Analysis of MicroRNA Regulation and Gene Expression Variability in Single Cell Data

Introduction

miRNAs are small non-coding RNA molecules that regulate gene expression in metazoan organisms. miRNAs function post-transcriptionally by regulating target genes through facilitated mRNA degradation or translational repression; this potentially reduces mRNA and protein levels [1,2,3]. In addition to affecting the level of gene expression, miRNAs reduce gene expression variability, or ‘noise’, particularly of less expressed genes [4,5]. This latter effect has been hypothesized to reduce stochasticity in gene expression and to confer robustness to genetic pathways [6].

Single-cell RNA sequencing (scRNA-seq) is a rapidly developing technology that enables direct profiling of gene expression in single-cell resolution. Quantification of cell-to-cell variation using scRNA-seq provides deep insights into expression level heterogeneity and stochastic gene expression [7]. A derivative of scRNA-seq, single-cell small RNA sequencing, reveals the expression pattern of the small-sized fraction of RNAs, including (among others) miRNAs, tRNAs, and small nucleolar RNAs [8]. The majority of studies that investigated the miRNA-mRNA regulatory interplay integrated miRNA-mRNA data from ‘bulk’ sequencing experiments (as opposed to single cell), and focused on the expected anti-correlative expression levels between the miRNAs and their target genes [9,10]. Those studies were limited in their ability to measure the effect of miRNAs on the variability of mRNA expression. However, advancements in single-cell mRNA and miRNA sequencing technologies now enable exploring the interplay between miRNA levels and noisy gene expression [11].

Drawbacks of scRNA-seq data include the occurrence of stochastic noise, which is due to the sampling method, the small amount of starting material, and sequencing inefficiency [12]. Approaches that have been proposed to resolve these technicalities include the use of unique molecular identifier (UMI) [13] and external RNA spike-ins [14]. However, technical noise is greater in scRNA-seq than in bulk RNA-seq (non-single cell); hence, accurate quantification and decomposition of technical and biological noise in scRNA-seq remain challenging. Our study evaluated the effect of miRNA on mRNA expression from single-cell RNA sequencing. We showed that the conclusions that can be derived from this analysis are limited, and proposed a few ideas for improvement.

Read More