Systematically labeling developmental stage-specific genes for the study of pancreatic β-cell differentiation from human embryonic stem cells

Abstract

The applications of human pluripotent stem cell (hPSC)-derived cells in regenerative medicine has encountered a long-standing challenge: how can we efficiently obtain mature cell types from hPSCs? Attempts to address this problem are hindered by the complexity of controlling cell fate commitment and the lack of sufficient developmental knowledge for guiding hPSC differentiation. Here, we developed a systematic strategy to study hPSC differentiation by labeling sequential developmental genes to encompass the major developmental stages, using the directed differentiation of pancreatic β cells from hPSCs as a model. We therefore generated a large panel of pancreas-specific mono- and dual-reporter cell lines. With this unique platform, we visualized the kinetics of the entire differentiation process in real time for the first time by monitoring the expression dynamics of the reporter genes, identified desired cell populations at each differentiation stage and demonstrated the ability to isolate these cell populations for further characterization. We further revealed the expression profiles of isolated NGN3-eGFP(+) cells by RNA sequencing and identified sushi domain-containing 2 (SUSD2) as a novel surface protein that enriches for pancreatic endocrine progenitors and early endocrine cells both in human embryonic stem cells (hESC)-derived pancreatic cells and in the developing human pancreas. Moreover, we captured a series of cell fate transition events in real time, identified multiple cell subpopulations and unveiled their distinct gene expression profiles, among heterogeneous progenitors for the first time using our dual reporter hESC lines. The exploration of this platform and our new findings will pave the way to obtain mature β cells in vitro.

Figure 1

The generation of stage-specific hESC reporter lines by labeling sequential developmental genes of β cells. (A) Schematic of a five-stage pancreatic endocrine differentiation protocol from hESCs represented by important marker genes. The genes labeled with eGFP are in green and those labeled with Tdtm are in red; unlabeled genes are in black. (B) The reporter cell lines constructed in this study. Dual-reporter cell lines were named with the second targeted gene locus followed by “DR” for example, the “FOXA2-DR” cell line stands for the “FOXA2-Tdtm/NGN3-eGFP dual-reporter” cell line. Abbreviations: DE (definitive endoderm); PG (primitive gut tube); PF (posterior foregut); PP (pancreatic progenitor); EP (endocrine progenitor); EC (endocrine cells); INS (INSULIN); GCG (GLUCAGON); Tdtm (TdTomato).

Figure 2

Expression dynamics and fidelity of individual reporter genes. (A) Flow cytometric analysis of the expression dynamics of individual reporter genes during hESC differentiation. The cell cultures were analyzed by flow cytometry at the end of each stage. The expression of FOXA2-Tdtm, SOX17-eGFP, PDX1-Tdtm, NKX6.1-Tdtm, NGN3-eGFP, NEUROD1-Tdtm, MAFB-Tdtm and PAX6-Tdtm was measured for stages (S) 1-4; whereas INS-Tdtm for S1-S5. The numbers in the figure indicate percentage of reporter-expressing cells at the end of specific stages. (B) Confocal imaging after co-staining of reporter gene expression with endogenous gene expression. The expressions of FOXA2-Tdtm and SOX17-eGFP were analyzed at S1, day (D) 4; PDX1-Tdtm at S3, D4; NGN3-eGFP was analyzed at S4, D1.5; NKX6.1-Tdtm and NEUROD1-Tdtm were analyzed at S4, D3; MAFB-Tdtm was analyzed at S4, D4; PAX6-Tdtm was analyzed at S4, D5, and INS-Tdtm at S5, D5. The endogenous protein staining is pseudocolored, and the eGFP expression of the dual-reporter cell lines is not shown. The superimposed images are shown in Supplementary information, Figure S2D. Scale bar, 50 μm. Abbreviations: INS (INSULIN); Tdtm (TdTomato).

Figure 3

Different gene reporters mark distinct cell populations at the different differentiation stages of hESCs. (A) At the end of stage (S)1, immunostaining analysis demonstrated that the FOXA2-Tdtm⁺ cells expressed the definitive endoderm marker SOX17, and the SOX17-eGFP⁺ cells expressed the definitive endoderm marker FOXA2. (B) At the end of S2, the FOXA2-Tdtm⁺ cells and SOX17-eGFP⁺ cells expressed the primitive gut-tube cell marker HNF1B. (C) At the end of S3, the PDX1-Tdtm⁺ cells expressed the posterior foregut endoderm markers HNF6 and FOXA2. (D) At the end of S4, the NKX6.1-Tdtm⁺ cells expressed the pancreatic progenitor markers PDX1 and SOX9; the NGN3-eGFP⁺ cells express the endocrine-associated genes NKX2.2 and CHGA. The NEUROD1-Tdtm⁺ cells, MAFB-Tdtm⁺ cells and PAX6-Tdtm⁺ cells represented different populations of endocrine precursors; most of them expressed NKX2.2, but accounted for different percentages of NKX2.2⁺ cells. They also expressed CHGA. (E) At the end of S5, the INS-Tdtm⁺ cells marked the INS-producing cells; these cells expressed the β-cell markers C-PEPTIDE and NKX2.2. Nuclear staining with DAPI (blue) is shown in the merged images. (F) Gene expression analysis of sorted cells at the end of stage 4 showed that MAFB-Tdtm⁺ cells expressed higher level of late-stage endocrine-associated genes (e.g., MAFA, PAX6, INS, etc.) than NEUROD1-Tdtm⁺ cells. Imaging was performed using confocal microscopy. Scale bar, 25 μm. Abbreviations: INS (INSULIN); CHGA (CHROMOGRANIN A); Tdtm (TdTomato); C-PEP (C-PEPTIDE).

Figure 4

Dual-reporter cell lines identify cell subpopulations. (A) The expression dynamics of two reporter genes (eGFP and Tdtm) in the dual-reporter cell lines, showing the existence of cell subpopulations and gene expression transitions between pairs of targeted genes. (B) Distinct cell populations express different combinations of fluorescent proteins in the dual-reporter cell line-derived cultures. The left five cell cultures were from the end of stage 4, and last culture is from stage 5. White arrows, arrow heads and blue arrows indicate eGFP⁻/Tdtm⁺, eGFP⁺/Tdtm⁻ and eGFP⁺/Tdtm⁺ cells, respectively. Scale bar, 25 μm. Abbreviations: INS (INSULIN); Tdtm (TdTomato).

Figure 5

Capture of cell-fate transitions and characterization of cell subpopulations using dual-reporter cell lines. (A) Real-time tracing of NEUROD1-DR cell cultures at stage 4 day 2 implies that NEUROD1-Tdtm⁺ cells were derived from NGN3-eGFP⁺ cells. The arrows indicate newborn Tdtm⁺ cells. (B) Q-PCR analysis of gene expression in isolated cell subpopulations from NKX6.1-DR cell cultures at the end of stage 4. (C) Q-PCR analysis of gene expression in isolated cell subpopulations from INS-DR cell cultures at the end of stage 5. Gene expression values in B and C were first normalized to GAPDH, and then the highest expression value of each gene among the corresponding group of cell subpopulations was set to 1. The color scale indicates the normalized expression values. Abbreviations: INS (INSULIN); Tdtm (TdTomato); Td (TdTomato); GFP (eGFP).

Figure 6

Gene profiling of NGN3-eGFP⁺ cells identified SUSD2 as a novel surface marker for endocrine progenitors and early endocrine cells. (A) Hierarchical clustering of 3 976 genes that are differentially expressed in NGN3-eGFP⁺ and NGN3-eGFP⁻ cells (left). The pancreatic progenitor-associated genes are enriched in NGN3-eGFP⁻ cells; the pancreatic endocrine-associated genes are enriched in NGN3-eGFP⁺ cells (right). The color scale indicates the normalized expression values. (B) Representative differentially expressed potential transcriptional factors, noncoding RNAs and transmembrane proteins. (C) Section staining demonstrated that both PLXNA2 and HRH1 are specifically expressed in GCG-producing cells, but not INS-producing cells in the 18-week human pancreas. (D) The gene expression dynamics of NGN3-eGFP cells during stage 4, showing the co-expression of SUSD2 and eGFP, and the co-expression of SUSD2 and endocrine-associated proteins, varied over time and by gene. (E) Immunostaining of NGN3-eGFP cell cultures at stage 4 day 4 showed that a high co-localization existed between SUSD2 and NGN3-eGFP and between SUSD2 and NKX2.2. Nuclear staining with DAPI (blue) is shown in the merged images. (F) FACS-based gene expression analysis further confirmed the enrichment of endocrine-associated genes in SUSD2⁺ cells and the enrichment of pancreatic progenitor-associated genes in SUSD2⁻ cells (mean ± SEM, n = 3). (G) Flow cytometry analysis showed that NKX2.2⁺ cells were highly enriched in the bound fraction from the MACS experiments using an anti-SUSD2 antibody. Cells at the end of stage 1 were used as cell control. Abbreviations: POS (positive); NEG (negative); INS (INSULIN); GCG (GLUCAGON); CHGA (CHROMOGRANIN A); SST (SOMATOSTATIN); PLXNA2 (PLEXIN A2); HRH1 (HISTAMINE H1 RECEPTOR).

Figure 7

SUSD2 marks endocrine progenitors and early endocrine cells in the developing human pancreas. (A) In the 11-week pancreas, SUSD2 was highly co-localized with NKX2.2, and ∼87% of NGN3⁺ cells also expressed SUSD2. (B) In the 18-week human pancreas, SUSD2 expression was restricted to dispersed NKX2.2^high cells and excluded from islet structures, indicated by clusters of CHGA⁺/NKX2.2^low cells (left) or by clusters of hormone-positive cells (middle); SUSD2⁺ cells expressed little PDX1(right). (C) In the adult pancreas, the expression of SUSD2 decreased to a very low level in pancreatic endocrine-associated cells. (D) RT-qPCR analysis of cells sorted from week 18 to 23 human fetal pancreas samples by MACS using an anti-SUSD2 antibody demonstrated that endocrine progenitor- and early endocrine cell-associated genes were enriched in the bound fractions, whereas pancreatic progenitor-, late-stage endocrine cell- and exocrine-associated genes were enriched in the unbound fractions (mean ± SEM, n = 3). Nuclear staining with DAPI (blue) is shown in the merged images. Scale bar, 50 μm. Imaging was performed using confocal microscopy. Abbreviations: INS (INSULIN); GCG (GLUCAGON); CHGA (CHROMOGRANIN A).