Background

Long noncoding RNAs (lncRNAs) play essential roles in biology but their mechanisms of action can be difficult to determine (Kopp and Mendell, 2018; Gil and Ulitsky, 2020). Relative to protein-coding genes, lncRNAs can evolve rapidly and are not constrained by codon usage (Cabili et al., 2011; Kutter et al., 2012; Necsulea et al., 2014; Schuler et al., 2014; Washietl et al., 2014; Hezroni et al., 2015; Chen et al., 2016; Ulitsky, 2016). Accordingly, they often lack easily identifiable domains that might otherwise provide insight into their molecular actions. Thus, for a given lncRNA, initial footholds into its molecular mechanism are often gained by examining its sub-cellular localization and its intracellular abundance (i.e., on average, how many molecules of the lncRNA are present in a single cell of a cell type of interest).

To these ends, single-molecule sensitivity RNA fluorescence in situ hybridization (smFISH) has been a boon, providing a convenient and cost-effective way to simultaneously investigate lncRNA sub-cellular localization and intracellular abundance (Cabili et al., 2015; Tsanov et al., 2016; Raj and Rinn, 2019). Nevertheless, while smFISH offers unparalleled benefits in regards to visualizing lncRNA sub-cellular distribution, it does have limitations in regards to estimating lncRNA intracellular abundance. In the simplest form of smFISH, the number of FISH puncta per cell is used as a proxy for a lncRNA’s intracellular abundance. However, especially when smFISH is performed using diffraction-limited microscopy, separate lncRNA molecules that are located in spatial proximity may not be individually resolved; instead, such lncRNAs may appear as single puncta. Thus, particularly for those lncRNAs that accumulate to high concentrations in specific subcellular regions (Chujo and Hirose, 2017; Ninomiya and Hirose, 2020), smFISH may underestimate intracellular abundance. This potential limitation can be overcome by performing smFISH using super-resolution microscopy and carefully quantifying signal intensity within individual puncta (Cerase et al., 2014; Smeets et al., 2014; Sunwoo et al., 2015). However, this latter approach requires high-end equipment and expertise that may not be readily accessible.

Here, we describe a sequencing-based approach that is orthogonal to smFISH and can provide estimates for the intracellular abundance of lncRNAs as well as mRNAs in cultured cells (Figure 1; Schertzer et al., 2019). The approach relies on RNA-Seq and requires minimal expertise beyond the ability to follow standard protocols in molecular biology and bioinformatics. The approach allows for the simultaneous quantitation of the intracellular abundance of all long RNAs (lncRNAs and mRNAs) that are expressed in a cell type of interest.

The sequencing-based approach is also likely to yield estimates of intracellular abundance that are more robust than those produced by single-gene approaches that rely on calibrated reverse transcription followed by gene-specific quantitative PCR (RT-qPCR [Schwaber et al., 2019]). A primary reason for this is because different qPCR amplicons spaced across the length of a transcript may vary dramatically in their efficiency of reverse transcription. In contrast, in RNA-seq, local, intra-transcript variations in reverse transcription efficiency are inherently averaged, owing to the chemical fragmentation of RNA that occurs just prior to reverse transcription of the RNA into cDNA (Hrdlickova et al., 2017). Moreover, in the sequencing-based approach, the use of ERCC Spike-In RNAs, which harbor diverse GC-contents, lengths, and abundances, obviates the need to explicitly estimate reverse transcription efficiency and instead allows transcript abundance to be estimated by comparison to a standard curve (Jiang et al., 2011).

Potential sources of errors in the sequencing-based approach include (1) errors associated with read alignment, which are most relevant if the lncRNA of interest contains sequence that is highly repetitive relative to other positions in the genome, (2) errors associated with transcript isoform uncertainty, which may arise if the predominant isoform of the lncRNA of interest is misannotated in the cell type of interest, and (3) errors associated with inaccurate or variable estimates of the total weight of RNA in the cell type of interest. Nevertheless, recently, the sequencing-based approach and smFISH performed using super-resolution microscopy have been shown to arrive at similar estimates of abundance for the lncRNA Xist (Smeets et al., 2014; Sunwoo et al., 2015; Schertzer et al., 2019), lending confidence that both approaches provide reasonable estimates of RNA abundance. The sequencing-based approach is additionally useful because in a single experiment, it can be used to estimate the abundance of all expressed mRNAs and lncRNAs simultaneously.


Figure 1. Overview of sequencing-based approach to quantify lncRNA and mRNA intracellular abundance