Toward the human cellular microRNAome

Abstract

MicroRNAs are short RNAs that serve as regulators of gene expression and are essential components of normal development as well as modulators of disease. MicroRNAs generally act cell-autonomously, and thus their localization to specific cell types is needed to guide our understanding of microRNA activity. Current tissue-level data have caused considerable confusion, and comprehensive cell-level data do not yet exist. Here, we establish the landscape of human cell-specific microRNA expression. This project evaluated 8 billion small RNA-seq reads from 46 primary cell types, 42 cancer or immortalized cell lines, and 26 tissues. It identified both specific and ubiquitous patterns of expression that strongly correlate with adjacent superenhancer activity. Analysis of unaligned RNA reads uncovered 207 unknown minor strand (passenger) microRNAs of known microRNA loci and 495 novel putative microRNA loci. Although cancer cell lines generally recapitulated the expression patterns of matched primary cells, their isomiR sequence families exhibited increased disorder, suggesting DROSHA- and DICER1-dependent microRNA processing variability. Cell-specific patterns of microRNA expression were used to de-convolute variable cellular composition of colon and adipose tissue samples, highlighting one use of these cell-specific microRNA expression data. Characterization of cellular microRNA expression across a wide variety of cell types provides a new understanding of this critical regulatory RNA species.

Figure 1.

A generalized overview of the 530 cells and tissues included in this study. (A) Representation of 46 main cell types. (B) Representation of 42 cancer or immortalized cell lines. (C) Representation of 26 tissues/organ types.

Figure 2.

Method and primary analysis of the cellular distribution of microRNAs. (A) Six hundred ninety-four total samples were processed through miRge, yielding 530 samples available for analysis and novel microRNA detection through miRDeep2 and miRanalyzer. (B) t-SNE distribution of 161 primary cells showing four main clusters (hematologic, epithelial, mesenchymal, and neural/stem cell) and subclustering by cell type. Cell types are color-coded, and round symbols indicate epithelial cells. (*) indicates an intestinal epithelial cell that was either contaminated or underwent mesenchymal transformation. (C) A selection of microRNAs that have unique expression to certain primary cell types. (*) indicates specificity for flow-sorted colonic epithelial (likely goblet) cells. (D) Across 334 microRNAs with >1000 RPM, both strands of a hairpin microRNA give rise to the dominant microRNA in fairly equal measures. (E) The presence of nearby superenhancers strongly correlates with high microRNA expression. (F) The individual cellular microRNA patterns can be used to de-convolute the cellular composition of tissue. (G) A representative hematoxylin & eosin (H&E) section of adipose with significant red blood cells (lower part of the panel) as an example of heterogeneous elements that can contribute microRNA expression (10× original magnification). (H) A H&E representative section of adipose with a small cluster of lymphocytes (lower part of the panel) that may be randomly sampled, modulating the tissue signal (10× original magnification).

Figure 3.

Some novel human microRNAs may still await characterization. (A) A nested Venn diagram was parsed down from 21,338 miRDeep2 identified novel microRNA loci to 2724 with ≥50 reads to 984 based on overlap with miRanalyzer-detected novel microRNAs to the 652 with both 5p and 3p mature microRNAs, and finally, with evidence of the RNA being Ago-bound, yielding 495 highest confidence novel microRNAs. (B) A histogram of PhyloP conservation scores averaged across the length of each mature microRNA. This collection of miRBase* v21 microRNAs has the miRGeneDB set removed. TJU novel microRNAs are from Londin et al. (2015). (C) Nine novel microRNAs were amplified that were predicted to be either cell-specific or ubiquitous. Most of these were lowly expressed. miR-21-5p was used as a control. (D) The predicted hairpin structure of novel microRNA JHU_ID_23828 is shown. (E) JHU_ID_23828 is located in the EGFL7 gene locus and shares a pri-miRNA with mir-126, an endothelial cell-enriched microRNA. (F) From the same sequencing batch, the average number of reads for JHU_ID_23828 among four endothelial cell types was 3673 and eight among 29 nonendothelial cell types ([*] P = 0.001, Mann-Whitney U test). (G) JHU_ID_23828 is present among primate species but is absent in lower mammals including Mus musculus.

Figure 4.

IsomiRs are a challenge to characterizing microRNA levels. (A) Among primary cells, the most abundant (dominant) sequence for many microRNAs differs in length from the canonical “C” miRBase.org v21 sequence by up to 4 bases (C−4 to C+4). Between cell types, length diversity is also present, as evidenced by microRNAs that are not entirely of one length. Eighteen representative microRNAs from 556 in total. (B) The general features of microRNA length among cancer/immortalized cells are similar, but the microRNA processing in these cells skews toward randomness. See also Supplemental Figure 12. (C) miR-150-5p, a lymphocyte-specific microRNA, shows a diversity of nontemplated nucleotide addition at the +1 site on the 3′ end. Cytosine is the templated (genomic) nucleotide at this position and is not shown. (D) miR-151a-3p, a ubiquitous microRNA, has marked variation in the first nontemplated nucleotide addition. Cytosine is again the templated nucleotide at this position.