The lncRNA DEANR1 facilitates human endoderm differentiation by activating FOXA2 expression

Abstract

Long non-coding RNAs (lncRNAs) regulate diverse biological processes, including cell lineage specification. Here, we report transcriptome profiling of human endoderm and pancreatic cell lineages using purified cell populations. Analysis of the data sets allows us to identify hundreds of lncRNAs that exhibit differentiation-stage-specific expression patterns. As a first step in characterizing these lncRNAs, we focus on an endoderm-specific lncRNA, definitive endoderm-associated lncRNA1 (DEANR1), and demonstrate that it plays an important role in human endoderm differentiation. DEANR1 contributes to endoderm differentiation by positively regulating expression of the endoderm factor FOXA2. Importantly, overexpression of FOXA2 is able to rescue endoderm differentiation defects caused by DEANR1 depletion. Mechanistically, DEANR1 facilitates FOXA2 activation by facilitating SMAD2/3 recruitment to the FOXA2 promoter. Thus, our study not only reveals a large set of differentiation-stage-specific lncRNAs but also characterizes a functional lncRNA that is important for endoderm differentiation.

Figure 1. Purification of samples for transcriptome analysis

(A) Purification of human ESCs, definitive endoderm, and pancreatic progenitors. Upper panel illustrates in vitro differentiation (R, Retinoic acid; N, Noggin; K, KGF), immuno-staining, and FACS sorting of human ESCs (SSEA4+/CD184−), definitive endoderm (CD184+/CD117+), and pancreatic progenitors (CD24^high/CD49f^med). (B) Flow cytometric analysis demonstrates that PDX1-positive cell populations are CD24^high/CD49f^med. (C) Comparison of gene expression of key pancreatic transcriptional factors between CD24^high/CD49f^med (Pos) and the remaining cells (Neg). Shown are three independent experiments. (D) Purification of the alpha and beta cells from human primary islets using HPi2 and HPa1. HPi2+/HPa1− mark beta cells while HPi2+/HPa1+ mark alpha cells. (E) RT-qPCR analysis of cell-type specific marker gene expression confirms the identity of sorted alpha, beta, and non-alpha/beta cells. Shown are three independent experiments.

Figure 2. Expression analysis of protein-coding genes across human endoderm and pancreatic cell lineages

(A) Cluster analysis of transcriptome across the human pancreatic cell lineages. (B) Top enriched Gene ontology terms of each group of stage-specific expressed genes are shown with gene counts and Fisher’s exact test p value. (C) Heatmap illustration of enriched KEGG signaling pathways based on pair-wise comparisons of the various developmental stages (Fisher’s exact test p value < 0.01). Color scale represents respective p value.

Figure 3. Dynamic regulation of lncRNAs during human endoderm and pancreatic cell lineage specification

(A) Hierarchical clustering of the 250 stage-specific lncRNAs during human pancreatic cell lineage specification (Red and blue represent high and low expression, respectively). (B) Top enriched gene ontology terms of the stage-specific lncRNA neighboring protein-coding genes. (C) RNA-seq reads alignment of representative stage-specific expressed lncRNAs. For each lncRNA, the vertical axis is scaled at the same expression level for all samples. The lncRNA transcripts are depicted as black bars with an arrow indicating transcription direction. (D) Relative percentage of stage-specific expressed lncRNAs (red bars) compared to protein-coding genes (gray bars) in five different human pancreatic cell lineage samples (p < 1E-16 for each comparison, Chi-Square test). (E) Endocrine islet-expressed lncRNAs show less correlation with the expression of their nearby genes compared to the lncRNAs in ES (black), DE (blue) and PP (purple) (p = 8.3E-15 compared to ES, p = 2.9E-16 compared to DE, p = 4.2E-8 compared to PP cells, Wilcoxon rank sum test).

Figure 4. Identification of DEANR1 as an endoderm-specific lncRNA

(A) Heatmap showing the expression levels of those lncRNAs differentially expressed in ESCs and DE cells. Criteria: higher FPKM > 5, log2 [fold-change] > 1.5, and p value < 0.05. (Red and blue represent high and low expression, respectively) (B) Relative enrichment of DEANR1 in DE compared with ESCs shown by FPKM. FOXA2 serves as a control. (C) Relative expression level of DEANR1 in various cell types quantified by RT-qPCR. Undifferentiated ESCs (ES), differentiated neuroectoderm (NE), spontaneously differentiated embryonic body (EB), and definitive endoderm (DE) samples were compared. (D) FOXA2 and DEANR1 exhibit a similar expression dynamics during definitive endoderm differentiation. The mean values are shown and error bars represent the s.d. from the mean (n=3). (E) Representative FACS analysis showing reduced CD117/CD184 double-positive definitive endoderm cell population upon DEANR1 knockdown. (F) Representative immunostaining of DE markers SOX17 and FOXA2 in DE differentiated control and DEANR1 knockdown cells. (G) RNA-seq analysis of differentiated DEANR1-knockdown endoderm cells and control endoderm cells. Compared with control (FDR < 0.001 and fold-change > 2), 632 genes were down-regulated (significantly enriched in DE signature genes) and 569 genes were up-regulated upon DEANR1 knockdown (significantly enriched in ES signature genes). Among those, 219 down-regulated genes in DEANR1-knockdown cells are DE-signature genes which should be up-regulated during endoderm differentiation (upper panel); 194 up-regulated genes in DEANR1-knockdown cells are ES-signature genes which should be down-regulated during endoderm differentiation (lower panel).

Figure 5. FOXA2 is a major functional target of DEANR1

(A) Genomic location and diagramatic presentation of RNA-seq results in DE for DEANR1 and FOXA2. (B) Correlation between DEANR1 and FOXA2 expression levels in human ESCs (ES), neuroectoderm (NE), embryonic body (EB), differentiated mesendoerm cells (ME), definitive endoderm cells (DE) and sorted DE. (C) Knockdown of DEANR1 in human ESCs reduces FOXA2 expression. Relative RT-qPCR results were presented with the FOXA2 level in control knockdown ESCs sets as 1. The shRNA#2 was used in C-F. (D) FACS-sorted DEANR1 knockdown DE cells exhibit decreased FOXA2 expression. (E) Scatterplot of the fold-change of the differentially expressed genes in DEANR1-knockdown cells and those in FOXA2-knockdown cells. (F) Venn diagram shows that the genes affected by DEANR1 knockdown largely overlap with those affected by FOXA2 knockdown. The common up-regulated genes in both knockdowns include pluripotent genes (NANOG, SOX2), ectoderm (SOX3, NES) and mesoderm marker genes (WNT5A, THY1); the common down-regulated genes include endoderm-specific genes (FOXA2, GATA3, SOX17 and HNF1B). (G) Representative FACS analysis results from control, DEANR1 knockdown, and DEANR1 knockdown rescue cells. The mean values are shown and error bars represent the s.d. from the mean (n=3). (H) Quantification of the DE differentiation efficiency from control, DEANR1 knockdown, and DEANR1 knockdown rescue cells from three independent experiments.

Figure 6. DEANR1 associates with and helps targeting SMAD2/3 to the FOXA2 promoter

(A) Fluorescent RNA in situ hybridization demonstrates nuclear localization of DEANR1 only in FOXA2-positive DE cells. (B) Dual RNA-DNA-FISH demonstrates DEANR1 transcripts (green signal) are localized to the FOXA2 gene locus (red signal). (C) SMAD2/3 antibodies immunoprecipitate DEANR1, but not U1 snRNA or lncRNA MALAT1. (D) ChIP analysis demonstrates binding of SMAD2/3 to the promoter region of the FOXA2 gene locus. IgG serves as a control. Genomic location of the analyzed regions is indicated in the diagram at the top of the panel. (E) Knockdown of DEANR1 reduces the binding of SMAD2/3 to the FOXA2 promoter, but not to another SMAD2/3 target EOMES. The mean values are shown and error bars represent the s.d. from the mean (n=3). The significance was determined by two-tailed Student’s t-test and p value was presented. (F) Hypothetical model illustrates how DEANR1 might regulate FOXA2 transcription in cis. The genomic locations of DEANR1 and FOXA2 genes are shown on the top line and the arrows indicate transcription direction. Upon DEANR1 activation, chromatin looping brings the DEANR1 locus spatially close to the FOXA2 promoter region (∼20kb between DEANR1 gene-body and FOXA2 promoter). The transcribed DEANR1 interacts with SMAD2/3 protein and brings SMAD2/3 to the FOXA2 promoter through forming DNA-RNA hydride for specific targeting of SMAD2/3 to the FOXA2 promoter. Together with other transcription machinery, binding of SMAD2/3 to the FOXA2 promoter initiates transcription.