A Caenorhabditis elegans RNA-Directed RNA Polymerase in Sperm Development and Endogenous RNA Interference

Short interfering RNAs (siRNAs) are a class of regulatory effectors that enforce gene silencing through formation of RNA duplexes. Although progress has been made in identifying the capabilities of siRNAs in silencing foreign RNA and transposable elements, siRNA functions in endogenous gene regulation have remained mysterious. In certain organisms, siRNA biosynthesis involves novel enzymes that act as RNA-directed RNA polymerases (RdRPs). Here we analyze the function of a Caenorhabditis elegans RdRP, RRF-3, during spermatogenesis. We found that loss of RRF-3 function resulted in pleiotropic defects in sperm development and that sperm defects led to embryonic lethality. Notably, sperm nuclei in mutants of either rrf-3 or another component of the siRNA pathway, eri-1, were frequently surrounded by ectopic microtubule structures, with spindle abnormalities in a subset of the resulting embryos. Through high-throughput small RNA sequencing, we identified a population of cellular mRNAs from spermatogenic cells that appear to serve as templates for antisense siRNA synthesis. This set of genes includes the majority of genes known to have enriched expression during spermatogenesis, as well as many genes not previously known to be expressed during spermatogenesis. In a subset of these genes, we found that RRF-3 was required for effective siRNA accumulation. These and other data suggest a working model in which a major role of the RRF-3/ERI pathway is to generate siRNAs that set patterns of gene expression through feedback repression of a set of critical targets during spermatogenesis.REPRESSION of gene expression by small RNAs of ∼20–30 nt in length is important for many aspects of multicellular eukaryotic development. A variety of classes of small RNA with distinct structural features, modes of biogenesis, and biological functions have been identified (reviewed in Hutvagner and Simard 2008). We are particularly interested in a class of small RNAs, called endogenous short interfering RNAs (siRNAs), that are similar to intermediates in exogenously triggered RNA interference (RNAi) in their perfect complementarity to mRNA targets. High-throughput sequencing technology has provided a valuable tool for characterization of endogenous siRNA populations from many diverse sources, including mouse embryonic stem cells (Babiarz et al. 2008), Drosophila heads (Ghildiyal et al. 2008), and Arabidopsis pollen (Slotkin et al. 2009). These siRNAs have been proposed to function in the regulation of both cellular processes and genome defense through downregulation of gene expression. Caenorhabditis elegans, like plants and fungi, utilizes RNA-copying enzymes called RNA-directed RNA polymerases (RdRPs) as part of the RNAi machinery (Smardon et al. 2000; Sijen et al. 2001). While two of the C. elegans RdRPs are nonessential (RRF-1 and RRF-2), mutations in either of the remaining two (EGO-1 or RRF-3) lead to fertility defects (Smardon et al. 2000; Simmer et al. 2002). RRF-3 is functionally distinct from EGO-1 in that the RRF-3 requirement in fertility is temperature dependent. In addition, RRF-3 activity has an inhibitory effect on exogenously triggered RNAi (resulting in an ERI, or enhanced RNAi, mutant phenotype in rrf-3 mutants). Mutants lacking either RRF-3 or another ERI factor, ERI-1, have been used as experimental tools because of their enhanced sensitivity in RNAi-based screens. One proposed mechanism for the enhancement in RNAi in rrf-3 and eri mutants has been a competition for cofactors between the exogenously triggered RNAi pathway and an endogenous RNAi pathway. Consistent with this hypothesis, siRNAs corresponding to several genes have been shown by Northern analysis to depend upon RRF-3 and other ERI factors for their accumulation (Duchaine et al. 2006; Lee et al. 2006; Yigit et al. 2006). Global microarray analyses have also been undertaken to identify messenger RNAs whose expression is affected by RRF-3 and ERI-1 (Lee et al. 2006; Asikainen et al. 2007).A functional significance of the RRF-3/ERI pathway has been inferred by the inability of rrf-3, eri-1, eri-3, and eri-5 mutant strains to propagate at a high growth temperature (Simmer et al. 2002; Duchaine et al. 2006). Rather than producing temperature-sensitive mutant protein effects, RRF-3 and other ERI proteins are thought to act in a temperature-sensitive process, as evidenced by the predicted truncated and presumed nonfunctional protein fragments that would result from the available deletion alleles and by their shared temperature-sensitive phenotypes. rrf-3 mutant animals have been observed to exhibit X-chromosome missegregation (Simmer et al. 2002) and an unusual persistence of a chromatin mark on the X chromosome during male spermatogenesis (Maine et al. 2005). X-chromosome missegregation and defective spermatogenesis have been referred to in previous studies of eri-1 (Kennedy et al. 2004) and eri-3 and eri-5 (Duchaine et al. 2006). Furthermore, eri-3 mutant sterility can be rescued by insemination with wild-type sperm (Duchaine et al. 2006).Here we investigated the role of RRF-3 during spermatogenesis. We found defects evident at multiple stages, including after fertilization, where defects in rrf-3 mutant sperm can produce subsequent nonviable embryos. By using high-throughput sequencing, we characterized a large population of siRNAs present in spermatogenic cells and found a strong enrichment for antisense siRNAs from genes with known mRNA expression during spermatogenesis. While the majority of siRNA production during spermatogenesis does not require RRF-3, we found a set of genes for which siRNA production was dependent upon RRF-3. Existing data indicate increased expression for these genes in rrf-3 and/or eri-1 mutants. Taken together, our analyses suggest a working model in which the RRF-3/ERI pathway generates siRNAs that downregulate specific genes during spermatogenesis, with this regulation playing a key role in generating functional sperm.