Background

The appearance of single-cell-omics has revolutionized our understanding of many biological processes and is a field that is rapidly evolving at both the experimental and computational level. Over the last decade, there has been a rapid increase in the number of features that can be measured within individual cells. Single-cell RNA-seq (scRNA-seq) has pioneered this endeavor, and it has become a routine experiment to perform in higher eukaryotes. While very close to the entirety of the repertoire of bulk experiments in higher eukaryotes now have single-cell counterparts, single-cell tools for microorganisms (beyond fluorescent reporter-based methods) are scarce. This technological gap is mainly due to technical limitations imposed by the intrinsic nature of yeast and other microorganisms.

Unlike the mammalian cells, in which scRNA-seq was developed, yeast cells are smaller in cell, genome and transcriptome size. The presence of a rigid cell wall, which needs to be removed prior to library preparation, has been one of the major technical challenges for single cell analysis. On top of that, the yeast genome is highly condensed and pervasively transcribed. Over 85% of their genome is expressed from both strands with extensive transcript isoform diversity per gene (David et al., 2006; Pelechano et al., 2013, 2014 and 2015), and this requires a sensitive, quantitative and strand-specific approach.

As was the case for higher eukaryotes, initial studies exploring scRNA-seq in yeast focused on the development and optimization of the method. Recently, three yeast single-cell RNA-seq approaches have been reported (Gasch et al., 2017; Nadal-Ribelles et al., 2019; Saint et al., 2019). Interestingly, each one of them uses a different cell-isolation strategy and library-preparation protocol. Cell isolation is one of the main differences across these studies that range from microfluidics (Gasch et al., 2017), micromanipulation (Saint et al., 2019) and index sorting (Nadal-Ribelles et al., 2019), the later described here. Microfluidic and micromanipulation-based approaches require specialized equipment and are labor intensive, but allow imaging of individual cells by microscopy as a way to measure phenotype. The protocol described here for yeast single-cell RNA-seq (yscRNA-seq), relies on index sorting as a strategy for recording phenotypic information such as cell morphology and any Fluorescent-Activated Cell-Sorting-compatible feature (fluorescent proteins, antibodies, etc.). Index sorting provides a seamless strategy for merging techniques for individual phenotyping and transcriptional profiling in cells.

A major tipping point for single-cell isolation was the appearance of droplet-based methodologies (Zhang et al., 2019). These technologies dramatically increase potential throughput by generating and isolating cDNA in barcoded gel emulsions, which encapsulate single cells on the order of thousands of cells per run, and thus the cost significantly decreases. Interestingly, a preliminary droplet-based study in yeast is underway (Gresham et al., 2019). While cell capturing boosts cell numbers to an unprecedented scale, the number of genes detected per cell tends to be much lower (Skinnider et al., 2019) and phenotypic information of individual cells is lost. In addition, customization of library preparation is complex and labor intensive in droplet based-approaches, which will most likely be bypassed as technologies are developed.

While cell-isolation strategies determine the number of cells and phenotypic information that can be recorded, the choice of library-preparation approach defines the kind of transcriptional profiling. Due to the intertwined nature of the yeast transcriptome, and the diversity of isoforms per gene, library preparation requires both gene- and strand-isoform-specific resolutions. YscRNA-seq is based on STRT-seq (Islam et al., 2014), and its library design bypasses the technical limitations of working with yeast, fulfilling the requirements for a high-resolution transcriptome profiling method. First, unique molecular identifiers (UMI) are incorporated during cDNA synthesis via a biotinylated template-switching oligo (TSO) at the 5′-end of the molecule to digitally count the absolute abundance of a gene, along with its transcription start site (TSS) position. Second, the use of a homemade Tn5 enzyme to incorporate cell-specific adaptors provides a cost-effective strategy that greatly reduces experimental costs, compared to commercial Tn5 (Hennig et al., 2018). Finally, biotinylated primers allow selectively recovering 5′ ends. Moreover, the high affinity between streptavidin and biotin leads to a sequestration of the biotinylated strand after a brief denaturing step, which releases the non-biotinylated strand into the supernatant for sequencing. These features make yscRNA-seq one of the most sensitive methods available, and the most sensitive method for yeast reported to date (Nadal-Ribelles et al., 2019).

We anticipate that the development of novel methodologies with different aims will soon start to emerge, and yscRNA-seq will be extended to any yeast species or microorganism. The protocol described here, along with future optimizations, could serve as a starting point for the development of new methods. There are many unanswered biological questions that influence our understanding of human health, such as the emergence of drug-resistant microbial phenotypes. Generation of solid frameworks to profile microorganisms at single-cell resolution promises to expand our understanding, and perhaps answer some of these questions once and for all.