Multi-site Neurogenin3 Phosphorylation Controls Pancreatic Endocrine Differentiation

Abstract

The proneural transcription factor Neurogenin3 (Ngn3) plays a critical role in pancreatic endocrine cell differentiation, although regulation of Ngn3 protein is largely unexplored. Here we demonstrate that Ngn3 protein undergoes cyclin-dependent kinase (Cdk)-mediated phosphorylation on multiple serine-proline sites. Replacing wild-type protein with a phosphomutant form of Ngn3 increases α cell generation, the earliest endocrine cell type to be formed in the developing pancreas. Moreover, un(der)phosphorylated Ngn3 maintains insulin expression in adult β cells in the presence of elevated c-Myc and enhances endocrine specification during ductal reprogramming. Mechanistically, preventing multi-site phosphorylation enhances both Ngn3 stability and DNA binding, promoting the increased expression of target genes that drive differentiation. Therefore, multi-site phosphorylation of Ngn3 controls its ability to promote pancreatic endocrine differentiation and to maintain β cell function in the presence of pro-proliferation cues and could be manipulated to promote and maintain endocrine differentiation in vitro and in vivo.

Keywords: diabetes; endocrine differentiation; insulinoma; neurogenin3; pancreatic development; pancreatic organoids; proneural bHLH transcription factors; β cells.

Graphical abstract

Figure 1

Ngn3 Is Phosphorylated on Multiple Sites (A) Six serine-proline (SP) sites in mouse Ngn3, showing conservation across species. (B) Schematic representation of the SP sites mutated in 6S-A Ngn3 and 2S-A Ngn3. (C) Western blot showing that Ngn3 is phosphorylated in mammalian HEK cells; treatment with and without phosphatase λ (λ-PP) is indicated. (D) Western blot showing that 6S-A and 2S-A Ngn3 are phosphorylated in insulinoma-derived MIN6 cells, with tubulin as a loading control. (E) Ngn3 immunostaining in E14.5 mouse embryonic pancreas; nuclei are counterstained with DAPI (blue). Scale bars, 50 μm (E) and 10 μm (E’). (F) Ngn3 expression and phosphorylation in HEK cells overexpressing Ngn3 (Ngn3 o/e) compared with E14.5 murine embryonic pancreas; λ-PP, phosphatase λ. Solid and open arrowheads in (C, D, and F) indicate un(der)phosphorylated and phosphorylated Ngn3, respectively. See also Figure S1.

Figure 2

Ngn3 Is Phosphorylated by Cyclin-Dependent Kinases (A) SDS-PAGE separation of in vitro translated (IVT) radiolabeled WT Ngn3 or 6S-A Ngn3 incubated in interphase (I) or mitotic (M) Xenopus extracts, treated with phosphatase λ (λ-PP), as indicated, or (B) incubated in I extract plus cyclin B Δ90; samples removed at increasing times. Solid and open arrowheads indicate un(der)phosphorylated and phosphorylated Ngn3, respectively. (C) In vitro kinase assay showing IVT WT and 6S-A Ngn3 proteins after incubation with human recombinant CYCLIN/CDKs, as labeled. (D) Schematic of the activity of cyclin/Cdks in the different phases of cell cycle. (E) Western blot of HA-tagged Ngn3 expressed in ductal mPAC cells after treatment with Cdk inhibitors, treatment of Ngn3 with λ-PP as a positive control for protein dephosphorylation, tubulin as loading control. (F) Graphs showing the relative amount of the slowest migrating band of phosphorylated Ngn3, compared with the total amount of Ngn3 protein. n = 3 independent experiments, a representative blot is shown. Mean ± SEM. Student's t test, ^∗p < 0.05, ^∗∗p < 0.01.

Figure 3

6S-A Ngn3 Enhances α and δ Cells in the Embryonic Pancreas (A) WT and 6S-A Ngn3 mouse model schematic. (B and C) Immunohistochemistry for eYFP (green), glucagon (red), and insulin (gray) in E16 embryonic pancreas from WT (B) and 6S-A (C) Ngn3 animals, nuclei counterstained with DAPI (blue). Scale bar, 200 μm. (D and E) Quantification of the percentage of areas that are eYFP+, insulin+, glucagon+ (D), somatostatin+ (Sst), and pancreatic polypeptide Y+ (Ppy) (E), in E16 embryonic pancreas. n = 4 mean ± SEM. Student's t test, ^∗p < 0.05, ^∗∗p < 0.01. (F) Relative hormone gene expression in adult pancreatic islets by qPCR, normalized to EF-1α expression. n = 3 mean ± SEM. See also Figures S2–S4.

Figure 4

6S-A Ngn3 Shows Enhanced Target Gene Expression in Pancreatic Organoids (A) Inducible lentiviral vectors (illustrated) were used to infect ductal organoid cultures, images 2 days post-induction. Scale bars, 100 μm. (B) Relative mRNA expression of Ngn3 and its downstream targets Neurod1 and Insm1 after 2 days of Ngn3 expression in organoids, normalized to β-actin. Data are mean ± SEM (n = 3). Student's t test, ^∗p < 0.05, ^∗∗p < 0.01. (C) Genome-wide transcriptomic analysis of pancreatic organoids expressing WT and 6S-A Ngn3. Graph showing the relative fold-change of expression in 6S-A Ngn3 organoids compared with WT Ngn3 fold-change (WT Ngn3 set as 1 unit) at 8 days after Ngn3 induction. Data represent mean fold change ± SEM (n = 3). (D) Average relative log2 fold-change of expression in potential Ngn3 targets compared with control (all genes excluding Ngn3 targets) at 8 days. Data represent average log2 fold change and error bars represent 95% confidence intervals of the mean (n = 3). ^∗∗p < 0.01. See also Figure S5.

Figure 5

Ngn3 Phosphorylation Controls Protein Stability and Binding to Target Genes (A–C) HA-tagged WT Ngn3 (A), 6S-A (B), and 2S-A Ngn3 (C) protein expression following cycloheximide (CHX) addition (in min). (D) Graph showing degradation rate for different Ngn3 mutants, normalized to tubulin. n = 3 independent experiments, mean ± SEM. Student's t test, ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗∗p < 0.0001. (E) Chromatin immunoprecipitation (ChIP) from ductal mPAC cell extracts containing normalized amounts of HA-tagged Ngn3 WT and 6S-A. Data represent mean ± SEM. (n ≥ 3) Student's t test, ^∗p < 0.05, ^∗∗∗p < 0.001. See also Figure S6.

Figure 6

6S-A Ngn3 Maintains Insulin Expression in Adult β Cells Expressing c-Myc (A) eYFP expression in live culture of Ngn3-eYFP islets. Scale bar, 40 μm. (B) Schematic representation of animal models used for in vivo insulinoma analysis. (C and D) Quantification of blood glucose measurements (in mmol/L) in Bcl-xL (C) and Ins-cMycER^TAM Bcl-xL (D) mice at 0, 4, and 7 days of tamoxifen treatment. n ≥ 4 mean ± SEM. Student's t test, ^∗p < 0.05. (E and F) Immunohistochemistry to detect insulin (red) and glucagon (green) in adult pancreatic sections from Bcl-xL only (E) or Ins-cMycER^TAM Bcl-xL (F) animals crossed with WT or 6S-A Ngn3 mice, as labeled, at 7 days of tamoxifen. Nuclei were counterstained with DAPI. Scale bar, 50 μm in (E, F, F′, and F’’). (G and H) Quantification of hormone intensity, divided by islet area in Bcl-xL (G) and Ins-cMycER^TAM Bcl-xL crossed with WT or 6S-A Ngn3 mice, as indicated (H) at 7 days of tamoxifen. Data represent mean ± SEM. n ≥ 3 different animals (islet analyses, 2–6 different sections for each animal for each condition, see STAR Methods for details). Student's t test, ^∗p < 0.05. See also Figure S7.

Figure 7

Ngn3 Dephosphorylation Enhances Cell-Cycle Exit in Pancreatic Ductal Cells (A) Graph showing the growth of pancreatic ductal mPAC cells at 24, 48, and 72 hr after transfection with WT and 6S-A Ngn3 and GFP or GFP only, counting GFP+ cells. Data are mean ± SEM of four independent experiments. Student's t test, ^∗p < 0.05, ^∗∗p < 0.01. (B) Representative images of mPAC cells 72 hr after transfection. Scale bar, 100 μm. (C) Chromatin immunoprecipitation (ChIP) from ductal mPAC cell extracts expressing HA-tagged WT and 6S-A Ngn3. Data represent mean ± SEM (n ≥ 3). Student's t test, ^∗p < 0.05. (D) Relative mRNA expression of Cdkn1a after 2 days of Ngn3 expression in ductal PAC cells, normalized to β-actin. Data are mean ± SEM (n = 3). Student's t test, ^∗p < 0.05, ^∗∗p < 0.01. (E) Model illustrating how Ngn3 phosphorylation controls the balance between proliferation and differentiation during development of the endocrine pancreas. Cdk, cyclin-dependent kinase; Cdki, Cdk inhibitor; P, phosphorylation.