Two Tales of Single-Cell RNA Sequencing: Gene Expression and Alternative Splicing in Mouse Kidney Development

Mammalian nephrogenesis is a process of cell fate acquisition and segregation driven by the reciprocal interactions between the ureteric bud (UB) and adjacent metanephric mesenchyme (MM).¹ In response to signals released from MM, the UB branches extensively, forming the collecting duct system. In turn, the tip of the UB induces the MM to self-organize into the Six2-expressing cap mesenchyme, which represents a pool of nephron progenitor cells. These Six2-expressing cells subsequently undergo a mesenchymal-to-epithelial transition, forming the nephron epithelium. As nephrogenesis proceeds, the distal ends of nephrons interconnect with the collecting duct system, giving rise to a network of continuous renal epithelial tubules. At the same time, vascular and other nonepithelial cells develop in parallel with epithelial cells, generating a highly organized and heterogeneous kidney tissue.

Nephrogenesis requires coordinated temporal and spatial gene expression and regulation. Growing evidence suggests that alternative splicing also plays a crucial role during development, including the kidney.² Alternative splicing is a post-transcriptional process in Eukaryotes that allows individual genes to generate more than one mRNA isoform. The resulting isoforms produced from a single gene can have unique properties or can be translated into proteins with different cellular functions. Thus, alternative splicing has greatly expanded transcriptome and proteome diversity. The advent of RNA sequencing (RNA-seq) has allowed investigators to characterize not only which genes are expressed but also, which alternatively spliced isoforms are produced. This is done computationally by identifying exon/intron-spanning reads. Because alternative splicing analysis requires a requisite sequencing depth and a requisite amount of RNA starting material, current analyses were largely limited to the bulk RNA-seq level, which can lose cell type–specific information.

Single-cell RNA sequencing (scRNA-seq; including single-nucleus RNA-seq) has emerged as a powerful tool to dissect transcriptomic heterogeneity and delineate cell types or cell states in complex tissues by measuring the expression of RNA in individual cells.³ The most popular methods for scRNA-seq use droplet-based platforms (e.g., 10× Chromium).⁴ These platforms have greatly increased the high-throughput capacity, allowing the analysis of tens of thousands of cells in a single run. However, they typically capture only the 3′ or 5′ end of the polyadenylated transcripts in the final sequencing library for quantification of gene expression, thus precluding recovery of full-length transcript information. Therefore, transcript alternative splicing analysis is not feasible in these 3′ or 5′ end-counting scRNA-seq data. Although sequencing the full transcript in one read (e.g., Pac-Bio technology) without the short-reads assembly step seems attractive, such technology is not widely available. An alternative is the full-length Smart-seq technology utilized in a paper in this issue of JASN from Wineberg et al.⁵ This method typically provides read sequence information throughout the length of transcripts, in contrast to droplet-based approaches. Wineberg et al.⁵ integrated the Smart-seq–based scRNA-seq method to investigate alternative splicing during mouse kidney development. This study takes advantage of a transgenic mouse line expressing GFP under the control of the promoter of Six2⁶ to enrich cap mesenchyme cells (GFP⁺) and other cell types from embryonic day 18.5 mouse kidney, followed by single-cell analysis. A total of 544 full-length, deep transcriptomes (1–2 million reads per cell) were obtained for both gene expression and alternative splicing analyses.

The authors used a panel of markers to detect clusters of several cell types in developing kidneys.⁵ Specifically, the largest cell cluster was enriched in Six2 (a homeobox transcription factor), confirming the identity as cap mesenchyme. Some clusters had limited abundance of cell types corresponding to cap mesenchyme derivatives (Cdh1 [E-cadherin]), including proximal tubule cells (Slc22a6), cells in the loop of Henle region (Slc12a1), and distal convoluted tubule–like cells. Another cell cluster that expressed Snail2 (mesenchyme cell marker) but not Six2 was likely uninduced mesenchyme. Also identified were cell clusters representing endothelial cells and macrophages. Overall, the authors confirm the expected cell diversity during embryonic kidney development. Given that the analysis was carried out only at embryonic day 18.5, when the cells are still in an active developmental phase, it would be interesting to see how these “cell types” differ from the terminally differentiated cells of the adult kidney. With the increasing availability of datasets from prior studies in both developing kidneys⁷ and adult kidneys,^8,9 such comparisons may provide insights into developmental programs. Resolution and information for cells of the collecting duct (principal cells and intercalated cells) and the connecting tubule were limited. This leaves an open question about how the ureteric lineage and mesenchyme lineage are interconnected.

Next, to obtain sufficient reads for alternative splicing analysis, the authors analyzed combined raw reads from cells from each cluster.⁵ This strategy revealed plenty of alternative splicing events between cell types, many of which were consistent with prior findings (e.g., Map3k7, Ctnnd1), confirming the power of the methodology. Of particular interest are the isoform switches between the mesenchymal and epithelial cells. A number of splicing events seem to be related to the mesenchyme-to-epithelium transition, showing a restrictive pattern in either mesenchymal or epithelial cells (e.g., Actn1, Tpm1). Additional supervised analysis revealed alternative splicing events that may relate to prior knowledge about kidney development (e.g., Fgfr2, Fat1).

Overall, these findings are novel and may provide the basis for the discovery of new mechanisms underlying kidney development. However, linking these alternative splicing events directly to protein functions and ultimately, to kidney functions remains to be investigated. Identifying protein domains that are differentially present in alternative isoforms could illuminate these links. However, alternative splicing events could have effects independent of protein structure, such as on mRNA stability, mRNA localization, or translation efficiency. Finally, coupled with gene expression and motif analysis, the potential splicing regulators were inferred. Specifically, Esrp1/2 and Rbfox1/2 splicing factors exhibited distinct distributions in the mesenchymal and epithelial cells, reinforcing observations that they might be involved in kidney development. However, regulation of these splicing factors remains to be investigated.

The pioneering integration of alternative splicing data with gene expression data at the single-cell level will likely provide new ways to model kidney development and disease. Further technical advances in RNA-sequencing platforms teamed with computational tools to perform multiple measurements in single cells (single-cell “multiomics”)¹⁰ will bring the ultimate understanding of kidney development closer.

Disclosures

The author has nothing to disclose.

Funding

None.