESCs: 2.2% \(\pm\) 0.5%) (Fig. 3C and 3D). The expression of mesendodermal markers remained mostly unchanged, but the expression of a few NE markers (Nestin, Pax6 and Emx1) was downregulated marginally.
FGF-ERK pathway inhibitors or overexpression of Rbm47 can reverse the PrE priming in Rbm47 knockdown mESCs
Since ESCs are in vitro models of the ICM, these cells are routinely cultured in a medium supplemented with leukemia inhibitory factor (LIF). However, they can be converted to a hypomethylated ground state of pluripotency (similar to preimplantation ICM) by culturing in a medium containing inhibitors of GSK3ß and MEK1/2 (termed as '2i') . To determine whether a return to the ground state would reverse Rbm47-depletion induced priming towards the PrE lineage, we cultured control and Rbm47-depleted mESCs for two passages in ‘2i’ supplemented ESC medium and analyzed these for PrE marker expression. We indeed observed that culture condition capturing the ground state was sufficient to reverse the PrE priming in Rbm47-depleted mESCs (Supplementary fig. S4).
It is well-documented that FGF4-ERK signaling is the central pathway in cell-fate determination of the ICM into NANOG-positive Epi and GATA6-positive PrE at the E4.5 stage of the mouse blastocyst (Fig. 3E). Blocking FGF signaling with inhibitors of FGF receptor (FGFR) and ERK is reported to convert ICM into Epi ; in contrast, overactivation of FGF signaling can transform ICM to PrE . To assess the possible effect of Rbm47 depletion in modulating the FGF-ERK pathway, we treated shRbm47 mESCs with PD0325901 (MEK1/2 inhibitor, MEKi) or PD173074 (Pan-FGFR inhibitor, FGFRi) for 48 h and profiled for expression of PrE markers by RT-qPCR and GATA4 + PrE-like cell population by immunostaining. The use of either inhibitor reduced the PrE marker levels significantly, suggesting the involvement of FGF-ERK signaling in increasing PrE-like subpopulation in Rbm47 knockdown mESCs (Fig. 3F and 3G). Further, to confirm the activation of this pathway, we quantified the expression of Fgfr1, Fgfr2, and Fgf4 mRNAs in Rbm47-depleted mESCs; however, there was no significant change in the mRNA levels of the FGF receptors and the ligand as compared with control cells (Supplementary fig. S5). Additionally, we did not observe a spike in phospho-ERK1/2 levels in Rbm47-depleted mESCs, a direct measure of FGF signaling (Supplementary fig. S5).
To overrule any off-target effect of Rbm47 knockdown, we overexpressed Rbm47 in mESCs. We electroporated shRbm47#1 mESCs (we did not use shRbm47#3 as it targets CDS) with pCAG-EGFP-N1 (empty vector) and pCAG-RBM47-EGFP-N1 using Neon transfection system and cultured for 48 hours. We then measured the expression of PrE markers in these cells. We found that overexpression of RBM47 in knockdown cells significantly reduced the expression of PrE markers and GATA4 + PrE-like cells, indicating the rescue of the phenotype (Fig. 3H, 3I and 3J). Collectively, our findings demonstrate that Rbm47-depleted mESCs displayed upregulated PrE related genes and contained an increased population of GATA4 + cells as compared with control mESCs. Overexpression of Rbm47 or culturing these cells in ESC medium supplemented with either 2i, MEKi, or FGFRi could reverse the priming towards PrE, suggesting that RBM47 is implicated in regulating the FGF-ERK pathway in mESCs.
Rbm47 depleted mESCs do not retain a complete multi-lineage differentiation potential
To evaluate the role Rbm47 on differentiation potential of mESCs in vivo, we xenografted shlacZ and shRbm47 mESCs subcutaneously into NSG mice for teratoma formation. Rbm47-depleted mESCs formed significantly smaller teratomas (mean teratoma volume: shRbm47#1 = 251.63 mm3; shRbm47#3 = 324.6 mm3) as compared with control mESC-derived teratoma (mean teratoma volume: 2081.36 mm3), indicating that Rbm47 might be necessary for self-renewal during the differentiation of mESCs (Fig. 4A and 4B). We then subjected the harvested teratomas for sectioning and histological observation following HE staining to look for any skews in the formation of primitive tissues from tri-lineage. Despite the notable gross reduction in the size of Rbm47 knockdown teratoma, we did not find any conclusive changes as primitive structures belonging to all three germ layers were identified (Fig. 4C). Because teratoma assays cannot solely explain the multi-lineage differentiation potential of PSCs due to inconsistencies in injection sites and data reporting formats, as well as the difficulty of quantifying in vivo differentiation kinetics, alternative assays in the form of spontaneous and directed in vitro differentiation have been suggested . We employed an in vitro spontaneous differentiation assay where mESCs were differentiated in the presence of serum for six days as monolayer cultures and the expression of many lineage-specific markers was analysed. We observed a consistent upregulation of extraembryonic endoderm (ExEn) markers such as Gata6, Gata4, Pdgfra, Sox17, and Foxa2 in shRbm47 differentiation cultures as shown in Fig. 4.D. Likewise, the neuroectoderm precursor markers Nestin, Pax6 and Cdh2 were downregulated; however, the expression of mesodermal markers was inconsistent when compared to control mESC differentiation (Fig. 6.2). Together, results from in vitro spontaneous differentiation experiment indicate that Rbm47 knockdown enhances the tendency of mESCs to differentiate into ExEn lineage with compromised neuroectodermal fate, thereby affecting their multi-lineage differentiation capacity.
Lineage-specific differentiation reveals Rbm47 is essential of neuroectoderm and endoderm fate of mESCs
To further probe the effects of Rbm47 depletion on differentiation, we profiled the expression of lineage-specific markers in shRbm47 and control mESCs differentiated into definitive endoderm (DE), mesoderm (MES), and neuroectoderm (NE). As shown in Fig. 1G and 1H, Rbm47 expression undergoes lineage-specific modulation, and the mRNA and protein levels were notably remained stable in wild-type mESC-derived DE. To determine whether Rbm47 is essential for DE formation, we plated control and shRbm47 mESCs to form EBs for 2-days in the serum-free formulation and then directed to DE lineage by supplementing the medium with Activin A. We observed that Rbm47-depleted EBs were significantly smaller and irregularly shaped than the control EBs (Fig. 5A and 5B). RT-qPCR profiling revealed a consistent downregulation of definitive endoderm markers (Sox17, Foxa2, and Hhex) (Fig. 5E), suggesting Rbm47 is necessary for proper differentiation of mESCs to DE.
We next specified Rbm47-depleted mESCs to MES by treating 2-day old EBs with Activin A, VEGF, and BMP4 for two days. A role for Rbm47 in the mesodermal specification was not expected since it was significantly downregulated in wild-type mESC-derived MES (Fig. 1G and 1H) and undetected in mesodermal tissues embryo and adult mouse . However, Rbm47-depleted mESCs not only formed significantly smaller EBs (Fig. 5C and 5D) but also displayed enhanced expression of mesodermal markers such as Mixl1, Kdr, Gsc, Mesp1, and Hand1 (Fig. 5F).
On the other hand, the involvement of Rbm47 in neuroectodermal differentiation was indicated since critical NE progenitor markers (Pax6, Nestin, and Cdh2) were suppressed upon Rbm47 knockdown (Fig. 3A and 4D), we expected to be necessary for proper NE differentiation. As expected, we observed a significant decrease in self-renewal of Rbm47-depleted mESCs on day 2 of NE induction with N2B27-high insulin medium, and the effect persisted throughout the duration (day 8) (Fig. 5G). RT-qPCR analysis of NE-differentiated shRbm47 mESCs revealed a significant downregulation of most NE markers, with a clear bias towards ExEn markers as compared with control mESCs (Fig. 5I). Immunostaining of these cultures with NE markers PAX6 and TUBB3 (B-3-Tubulin) and ExEn marker GATA4 correlated with the RT-qPCR data (Fig. 5H). Collectively, these findings from the lineage-specific differentiation models successfully demonstrated that Rbm47 is essential for fine-tuning the cell-fate decisions and lineage specification of mESCs. Rbm47 knockdown strongly affects the DE and NE differentiation programs.