Profiling of REST-Dependent microRNAs Reveals Dynamic Modes of Expression

Abstract

Multipotent neural stem cells (NSCs) possess the ability to self-renew and differentiate into both neurons and glia. However, the detailed mechanisms underlying NSC fate decisions are not well understood. Recent work suggests that the interaction between cell type specific transcription factors and microRNAs (miRNAs) is important as resident neural stem/progenitor cells give rise to functionally mature neurons. Recently, we demonstrated that the transcriptional repressor REST (RE1-silencing transcription factor) is essential to prevent precocious neuronal differentiation and maintain NSC self-renewal in the adult hippocampus. Here we show that REST is required for orchestrating the expression of distinct subsets of miRNAs in primary mouse NSC cultures, a physiologically relevant cell type. Using miRNA array profiling, we identified known REST-regulated miRNA genes, as well as previously uncharacterized REST-dependent miRNAs. Interestingly, in response to proliferation and differentiation stimuli, REST-regulated miRNAs formed distinct clusters and displayed variable expression dynamics. These results suggest that REST functions in a context-dependent manner through its target miRNAs for mediating neuronal production.

Keywords: adult neurogenesis; epigenetic; microRNAs; neural stem cell; repressor; transcription.

Figure 1

Context-dependent regulation of REST miRNA targets in mouse NSCs. (A) Successful derivation of multipotent REST cKO NSCs was confirmed using NSC markers (Nestin and Sox2), neuronal (Tuj1) and astrocytic markers (GFAP). Scale bar: 40 μm. (B) REST deletion by Ad-Cre-GFP viral infection in vitro resulted in upregulation of its target genes and neuronal genes, as well as precocious neuronal differentiation (not shown). (C–I) Hierarchical clustering analysis showed that known REST miRNA targets could be categorized in six different clusters (labeled 1–6) according to their expression dynamics during differentiation and responsiveness to REST ablation.

Figure 2

REST ablation modulates the expression of numerous miRNAs during differentiation. (A) The top expressed miRNAs included the let-7 family, miR-709, -9, -21, and -1937a/b. Most of the top expressed miRNAs are either ubiquitously expressed or neural specific/enriched species and many undergo dynamic changes over the course of neural differentiation or response to REST ablation. (B) miRNAs from the oncomiR miR-17-92 and its two paralogous clusters are dynamically regulated during RA and FSK induced NSC differentiation. Loss of REST prevents the initial downregulation mediated by RA and FSK.

Figure 3

Identification of novel REST miRNA targets. (A,B) In unsupervised clustering analyses, known REST miRNA targets formed clusters that also contained other miRNAs. Examples of clusters containing known REST miRNA targets (underlined in red) (C) Combination of bioinformatic analysis and REST ChIP-seq identified novel REST miRNA target genes in Cluster 1 and 2. (D) REST ChIP-QPCR analysis confirmed the binding of REST and CoREST to novel miRNA target genes (seven out of nine randomly chosen), except miR-24-1 and -26a that fell below the twofold cut-off. MiR-129 and -149 were included as known miRNA target controls.

Figure A1

Array profiling of miRNAs in REST cKO NSCs. Whisker box plot analysis demonstrated there was no significant global expression differences among the GFP and Cre-GFP groups regardless of proliferation versus differentiation conditions.

Figure A2

Distribution of the number of miRNAs examined in the array analysis. In NSCs cultured under proliferation and differentiation conditions, only a quarter of the 1040 miRNAs examined are expressed at a significant level (with an intensity above 32). Out of these significantly expressed groups, about 60 miRNAs were expressed at levels between 500 and 5000 and about 20 miRNAs were above 5000.

Figure A3

Expression distribution of miRNAs examined in the array analysis. The top expressed miRNAs account for the majority of the total miRNA expression: about 6% with an expression value between 500 and 5000 count for more than 20–30% of the total miRNA expression and about 2% with an intensity score above 5000 comprised 60–70% of the total expression.

Figure A4

Validation of the array profiling by examining specific miRNA clusters. All miRNAs from the miR-297-669 loci displayed similar expression dynamics demonstrating the validity of the array analysis. Similar expression dynamics were also observed for other miRNA clusters (Figure 2 and Table A2 in Appendix).

Figure A5

Validation of the clustering approach to identify potential novel REST miRNA targets. Regardless of the clustering algorithm, miRNAs from miR-297-669 locus readily form clusters, confirming the validity of our clustering approaches.

Figure A6

Clustering approach to identify potential novel REST miRNA targets. In unsupervised clustering analyses, known REST miRNA targets formed clusters that also contained other miRNAs. Cluster 1 and 2 were chosen as representative examples and presented in Figure 3. Through bioinformatic and ChIP-seq analysis, we identified numerous potential novel REST miRNA targets residing within the same clusters as known REST miRNA targets (refer to Table A7 in Appendix).