Background

Endogenous retroviruses (ERV) are retrovirus-like mobile DNA elements that form ~10% of the murine genome, and are expressed in both healthy tissues and disease states (Kazazian and Moran, 2017; Johnson, 2019). Expression of ERV families is tissue-specific (Faulkner et al., 2009), and many families of ERV are temporally and differentially regulated throughout embryogenesis (Peaston et al., 2004; Rowe and Trono, 2011). We observed that proviral ERV are globally dysregulated in C57BL/6N, but not C57BL/6J, mice, secondary to loss of two transcriptional repressors, suppressor of non-ecotropic ERV (SNERV)-1, and SNERV-2, that promote heterochromatinization at ERV loci (Treger et al., 2019). We observed that in the absence of SNERVs in C57BL/6N mice, ERV are upregulated in a variety of immune cells, as well as in murine embryonic fibroblasts and embryonic stem cells. To identify whether ERV dysregulation is similarly present in gametes, we needed a cost-effective method to successfully generate cDNA sequencing libraries from pooled samples of growing oocytes.

While primordial oocytes can now be efficiently isolated from fetal and neonatal ovaries by flow cytometry (Stewart et al., 2015), antral germinal vesicle (immature) and metaphase II-arrested (mature) oocytes require a technique employing microdissection of superovulated ovaries (Seli et al., 2005; Duselis and Vrana, 2007; Litscher and Wassarman, 2010; Stein and Schindler, 2011; Guzeloglu-Kayisli et al., 2012;). While superovulation increases the yield of immature and mature oocytes, only ~20-30 ovulated oocytes per mouse can be reliably obtained through this method (Litscher and Wassarman, 2010). Prior studies have generated mRNA sequencing libraries from ~700-1,500 oocytes ( Smallwood et al., 2011; Veselovska et al., 2015; Gahurova et al., 2017), whose isolation represents a significant investment of both time and money.

The transcriptomes of immature (GV) and mature (MII) oocytes have been successfully sequenced by single-cell RNA-seq (scRNA-sesq) technologies (Tang et al., 2010; Xue et al., 2013). scRNA-seq permits the identification of cell-to-cell heterogeneity in a sample by detecting high-abundance cellular transcripts (Liu and Trapnell, 2016; Picelli, 2016; Choi and Kim, 2019), and circumvents the need to isolate large numbers of oocytes for bulk sequencing library preparation. However, our aim was to sensitively detect even low-abundance retroelement transcripts in growing oocytes and to detect differential expression between substrains. For this goal, scRNA-seq was not a suitable technique. Additionally, a high percentage of repeat element sequencing reads map to multiple locations in the genome and transcriptome, and their mapping accuracy is improved by using longer sequencing read lengths and paired-end reads (Treangen and Salzberg, 2011). Importantly, while many scRNA-seq methods amplify only the 3′ end of polyadenylated transcripts, the Smart-seq2 method is a low-cost protocol that generates full-length cDNA from mRNA transcripts (Picelli et al., 2014; Haber et al., 2017; Ziegenhain et al., 2017) and could further improve mappability of repeat-associated reads. For these reasons, we developed a modified Smart-seq2 protocol to generate full-length cDNA libraries from low numbers of pooled oocytes for bulk sequencing to permit subsequent analysis of transposable element expression.

This Smart-seq2-based protocol provides a cost-effective method to generate libraries for bulk mRNA-sequencing of oocytes, which results in data suitable for downstream analysis of population-level differences in retroelement expression. This protocol can be further modified to generate sequencing libraries from any rare or difficult-to-isolate population, with broad applications ranging from analysis of transduced primary cells with low transduction efficiency to investigation of transcriptional differences in scarce tissue-resident immune cell populations.