Arginine 65 methylation of Neurogenin 3 by PRMT1 is a prerequisite for development of hESCs into pancreatic endocrine cells

In the developing pancreas, Neurogenin 3 (NGN3) is a key transcription factor in the cell fate determination of endocrine progenitors (EPs). Although the activation and stability of NGN3 are regulated by phosphorylation, the role of arginine methylation of NGN3 is poorly understood. Here, we report arginine 65 methylation of NGN3 is absolutely required for the pancreatic lineage development of human embryonic stem cells (hESCs) in vitro. First, we found inducible protein arginine methyltransferase 1 (PRMT1)-knockout (P-iKO) hESCs did not differentiate from EPs to endocrine cells (ECs) in the presence of doxycycline. Loss of PRMT1 caused an accumulation of NGN3 in the cytoplasm of EPs and blocked NGN3’s transcriptional activity in PRMT1-KO EPs. We also found that PRMT1 specifically methylates NGN3 arginine 65, and this modification is a prerequisite for ubiquitin-mediated NGN3 degradation. Our findings indicate arginine 65 methylation of NGN3 is a key molecular switch in hESCs in vitro permitting the differentiation into pancreatic endocrine lineages.

Introduction methyltransferase in mammalian cells, methylates arginine residues in conserved glycine and 48 arginine-rich (GAR) regions of histone and non-histone proteins (Bedford and Richard, 2005;49 Tang et al., 2000). Through its role in arginine methylation, PRMT1 contributes to diverse 50 cellular processes, such as epithelial-mesenchymal transition (EMT), cell cycle, DNA repair, Next, we investigated the developmental competence of P-KO EP cells to become pancreatic 142 ECs. We found some EC-associated genes (i.e., NKX2.2 and CHGA) with reduced expression 143 in P-KO ECs and others (i.e., PDX1, HNF1b, and MAFA) with normal expression ( Figure 3F). 144 Transcripts of genes downstream of NGN3 (i.e., NEUROD1, PAX6, and PAX4) were reduced 145 in P-KO ECs ( Figure 3F). Among the pancreatic endocrine hormone genes, we found specific we examined NGN3 expression in the cytoplasmic and nuclear compartments of P-iKO EP 151 cells. When compared to untreated cells, dox-treated P-iKO EP cells showed higher levels of 152 NGN3 protein in whole cell lysates (WCL) and in the cytoplasm (Cyt) but no difference in 153 the nuclear compartment ( Figure 3I). On western blot, we observed multiple nuclear NGN3 154 bands in P-iKO EP cells but only a single cytoplasmic NGN3 band. Thus, NGN3 accumulates 155 in the cytoplasm of P-KO EP cells. These data suggest PRMT1 KO causes NGN3 156 accumulation in EP cells, leading to impaired EC differentiation.
Next, we asked whether arginine methylation of NGN3 influences the expression of its 189 downstream genes. After transfecting FLAG-tagged WT-NGN3 and R65A-NGN3 into HEK 190 cells, we found R65A-NGN3 triggered less activation of NEUROD1 than WT-NGN3 ( Figure   191 S5A). We then confirmed this reduced transcriptional activity of the R65A-NGN3 mutant via 192 luciferase assay. To do so, we co-transfected HEK cells with each expression vector and a 193 firefly luciferase reporter containing the NEUROD1 promoter (Huang et al., 2000;Wang et 194 al., 2006). We first confirmed that the expression of each FLAG-tagged version of NGN3 was 195 similar ( Figure S5B) and then normalized the firefly luciferase signal to the renilla luciferase 196 signal. As expected, we observed significantly higher activity triggered by the  transfectants than the R65A-NGN3 transfectants ( Figure 4G, p<0.05). This result indicates 198 that the R65A mutation blocks NGN3 transcriptional activity. Together, our findings suggest 199 methylation of NGN3 arginine 65 is essential for its transcriptional activity.

200
Arginine 65 methylation of NGN3 is a prerequisite for its degradation 201 NGN3 is transiently expressed in pancreatic EP cells, rapidly disappearing during pancreatic 202 lineage development (Apelqvist et al., 1999). We therefore asked whether arginine 203 methylation of NGN3 is associated with its degradation. To address this question, we 204 transfected P-iKO PE cells with an NGN3 overexpression vector (pCAG-FLAG-NGN3) and 205 incubated the transfected cells for 1-2 d ( Figure 5A). Although ectopic NGN3 was 206 significantly reduced 2 days after transfection in P-iKO PE cells that were not treated with 207 doxycycline, similar cells treated with doxycycline maintained consistent levels of NGN3 208 ( Figure 5B). This suggests PRMT1 is associated with NGN3 degradation in hESCs during 209 pancreatic lineage differentiation.

210
To investigate the role PRMT1 plays in NGN3 degradation, we transfected HEK cells with 211 PRMT1-specific siRNAs, waited 48 h, further transfected them with a FLAG-NGN3 expression vector, waited another 24 h, and then treated them with cycloheximide (CHX) to 213 block protein synthesis. We observed a significant delay in FLAG-NGN3 degradation in the 214 siPRMT1 group compared to the siCTL group ( Figure 5C, p < 0.001). We also found that 215 FLAG-NGN3 R65A showed a longer half-life than FLAG-NGN3 WT ( Figure 5D, p < 0.001).

216
Unlike R65A-NGN3, WT-NGN3 appeared in multiple distinct bands on western blot ( Figure   217 5D). In addition, we observed differences in the various phosphorylation bands for FLAG-218 tagged WT-NGN3 and R65A-NGN3 ( Figure S6A, arrowheads) that were sensitive to 219 phosphatase treatment ( Figure S6B, arrowheads). These results indicate arginine methylation 220 contributes to NGN3 stability by altering its phosphorylation.

221
To examine the correlation between arginine methylation and polyubiquitination of NGN3, 222 we transfected HEK cells with FLAG-tagged WT-NGN3 plasmids and siPRMT1 and then 223 treated the transfected cells with the proteasome inhibitor MG132 (20 μM) for 2 h. In the 224 presence of proteasome inhibitor, the ubiquitinated NGN3 bands were slightly reduced in the 225 siPRMT1 group compared to the siCTL group ( Figure 5E). R65A-NGN3 also showed 226 reduced ubiquitination bands compared to FLAG-tagged WT-NGN3 ( Figure 5F). These 227 results suggest PRMT1-mediated arginine methylation is associated with NGN3 228 polyubiquitination. Together, our results indicate PRMT1 specifically methylates arginine 65 229 of NGN3 in hESCs during pancreatic EC development. This methylated NGN3 then 230 undergoes ubiquitin-mediated degradation.

233
Here, we report for the first time that arginine 65 methylation of NGN3 is essential in the 234 pancreatic lineage differentiation of hESC-derived EP cells to ECs. While we did find that  Figure 3G). Since β-and δcells arise from the same ancestor (DiGruccio et al., 258 2016;Mfopou et al., 2010), it is clear that PRMT1-KO contributes to the specification of EP 259 cells into β-and δcells. We found that arginine 65 of NGN3 is a conserved methylation 260 motif among mammals whose methylation is specifically catalyzed by PRMT1 (Figure 4).

261
Furthermore, unmethylated NGN3 was not degraded via ubiquitin-mediated proteolysis 262 ( Figures 5E and 5F). Our results indicate NGN3 arginine 65 methylation is crucial for EC 263 fate determination in pancreatic lineage development.

264
To induce PRMT1 loss of function at a specific stage of pancreatic development, we 265 generated a PRMT1-inducible KO (P-iKO) system in hESCs using a "safe harbor" AAVS1 266 locus. We selected this AAVS1 locus because it does not produce any adverse effects on 267 differentiation, transgene silencing, or proliferation in gene-edited hiPSCs (Hockemeyer et al., 268 2011;Smith et al., 2008). We performed homology-directed repair (HDR) using ZFNs to 269 avoid random transgene integration. When we cultured P-iKO hESCs in the presence of 270 doxycycline, most died within 2 d ( Figure 1C). This implies PRMT1 is essential for the 271 viability and proliferation of hESCs, which is consistent with the embryonic lethal phenotype 272 of PRMT1-KO mice (Pawlak et al., 2000). In the absence of dox, however, P-iKO hESCs 273 maintained pluripotency, differentiation competence, and Cas9 mRNA expression for more 274 than 3 months (data not shown). Thus, our use of the AAVS1 locus with these transgenes 275 permitted stable enough gene expression for the maintenance and differentiation of hESCs.

276
NGN3 protein is rapidly degraded by the canonical ubiquitination pathway (Vosper et al.,277 2009). In the cytoplasm of mouse EPs, NGN3 is degraded by the ubiquitin-proteasome 278 system after phosphorylation (Krentz et al., 2017;Vosper et al., 2009). As in mouse EPs, we 279 observed multiple NGN3 phosphorylation bands in the nuclear fraction of EP cells derived from hESCs ( Figure 3I). We observed a different electrophoretic mobility for the NGN3 281 R65A mutant compared to WT NGN3 ( Figure S6, arrowheads), suggesting arginine 65 282 methylation of NGN3 affects its phosphorylation. In addition, we found that most of the 283 endogenous NGN3 protein in P-KO EP cells was localized to the cytoplasm ( Figure 3I). As 284 with the NGN3 R65A mutant ( Figure 5D), KD of PRMT1 slowed NGN3 degradation in HEK 285 cells ( Figure 5C). Therefore, we concluded arginine 65 methylation of NGN3 is associated 286 with NGN3 phosphorylation, leading to NGN3 degradation. This degradation is required for 287 the cell fate transition of EPs to ECs during human pancreatic lineage development.

288
From these results, we propose a model for the role of PRMT1 in pancreatic endocrine 289 lineage development ( Figure 5G). In this model, NGN3 arginine 65 is specifically methylated

634
Overall protocol for the differentiation of hESCs into pancreatic endocrine cells.

Figure S2. Differentiation of P-iKO hESCs into pancreatic ECs
(A) Cellular morphology at the various stages through which P-iKO hESCs progress as they differentiate into ECs. Scale bars, 200 μm.