Mut-16 and Other Mutator Class Genes Modulate 22G and 26G Sirna

mut-16 and other mutator class genes modulate 22G INAUGURAL ARTICLE and 26G siRNA pathways in Caenorhabditis elegans

Chi Zhanga,1, Taiowa A. Montgomerya,1, Harrison W. Gabela,1,2, Sylvia E. J. Fischera, Carolyn M. Phillipsa, Noah Fahlgrenb, Christopher M. Sullivanb, James C. Carringtonb, and Gary Ruvkuna,3 aDepartment of Molecular Biology, Massachusetts General Hospital, Department of Genetics, Harvard Medical School, Boston, MA 02114; and bCenter for Genome Research and Biocomputing, Oregon State University, Corvallis, OR 97331

This contribution is part of the special series of Inaugural Articles by members of the National Academy of Sciences elected in 2008.

Contributed by Gary Ruvkun, December 16, 2010 (sent for review October 20, 2010) Argonaute-associated siRNAs and Piwi-associated piRNAs have also required for CSR-1 class 22G siRNA formation (7). RRF-3 is overlapping roles in silencing mobile genetic elements in animals. required for ERGO-1 and ALG-3/4 class 26G siRNA biogenesis In Caenorhabditis elegans, mutator (mut) class genes mediate (10–13). 26G siRNAs are unique among endogenous siRNAs in siRNA-guided repression of transposons as well as exogenous their requirement for DICER and enhancer of RNAi (eri)class RNAi, but their roles in endogenous RNA silencing pathways are genes (10, 12, 13). ERGO-1 class 26G siRNAs are enriched in not well-understood. To characterize the endogenous small RNAs oocytes, embryos, and somatic cells, whereas ALG-3/4 class 26G dependent on mut class genes, small RNA populations from a null siRNAs are enriched in spermatogenic cells (10–13). In addition to allele of mut-16 as well as a regulatory mut-16(mg461) allele that endogenous siRNAs, microRNAs (miRNAs) and Piwi-interacting disables only somatic RNAi were subjected to deep sequencing. RNAs (piRNAs; also called 21U RNAs in C. elegans)alsohave Additionally, each of the mut class genes was tested for a require- broad roles in RNA silencing in C. elegans (14–20). ment in 26G siRNA pathways. The results indicate that mut-16 is A conserved function of endogenous RNAi pathways in eu- an essential factor in multiple endogenous germline and somatic karyotes is to silence transposons and other invasive nucleic acids siRNA pathways involving several distinct Argonautes and RNA- – fi (6, 21 27). mutator (mut) class genes were identi ed in screens for GENETICS dependent RNA polymerases. The results also reveal essential mutations that desilence transposons in the C. elegans germline roles for mut-2 and mut-7 in the ERGO-1 class 26G siRNA pathway (21, 28, 29). Mutations in any of the six characterized mut genes and less critical roles for mut-8, mut-14, and mut-15. We show that (mut-2/rde-3, mut-7, mut-8/rde-2, mut-14, mut-15/rde-5, and mut- – transposons are hypersusceptible to mut-16 dependent silencing 16) also cause an increased incidence of males because of chro- and identify a requirement for the siRNA machinery in piRNA mosomal nondisjunction, and they are temperature-sensitive ster- biogenesis from Tc1 transposons. We also show that the soma- fi ile (21, 29). The mut class genes are also required for exogenous speci c mut-16(mg461) mutant allele is present in multiple C. ele- RNAi and the accumulation of at least a subset of endogenous gans laboratory strains. siRNAs (9, 21, 29–33). mut-2, mut-7, and mut-14 are conserved in animals and encode a nucleotidyltransferase, a 3′–5′ exonuclease NAi can be elicited by the introduction of dsRNAs, which and an RNA helicase, respectively (21, 28, 30, 31). mut-16 encodes Rtrigger sequence-specific degradation of homologous mRNAs a protein with glutamine/asparagine (Q/N)-rich domains, sug- (1). Exogenous dsRNAs are processed into ∼22-nt siRNA gesting that it may mediate specific protein–protein interactions, duplexes by the RNase III enzyme DICER (2–4). One strand of whereas mut-15 and mut-8 encode proteins that lack known the siRNA duplex is loaded into a silencing complex (RISC) functional domains (29, 32–34). containing an Argonaute protein and accessory factors, where it To assess the role of mut class genes in endogenous RNAi- serves as a guide to silence complementary mRNAs (reviewed related pathways, small RNAs from WT C. elegans and a null mut- in ref. 5). 16 mutant strain were subjected to deep sequencing. We also High-throughput sequencing has revealed an extensive repertoire analyzed (by deep sequencing) the small RNA defects of a regu- of endogenous siRNAs in plants, animals, and fungi. These endog- latory mutation in mut-16 that decouples its functions in somatic enous siRNAs play roles in maintaining genome integrity at both and germline RNAi. Additionally, we tested the requirement of transcriptional and posttranscriptional levels by suppressing trans- each of the other mut class genes in the 26G siRNA pathways. poson mobilization, silencing aberrant transcripts, regulating gene Our results indicate that mut-16 and other mut class genes are expression, and promoting heterochromatin formation (reviewed in essential components of the endogenous RNAi machinery af- ref. 6). In the nematode Caenorhabditis elegans, endogenous siRNAs fecting multiple classes of endogenous siRNAs. can be broadly categorized according to their length and 5′ nucleotide. 22G siRNAs are predominantly 22 nt in length and contain a5′G that is triphosphorylated, whereas 26G siRNAs are pre- Author contributions: C.Z., T.A.M., H.W.G., S.E.J.F., C.M.P., and G.R. designed research; dominantly 26 nt long and have a 5′G that is monophosphorylated. C.Z., T.A.M., H.W.G., S.E.J.F., and C.M.P. performed research; N.F., C.M.S., and J.C.C. con- These distinct siRNA classes are bound to particular subtypes of tributed new reagents/analytic tools; C.Z., T.A.M., H.W.G., S.E.J.F., C.M.P., and G.R. analyzed data; and C.Z., and T.A.M., and G.R. wrote the paper. the 27 different C. elegans Argonaute proteins (WAGOs, CSR-1, fl ALG-3/4, and ERGO-1), which are largely responsible for confer- The authors declare no con ict of interest. ring unique functionality to the different classes of siRNAs. The Freely available online through the PNAS open access option. majority of endogenous siRNAs in C. elegans bypass DICER pro- Data deposition: The data reported in this paper have been deposited in the Gene Ex- pression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE26165). cessing and instead, derive from short RNA-dependent RNA 1C.Z., T.A.M., and H.W.G. contributed equally to this work. polymerase (RdRP) transcripts, which likely undergo additional 2Present address: Department of Neurobiology, Harvard Medical School, Boston, nuclease-mediated processing to facilitate association with one MA 02115. particular type of Argonaute protein (7–9). C. elegans contains four 3To whom correspondence should be addressed. E-mail: [email protected]. RdRPs (ego-1, rrf-1, rrf-2,andrrf-3) with overlapping roles in the edu. various siRNA pathways. EGO-1 and RRF-1 have partially re- This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. dundant roles in WAGO class 22G siRNA formation (9). EGO-1 is 1073/pnas.1018695108/-/DCSupplemental. www.pnas.org/cgi/doi/10.1073/pnas.1018695108 PNAS | January 25, 2011 | vol. 108 | no. 4 | 1201–1208 Results The presence of the mut-16(mg461) allele in strains originating RNAi-Deficient mut-16(mg461) Allele Is Present in Many C. elegans from different laboratories suggested that it could be a common Laboratory Strains. To characterize the genetic factors required mutation in C. elegans laboratory stocks. To assess the prevalence for RNAi in C. elegans, we obtained mutant alleles for a subset of of this mutation in laboratory strains and to determine its origin, genes that were identified in RNAi screens for components we genotyped more than 100 arbitrarily selected C. elegans strains (Datasets S1 and S2). In total, 18 strains from six laboratories of RNAi pathways (33). We identified RNAi defects in several of were found to contain the mut-16(mg461) deletion (Table 1). In ncl-1 e1865 inx-22 these strains, including those containing ( ), addition, the WT N2 strains used by at least two laboratories carry (tm1661), and inx-22(tm1661);fog-2(q71). However, after back- this deletion. The dimorphism of presumed WT strains was likely crossing the ncl-1 and inx-22 mutants to WT, the RNAi defects caused by a deletion that yielded the mut-16(mg461) allele as failed to cosegregate with the ncl-1 and inx-22 mutant alleles, opposed to an insertion in the canonical Bristol N2 strain, be- suggesting that a background mutation was responsible for the cause the closely related C. elegans isolate Hawaiian CB4856 does RNAi-deficient phenotype. From a combination of genetic map- not have the deletion (Dataset S2). ping and candidate gene sequencing, the causal mutation in each of these three strains was identified as a regulatory mutation in mut-16(mg461) Is a Soma-Specific RNAi-Defective Allele of mut-16. mut-16 (Methods). The mut-16 mutant allele (mg461) contains Presumptive null alleles of mut-16 are defective for both so- a 451-bp deletion located about 500 bp upstream of the predicted matic and germline RNAi (29). mut-16 is also required for an endogenous RNAi-related pathway that directs transposon si- start codon (Fig. 1A). Although this mutation is common to un- lencing in the germline (29). To determine if the mut-16(mg461) related strains, the DNA sequence around the deletion does not allele causes RNAi defects similar to that of null alleles, genes contain any obvious repetitive or transposon sequences that would expressed in particular tissues were targeted by dsRNAs in WT, suggest that the frequency of spontaneous mutation for this se- a probable null allele of mut-16(pk710) containing an early stop quence is unusually high. codon, and the mut-16(mg461) allele. WT animals were sensitive to and mut-16(pk710) mutant animals were resistant to RNAi targeting each of the somatic and germline genes tested (Fig. 1B). A mut-16 In contrast, although the mut-16(mg461) allele displayed sensi- Δ tivity to RNAi targeting germline genes similar to that of WT, the pk710 efficiency of RNAi targeting somatic genes was strongly reduced mg461 W189Amber 1 kb (Fig. 1B). When introduced into mut-16(pk710), an 8,422-bp B RNAi Susceptibility in Various Tissues DNA amplicon containing the mut-16(mg461) locus that includes Intestine Muscle Seam, Vulva, Hypodermis Germline the 451-bp deletion in the upstream regulatory region rescued the flr-1 elt-2 unc-22 lin-29 dpy-11 nhr-23 pos-1 mom-2 germline RNAi defects of mut-16(pk710) but failed to rescue the ++ ++ +++ +++ ++ +++ +++ ++ wt (N2) somatic RNAi defects, whereas an analogous 8,873-bp amplicon mut-16(pk710) - - + - - - - - mut-16(mg461) + - ++ - + - +++ ++ containing the WT mut-16(+) allele rescued defects in both the soma and germline (SI Appendix, Table S1). Defects in endogenous siRNA pathways have been linked to 300 20˚C Percent Male Progeny C D mut-16 250 26˚C temperature-sensitive sterility, increased incidence of male prog- wt (N2) pk710 mg461 eny, and increased transposition in the germline (9, 21). mut-16 200 20˚C 0.12 2.1 0.05 150 26˚C 0.42 5.9 0.82 (pk710) mutants were temperature-sensitive sterile at 26 °C and

Brood Size yielded an ∼17.5- and ∼14-fold increase in male progeny at 20 °C 100 unc-22::Tc1 Reversion Frequency and 26 °C, respectively, relative to WT (Fig. 1 C and D) (29). In 50 E wt (N2) None detected -4 -5 contrast, mut-16(mg461) mutant animals had a brood size in- 0 mut-16(pk710) 3X10 - 5X10 wt (N2) pk710 mg461 mut-16(mg461) None detected mut-16 Table 1. C. elegans strains identiﬁed as containing the mut-16 1.4 (mg461) allele F Small RNA Northern Blot Assay G N2 1.2 mut-16 mg461 Strain Allele 1.0 wt (N2) eri-1 pk710 mg461 mut-7 0.8 ~45% CB3388 ncl-1(e1865) III X -cluster mRNA to wt (N2) 0.6 XM1011 inx-22(tm1661)I 0.4

T01A4.3 mut-16 0.2 ~75% XM1012 inx-22(tm1661)I;fog-2(q71)V 0.0 JK1277 lag-2(q420)V 5S rRNAs glp-4 glp-4 Ratio of JK816 fem-3(q20)IV 1 2345 +/+ -/- TR1151 unc-22(r743::Tc3)IV; mut-2(r459)I Fig. 1. The mut-16(mg461) allele specifically affects the somatic activity of TR1160 unc-22(r750::Tc3) IV; mut-2(r459) I mut-16.(A) A diagram of the gene structure of mut-16 displaying the mg461 YY011 dcr-1(mg375Eri) III ′ deletion. Black boxes, exons; gray box, 3 UTR. The mut-16(mg461) allele TR1326 smg-2(r863) unc-54(r293)I ∼ carries a 451-bp deletion 500 bp upstream of the predicted start codon. The TR1332 smg-2(r863)I presumptive null allele mut-16(pk710) has an early stop codon. (B) mut-16 TR1696 unc-54(r293)I;smg-3(r930)IV (pk710) and mut-16(mg461) sensitivity to dsRNA clones targeting germline- and soma-specific genes. +, sensitive; −, resistant. (C) Brood size from WT SM196 smg-3(r867)IV and mut-16 mutant strains (pk710 and mg461) grown at 20 °C or 26 °C (n = TR135 smg-5(r860)I 15). (D) Incidence of male progeny from WT and mut-16 mutant strains TR1324 smg-5(r860) unc-54(r293)I grown at 20 °C or 26 °C. (E) Frequency of reversion of unc-22::Tc1 in WT and TR1335 smg-5(r860)I mut-16 mutant backgrounds. (F) Northern blot assays of X-cluster and eri-6 TR1336 smg-6(r896) III (T01A4.3) siRNAs. 5S rRNAs stained with ethidium bromide are shown as TR2264 unc-54(r293)I;smg-6(r1217) III a loading control. (G) qRT-PCR of mut-16 mRNA levels in N2 and mut-16 TR2230 unc-54(r293)I;smg-7(r1197)IV (mg461) in either WT or glp-4(bn2) germline-deficient backgrounds (WT = 1.0). Error bars are SDs calculated from three technical replicates. Determined by PCR-based genotyping.

1202 | www.pnas.org/cgi/doi/10.1073/pnas.1018695108 Zhang et al. distinguishable from WT at both 20 °C and 26 °C and similar in- somatic X-cluster siRNAs (Fig. 1F), suggesting that the mg461 cidence of male progeny to WT (Fig. 1 C and D). The mut-16 allele specifically affects somatic siRNA pathways. INAUGURAL ARTICLE (pk710) temperature-sensitive sterility phenotype was only par- To determine if the mg461 allele causes a differential reduction tially rescued by male mating (∼20%; relative to WT) and thus, is in mut-16 expression in the soma relative to the germline, we likely caused by defects in both sperm and oocyte small RNA measured the relative levels of mut-16 mRNA in mut-16(mg461)in pathways. Finally, unlike the mut-16(pk710) null allele, which an otherwise WT background or a temperature-sensitive germline- causes reversion of an unc-22 mutant allele containing a Tc1 deficient glp-4(bn2) background by quantitative RT-PCR (qRT- transposon to WT caused by germline transposition of the Tc1 PCR) (Fig. 1G). mut-16 mRNA levels were reduced by ∼45% in sequence out of the unc-22 gene, no revertants were observed in mut-16(mg461) compared with WT. However, mut-16 mRNA either WT or mut-16(mg461) mutants (Fig. 1E). levels in mut-16(mg461) in the glp-4(bn2) germline-deficient To determine if the mut-16(mg461) hypomorphic allele causes background were reduced by ∼75% relative to control glp-4 mutant defects in endogenous siRNA pathways, Northern blot assays animals (Fig. 1G). Thus, the reduction in mut-16 mRNA levels in were done for siRNAs derived from the somatically expressed X- mut-16(mg461) was ∼1.7-fold greater in germline-deficient glp-4 cluster region and the germline-expressed gene eri-6 (T01A4.3). (bn2) than in WT, suggesting that mut-16(mg461) primarily affects The mut-16(pk710) null allele, as well as eri-1 and mut-7 mutants, somatic expression. These data suggest that the mut-16(mg461) was tested in parallel. mut-16(pk710) and mut-7 mutants had allele results in soma-specific defects in RNAi-related pathways, decreased levels of siRNAs from both loci tested, whereas eri-1 while having little or no effect on the germline function of mut-16. only affected the somatically expressed X-cluster siRNAs (Fig. 1F) (16, 35). Similar to eri-1 mutants, mut-16(mg461) had WT mut-16 Is Required for the Accumulation of WAGO Class 22G siRNAs. levels of germline eri-6 (T01A4.3) siRNAs and reduced levels of To more comprehensively characterize the role of mut-16 in

A wt (N2) mut-16(pk710) mut-16(mg461) D Germline and Soma E WAGO-Class F NRDE-3 Associated 0.5 5'G miRNAs Enriched siRNAs siRNAs siRNAs 0.4 5'C piRNAs 1.0 pk710 1.0 1.0 Transposons mg461

5'U GENETICS 0.3 5'A Coding genes 0.8 0.8 0.8 Pseudogenes 0.2 tRNAs/rRNAs 0.6 0.6 0.6 snRNAs/snoRNAs 0.1 0.4 0.4 Intergenic/other 0.4 Proportion of Reads 0.0 0.2 0.2 18 20 22 24 26 28 18 20 22 24 26 28 18 20 22 24 26 28 0.2

Ratio of Reads to wt (N2) 0.0 Ratio of Reads to wt (N2) Size (nt) Ratio of Reads to wt (N2) 0.0 0.0 20 Size (nt) B Enriched in mut-16(pk710) 19 20 21 22 Soma Soma pk710 pk710 mg461 Germline Germline mg461 pk710 10 23 24 25 26 RDE-1 Target Genes G Y47H10A.5 pk710 0 mg461 W06H8.8 Change in

-10

Fold 0.0 0.5 1.0 1.5 2.0 2.5 2 Depleted in mut-16(pk710) Ratio of siRNA Reads to wt (N2) Log -20 I II III IV VX Chromosome H CSR-1-Class siRNAs I CSR-1-Associated Target Genes 1.0 klp-16 pk710 20 Size (nt) C 0.8 mg461 Enriched in mut-16(mg461) 19 20 21 22 bub-1

mg461 10 23 24 25 26 0.6 cdc-48.1 cgh-1 0.4 0 hcp-1 Change in 0.2 -10 hcp-3 Fold Ratio of Reads to wt (N2)

2 0.0 Depleted in mut-16(mg461) 0.0 0.5 1.0 1.5

Log -20 I II III IV VX Ratio of siRNA Reads to wt (N2) pk710 Chromosome mg461 CSR-1 Target Genes Non-CSR-1 Target Genes J 1 wt K 1 wt 40 wt 1 wt 0 0 0 0

4 6 80 5 mut-16(pk710) 1 mut-16(pk710) 1 1 mut-16(pk710) 40 mut-16(pk710) 0 0 0 0 Reads Per Million Reads Per Million Reads Per Million Reads Per Million 5 4 6 80 1 2500 5000 1 1500 3000 1 3000 6000 1 6000 12000 Position on hcp-1 Position on bub-1 Position on B0001.6 Position on eri-6/T01A4.3

Fig. 2. mut-16 is required for WAGO class 22G siRNA accumulation. (A) Small RNA size and 5′ nucleotide distribution in WT and mut-16 mutants (pk710 and mg461). (Insets) Pie charts display the proportion of reads aligning to each genomic feature. (B and C) Log2 ratio of small RNA reads in mut-16 mutants to WT plotted across each chromosome (5-kb bins and 1-kb scroll). (D) Ratio of germline- and soma-enriched siRNA reads in mut-16 mutants to WT (WT = 1.0). (E) Ratio of WAGO class siRNA reads in mut-16 mutants to WT (WT = 1.0). (F) Ratio of NRDE-3–associated siRNA reads in mut-16 mutants to WT (WT = 1.0). (G) Ratio of siRNAs derived from individual RDE-1 target genes in mut-16 mutants to WT (WT = 1.0). (H) Ratio of CSR-1 class siRNA reads in mut-16 mutants to WT (WT = 1.0). (I) Ratio of siRNA reads from individual CSR-1 targets in mut-16 mutants to WT (WT = 1.0). (J) siRNA distribution across two representative CSR-1 targets. Position is relative to annotated transcription start site. Sense siRNAs are in orange; antisense siRNAs are in blue. (K) siRNA distribution across two representative WAGO class siRNA targets that are not also targeted by CSR-1. Position is relative to annotated transcription start site. Sense siRNAs are in orange; antisense siRNAs are in blue.

Zhang et al. PNAS | January 25, 2011 | vol. 108 | no. 4 | 1203 endogenous siRNA pathways, small RNA cDNA libraries were The fact that CSR-1 class siRNAs were reduced in mut-16 constructed from adult WT, mut-16(pk710), and mut-16(mg461) (pk710) was unexpected given that mut-16 mutants do not have strains and sequenced on an Illumina Genome Analyzer plat- obvious chromosome segregation defects, unlike CSR-1 pathway form (SI Appendix, Fig. S1A). Small RNAs were captured using mutants, aside from modest nondisjunction at the X chromosome a method that allows for either a 5′ monophosphate (e.g., miR- (7, 8, 29). It is possible that mRNAs targeted by CSR-1 to direct NAs, piRNAs, and 26G siRNAs) or a 5′ triphosphate (e.g., 22G chromosome segregation are also routed into a distinct RNA si- siRNAs) (9). Small RNAs containing a 5′G that were 22 nt in lencing pathway involving the WAGOs. Indeed, of 937 genes length (22G siRNAs) were the most abundant class in WT ani- depleted by twofold or greater of siRNAs in csr-1 mutants (7), 383 mals (Fig. 2A), consistent with previous 5′ monophosphate- were also depleted of siRNAs in a C. elegans strain mutant for all independent sequencing results (9). These small RNAs were most 12 of the wago class genes (9). We removed these 383 genes from commonly derived from coding genes, pseudogenes, transposons, our analysis as well as genes that were not enriched for siRNAs and unannotated regions of the genome (Fig. 2A). bound by CSR-1 as determined by coimmunoprecipitation assays In mut-16(pk710) mutants, 22G siRNAs were largely depleted, (7). siRNAs from the remaining 392 CSR-1 target genes were whereas miRNAs and 21U RNAs (i.e., piRNAs) seemed to be depleted by only ∼60% in mut-16(pk710)(SI Appendix, Fig. S2). enriched because of, at least in part, depletion of the abundant A subset of predicted CSR-1 target genes (daf-21, vig-1, klp-16, 22G siRNAs (Fig. 2A). In contrast, mut-16(mg461) mutants bub-1, cdc-48.1, cgh-1, hcp-1, and hcp-3) was shown to bind to showed only a slight reduction in total 22G siRNA levels (Fig. CSR-1 and thus, represents a high-confidence set of CSR-1 tar- 2A). Five thousand thirteen annotated coding genes, representing gets (7). For the six of these genes that we obtained greater than ∼25% of all unique coding genes in C. elegans, yielded ≥10 siRNA ∼35 reads per million total small RNA reads in our WT library, reads per million total small RNA reads (RPM) in WT. Of these, siRNA levels were only moderately affected or unaffected in siRNAs from 2,803 were reduced by ≥67% in mut-16(pk710) mut-16(pk710) (Fig. 2I). A closer inspection of these six high- (Dataset S3). Additionally, 198 annotated pseudogenes and 254 confidence CSR-1 target genes revealed that, although most in- annotated transposons yielding ≥10 RPM in WT animals dis- dividual siRNAs were reduced, some siRNAs were unaffected or played a ≥67% reduction in siRNA levels in mut-16(pk710) in certain cases, particularly near the 3′ ends of genes, elevated in (Dataset S4). In contrast, only 177 features yielding ≥10 RPM in mut-16(pk710) (Fig. 2J and SI Appendix, Fig. S3A). In contrast, WT animals had siRNAs levels reduced by ≥67% in mut-16 genes targeted by the WAGO pathway and that do not yield (mg461) relative to WT (Dataset S5). 22G siRNAs were depleted siRNAs shown to associate with CSR-1 were uniformly depleted across each of the six chromosomes in mut-16(pk710) mutants of siRNAs (Fig. 2K and SI Appendix, Fig. S3B). Based on these (Fig. 2B), whereas mut-16(mg461) mutants displayed only mod- results, we conclude that mut-16–dependent siRNAs derived est, although widespread, reductions in 22G siRNAs (Fig. 2C). from CSR-1 targets are distinct from those required for CSR-1– Small RNA reads derived from genes that yield 22G siRNAs mediated chromosome segregation and that mut-16 is not re- shown to be germline-enriched (9) were reduced by ∼85% in mut- quired for CSR-1 class 22G siRNA formation or function. 16(pk710) but only by ∼15% in mut-16(mg461) mutants (Fig. 2D). Small RNA reads derived from genes yielding siRNAs shown to Transposons Are Hypersusceptible to mut-16–Dependent Silencing. be soma-enriched (9) were depleted by >99% in mut-16(pk710) Transposon-derived siRNAs suppress transposase activity by an and ∼80% in mut-16(mg461) (Fig. 2D). This represents a greater RNAi-related mechanism in the C. elegans germline (9, 21, 22, than fivefold reduction in soma-enriched siRNAs relative to 39). To determine if transposons are hypersusceptible to siRNA- germline-enriched siRNAs in mut-16(mg461) but a similar re- mediated silencing involving mut-16, the numbers of small RNA duction in soma and germline siRNAs in mut-16(pk710), sup- reads per annotated coding gene, pseudogene, and transposon porting the conclusion that the mg461 regulatory mutation were calculated. In WT C. elegans, the median number of reads primarily affects mut-16 activity in the soma. from transposons was ∼18-fold greater than from coding genes 22G siRNAs fall into one of two classes determined by their and ∼108-fold greater than from pseudogenes (Fig. 3A). Fur- Argonaute binding partners, either CSR-1 or the WAGO family thermore, ∼62% of 408 annotated transposons in C. elegans yiel- (7–9). siRNAs derived from WAGO class 22G siRNA target ded ≥10 siRNA reads per million total small RNA reads, whereas genes were reduced by >99% in mut-16(pk710) mutants but only only ∼25% of 20,163 annotated coding genes and ∼14% of 1,524 slightly reduced in mut-16(mg461) mutants relative to WT (Fig. annotated pseudogenes yielded ≥10 siRNA reads per million total 2E). A subset of WAGO class 22G siRNAs interacts with the so- small RNA reads in our WT small RNA library after correcting for matically expressed Argonaute NRDE-3 to direct cotranscrip- sequences with multiple genomic loci. When small RNA reads per tional gene silencing (36, 37). In mut-16(pk710) and mut-16 individual feature were plotted for mut-16(pk710) (y axis) vs. WT (mg461), NRDE-3–interacting siRNAs were reduced by >99% (x axis), the vast majority of coding genes, pseudogenes, and and ∼60%, respectively, relative to WT (Fig. 2F). The Argonaute transposons clustered to the right of the y = x line, indicating de- RDE-1 associates with primary siRNAs derived from exogenously pletion in mut-16(pk710), with transposons clustering farthest delivered RNAs and affects only a small subset of endogenous from the line (Fig. 3B). In contrast, individual 21U RNAs and siRNAs, most notably soma-enriched WAGO class 22G siRNAs miRNAs tended to be slightly elevated in mut-16(pk710) (i.e., to triggered by miR-243–guided cleavage of Y47H10A.5 (38). siR- the left of the y = x line), likely because these classes of small RNAs NAs from Y47H10A.5 were reduced by >99% and ∼90% in mut- are enriched because of depletion of siRNAs (Fig. 3B). 16(pk710) and mut-16(mg461), respectively (Fig. 2G). A second Tc1 and Tc3 are the most abundant active DNA transposons in RDE-1 target, W06H8.8, yields siRNAs that are not somatically C. elegans. siRNA reads from Tc1 and Tc3 were strongly reduced enriched, and it displayed siRNA levels that were reduced by in mut-16(pk710) but unchanged in mut-16(mg461) (Fig. 3 C and ∼90% in mut-16(pk710) and somewhat elevated in mut-16(mg461) D). siRNA reads from the gypsy-like retrotransposon gene retr-1 (Fig. 2G). These results indicate that mut-16 is essential for the were also reduced in mut-16(pk710) mutants and to a lesser de- formation or stability of WAGO class 22G siRNAs derived from gree, in mut-16(mg461) mutants (Fig. 3E) (40). Two 21U RNAs both germline and somatically expressed genes, including siRNAs (i.e., piRNAs) derive from the transposase region of Tc3 (18, 19). that associate with NRDE-3 to direct cotranscriptional silencing. We mapped a 21U RNA (21UR-6046) to the terminal inverted Small RNAs derived from CSR-1 class 22G siRNA target repeat (TIR) sequences of each of the 31 Tc1 transposons and id- genes (i.e., genes depleted of siRNAs by twofold or greater in entified a 21U RNA (SI Appendix, SI Text) that maps to the long csr-1 mutants) (7) were reduced by nearly 90% in mut-16(pk710) terminal repeat (LTER) regions of retr-1 (Fig. 3 C and E) (18). and ∼15% in mut-16(mg461) mutants relative to WT (Fig. 2H). Surprisingly, 21U RNA reads from Tc1 were depleted by ∼78% in

1204 | www.pnas.org/cgi/doi/10.1073/pnas.1018695108 Zhang et al. Log siRNA Reads Per Million Per Feature Log Small RNA Reads Per Million Per Feature in mut-16(pk710) and Wild Type A 2 B 2

-20 -10 0 10 20 INAUGURAL ARTICLE 15 y = x 20 pk710

wt (N2) pk710 Coding Genes pk710 mg461 Pseudogenes = mean -15 15 -20 20 = median Coding Genes 21U RNAs y = x siRNA Reads in Pseudogenes miRNAs 2

Transposons Small RNA Reads in

Transposons 2 Log -15 -20 Log siRNA Reads in wt Log Small RNA Reads in wt

2 Log 2

C Tc1 Small RNA Distribution D Tc3 Small RNA Distribution E gypsy-Like Transposon Small RNA Distribution wt (N2) wt (N2) 20 wt (N2) 40 S 20 S S 0 0 0 40 21U RNA AS 20 21U RNA AS AS 22G siRNA/other 22G siRNA/other 20 80 40 20 mut-16(pk710) 40 mut-16(pk710) S 20 mut-16(pk710) S S 0 0 0 40 AS AS AS 20 20

80 40 Reads Per Million 20 mut-16(mg461)

Reads Per Million Reads Per Million S 40 mut-16(mg461) 20 mut-16(mg461) S S 0 0 0 putative 21U RNA AS 40 AS 20 AS 20 22G siRNA/other Tandem 80 TIR Tc1 Transposase TIR 40 TIR Tc3 Transposase TIR LTER retr-1 (reverse transcriptase-like) Repeats

10529.8K 10529.0K 10528.2K 601K 600K 599K 8852K 8856K 8860K 8864K Position on Chromosome I Position on Chromosome I Position on Chromosome III

Fig. 3. Transposons are hypersensitive to mut-16–dependent silencing. (A) Box plot displaying distribution of siRNA reads per coding gene, pseudogene, and GENETICS transposon after log2 transformation in WT and mut-16 mutants (pk710 and mg461). (B) Scatter plots display log2 small RNA reads per million total reads for each annotated coding gene, pseudogene, and transposon (Left) and miRNA and 21U RNA (Right). (C and D) Small RNA distribution across representative Tc1 (F59C6.1) and Tc3 (F56A6.3) transposon loci in WT and mut-16 mutants. TIR, terminal inverted repeats; S, sense strand small RNAs; AS, antisense strand siRNAs. (E) Small RNA distribution across the gypsy-like retrotransposon locus in WT and mut-16 mutants. LTER, long terminal repeats; S, sense strand small RNAs; AS, antisense strand small RNAs. mut-16(pk710) relative to WT (Fig. 3C). Tc1-derived 21U RNA 4A). ERGO-1 class 26G siRNAs are enriched in embryos (10). reads were also depleted in small RNA deep-sequencing datasets Thus, to provide greater ERGO-1 class 26G siRNA sequencing from mut-2 and mut-7 mutants relative to a WT control dataset depth and thereby allow for a more comprehensive assessment of (9), suggesting that 21U RNA biogenesis from Tc1 elements is mut-16 dependence, we prepared additional small RNA cDNA dependent, possibly indirectly, on the siRNA machinery. In libraries from WT and mut-16(pk710) embryos (SI Appendix, Fig. contrast, 21U RNA reads from Tc3 and retr-1 were elevated in S1B). Consistent with the results obtained from adult libraries, mut-16(pk710) (Fig. 3 D and E). Tc1 and Tc3 transposase mRNA ERGO-1 class 26G siRNAs derived from 57 annotated coding levels were elevated by approximately three- and sixfold, re- genes (10, 13) were reduced by ∼88% in the mut-16(pk710) spectively, in mut-16(pk710) but were unaffected in mut-16 embryo library relative to the WT embryo library (SI Appendix, (mg461) relative to WT (SI Appendix, Fig. S4). These results Fig. S5). Of 57 ERGO-1 target genes, 54 were depleted of 26G suggest that certain features trigger entry of transposons into the siRNA reads in mut-16(pk710)(Dataset S6). mut-16–dependent RNA silencing pathway more efficiently than To test whether or not other mut class genes are required for other classes of genes. However, the fact that many coding genes 26G siRNA accumulation, Taqman qRT-PCR was done to and pseudogenes also yield relatively high levels of siRNAs measure the levels of individual 26G siRNAs in mutants for each indicates that additional nontransposon-specific features confer of the six mut class genes (10, 41). As a control, eri-1, a factor hypersusceptibility to RNA silencing pathways. required for both ERGO-1 and ALG-3/4 class 26G siRNA biogenesis, was tested in parallel (10–13). In mut-2, mut-8, mut-14, mut-16 and Other mut Class Genes Are Required for the Accumulation mut-15, and mut-16(mg461) mutants, the levels of two 26G siR- of ERGO-1 Class 26G siRNAs. 26G siRNAs are primary siRNAs that NAs, S4 and S5, derived from desp-1 and ssp-16 genes, re- trigger secondary 22G siRNA formation from their targets (10– spectively, were indistinguishable from WT (Fig. 4B). A moderate 13). 26G siRNAs can be distinguished by their association with decrease in S4 and S5 26G siRNA levels was observed in mut-7 either ERGO-1 or ALG-3 and ALG-4 Argonautes. To deter- mutants, whereas in mut-16(pk710) mutants, both siRNAs were mine if mut-16 is required for the accumulation of ALG-3/4 and moderately increased relative to WT (Fig. 4B). Consistent with ERGO-1 class 26G siRNAs, small RNA reads corresponding to previous studies, S4 and S5 26G siRNA levels were strongly re- previously identified 26G targeted mRNAs were extracted from duced in eri-1 (Fig. 4B) (10). our deep-sequencing datasets (10, 11, 13). ALG-3/4 class 26G In contrast to ALG-3/4 class 26G siRNAs, the levels of two siRNAs were moderately increased in mut-16(pk710) relative to ERGO-1 class 26G siRNAs, O1 and O2, derived from C40A11.10 WT (Fig. 4A). In contrast, 22G siRNA reads derived from ALG- and E01G4.7 loci, respectively, were strongly reduced (∼16- to 3/4 target genes were depleted by ∼70% in mut-16(pk710), sug- 100-fold) in mut-2, mut-7, mut-15, and mut-16 and moderately gesting that, although ALG-3/4 class 26G primary siRNAs are reduced (approximately two- to fourfold) in mut-8 and mut-14 mut-16–independent, the secondary 22G siRNAs are at least mutants relative to WT (Fig. 4C). In mut-16(mg461), O1 and O2 partially mut-16–dependent (Fig. 4A). 26G siRNA levels were indistinguishable from WT (Fig. 4C). ERGO-1 class 26G siRNAs and secondary 22G siRNAs were ERGO-1 class 26G mRNA targets were strongly up-regulated in both strongly depleted in mut-16(pk710) relative to WT (Fig. mut-2, mut-7,andmut-16 (∼4- to 13-fold) as well as in eri-1 and

Zhang et al. PNAS | January 25, 2011 | vol. 108 | no. 4 | 1205 A C D 26G-Target siRNA Levels in ERGO-1-Class 26G siRNAs (qRT-PCR) ERGO-1-Class 26G-Target mRNAs 25 mut-16(pk710) (smRNA-seq) F39E9.7 Log2 Ratio of Reads to wt (N2) 1.6 -8 -7 -6 -5 -4 -3 -2 -1 0 F55C9.5 ALG-3/4-Class 20 E01G4.7 ERGO-1-Class 1.2 26G O1 wt K02E2.6 26G O2 mut-2 15 E01G4.5 0.8 mut-7 10 0.4 mut-8 mut-14 Ratio of Reads to wt (N2) 0.0 5 22 26 18-28

mut-15 Normalized Expression Ratio Size (nt) pk710 0 mut-16 wt mut-2 mut-7 mut-8 mut-14 mut-15 pk710 mg461 eri-1 rde-4 mg461 mut-16 B E F ALG-3/4-Class 26G siRNAs (qRT-PCR) RRF-3-Target siRNAs (smRNA-seq) RRF-3-Target mRNAs 3.0 2.0 16 26G S4 pk710 E01G4.5 2.5 26G S5 mg461 14 W04B5.1 1.5 2.0 12 H09G03.1 10 C44B11.6 1.5 1.0 8 1.0 0.5 6 0.5 4 0.0 Ratio of Reads to wt (N2) 0.0

Normalized Expression Ratio 2 wt

Normalized Expression Ratio 0 eri-1 mut-2 mut-8 mut-7 pk710 wt pk710 mg461 mg461 mut-14 mut-15 F14F7.5 F39E9.7 F55A4.4 ZK380.5 T08B6.2 F52D2.6 K02E2.6 K06B9.6 F07G6.6 C18D4.4 E01G4.5 W04B5.2 W04B5.1 C36A4.11 C36A4.11 C44B11.6 Y43F8B.9 H16D19.4 Y17D7B.4 H09G03.1 Y17D7C.1 W05H12.2 mut-16 Y37E11B.2 mut-16

Fig. 4. Role of mut class genes in the 26G siRNA pathways. (A) Ratio of siRNA reads from ALG-3/4 and ERGO-1 class 26G siRNA target genes in mut-16(pk710) to WT (WT = 1.0). (B) Ratio of ALG-3/4 class 26G siRNAs S4 and S5 in mut class mutant strains as determined by Taqman qRT-PCR (WT = 1.0). (C) Ratio of ERGO-1 class 26G siRNAs O1 and O2 in mut class mutant strains to WT after log2 transformation as determined by Taqman qRT-PCR (WT = 0; i.e., log2 1.0]). (D) Ratio of ERGO-1 class 26G siRNA target mRNA levels in mut class mutant strains to WT as determined by qRT-PCR (WT = 1.0). (E) Ratio of siRNA reads from a representative subset of RRF-3 somatic target genes in mut-16 mutants to WT (WT = 1.0). (F) Ratio of RRF-3 target mRNA levels in mut-16 mutants to WT (WT = 1.0) as determined by qRT-PCR. All error bars are SDs calculated from three technical replicates. rde-4, two genes required for 26G siRNA accumulation (Fig. 4D) possible that siRNA levels were not sufficiently reduced in mut-16 (10, 11, 13). mut-8, mut-14,andmut-15 had only slightly to mod- (mg461) to effectively inhibit target mRNA silencing. Another erately elevated levels of 26G siRNA target mRNAs or affected possibility is that silencing of RRF-3 target mRNAs is initiated by only a subset of the targets analyzed (Fig. 4D). E01G4.5-derived germline-derived siRNAs, which are presumably unaffected in siRNAs are strong NRDE-3 interactors and are required to silence mut-16(mg461) and maintained by somatically produced siRNAs both the precursor mRNA (pre-mRNA) and mature mRNA of such that the relatively low levels of siRNAs observed in mut-16 E01G4.5 (36). E01G4.5 mRNA levels were elevated by ∼8- to 12- (mg461) mutants are sufficient to maintain mRNA silencing. fold in mut-16(pk710) and a slightly lesser degree in mut-2 and mut- These results indicate that mut-16, mut-2, and mut-7 are es- 7 (Fig. 4D). Together with the deep-sequencing results described sential factors in the ERGO-1 class 26G siRNA pathway, whereas above (Fig. 2F), our data suggest that mut-16, mut-2,andmut-7 are mut-8, mut-14, and mut-15 have only minor or redundant roles. required for the formation or stability of NRDE-3–interacting We conclude that mut-16 and other mut class genes regulate siRNAs. This is consistent with the requirement for mut-2 and mut- multiple endogenous siRNA pathways mediating RNA silencing 7 in siRNA-mediated NRDE-3 nuclear localization (36). but are largely dispensable for the CSR-1 siRNA pathway in- RRF-3 is an RNA-dependent RNA polymerase that is re- volved in chromosome segregation (SI Appendix, Fig. S6). quired for ERGO-1 class 26G siRNA biogenesis and somatic si- RNAs derived from ERGO-1 targets (10, 12, 13). If mg461 is a Discussion soma-specific allele of mut-16, we predicted that mut-16(mg461) We identified a soma-specific RNAi-defective allele of mut-16, mutants would be deficient for somatic RRF-3–dependent siRNAs. mg461, present in the background of multiple C. elegans laboratory To test this, we compared the levels of siRNAs that mapped to strains. The prevalence of this allele in the presumed WT and a subset of somatic RRF-3 targets in our WT and mut-16 mutant various mutant strains from the initial American immigrant wave libraries (12). siRNA reads mapping to each of the RRF-3 so- of C. elegans laboratories from the Brenner lab in England suggests matic target genes were strongly depleted (∼88% to >99%) in that the allele arose early in C. elegans research and subsequently, mut-16(pk710) (Fig. 4E). siRNA reads mapping to 15 of 22 genes propagated among laboratory stocks through a founder effect. analyzed (∼70%) were reduced by >50% in mut-16(mg461), Strains that contain the mut-16(mg461) allele have been used consistent with our earlier conclusion that the mg461 allele is in studies characterizing both exogenous and endogenous RNAi defective in somatic siRNA pathways. Interestingly, 22G but not pathways. Strains bearing mutations in smg-2, smg-5, and smg-6, 26G siRNAs derived from RRF-3 targets were reduced in mut-16 genes that encode proteins involved in the nonsense-mediated (mg461), whereas both classes were reduced in mut-16(pk710). To decay (NMD) pathway, were shown to affect the persistence of determine if the reduction in RRF-3 target gene siRNA levels in RNAi in C. elegans, and strains bearing smg-2 and smg-5 have also mut-16 mutants resulted in increased levels of the corresponding been shown to affect endogenous siRNA pathways (9, 42). The mRNA, qRT-PCR was done for 4 of 22 RRF-3 targets. Each mut-16(mg461) mutation is present in each of the smg-2, smg-5, target mRNA was up-regulated by approximately fivefold in mut- and smg-6 strains used in these studies (Table 1), which may 16(pk710) but was unaffected in mut-16(mg461) (Fig. 4F). It is provide an alternative explanation for why RNAi pathways are

1206 | www.pnas.org/cgi/doi/10.1073/pnas.1018695108 Zhang et al. disrupted in these NMD mutants. The mut-16(mg461) allele was siRNAs, which regulate a subset of spermatogenesis-enriched also found in strains bearing the dcr-1 and fem-3 mutant alleles mRNAs (10, 11, 44). In contrast, the temperature-sensitive ste- INAUGURAL ARTICLE mg375Eri and q20, respectively (Table 1), which affect the accu- rility phenotype observed in mut class mutants was only partially mulation of a subset of endogenous siRNAs (12, 43). The pres- rescued by male mating, suggesting that it is caused by broader ence of the mut-16(mg461) mutation in these strains may con- defects in both sperm and oocytes. It will be important to decipher found some of the results of those studies. Taken together, our the individual contributions of 22G and 26G siRNA pathways to data suggest that the RNAi-defective mut-16(mg461) allele is sperm and oocyte viability. a common background mutation in laboratory strains of C. ele- Our study showed that, in addition to the requirement for mut- gans. We suggest that strains showing defects in small RNA 16 and other mut class genes in siRNA pathways in the germline, pathway functions be tested for the mut-16(mg461) allele using they also play important roles in RNAi-related mechanisms in the PCR primers described. somatic tissues. mut-16 is expressed in the soma and is required mut-16 and other mut class genes were shown to be essential for the response to exogenous dsRNAs targeting somatic genes. components of the WAGO class 22G and ERGO-1 class 26G In oocytes, embryos, and larval stages, mut-16 and other mut siRNA pathways but dispensable for the formation or stability of class genes are required for formation or stability of ERGO-1 ALG-3/4 class 26G siRNAs. Given the biochemical similarity be- class 26G siRNAs, which guide silencing of target mRNAs and tween the two classes of 26G siRNAs, it is surprising that ALG-3/4 trigger amplification of secondary 22G siRNAs. Secondary 22G class 26G siRNAs somehow bypass the requirement for mut class siRNAs associate with WAGO class Argonautes expressed in genes. Interestingly, 22G siRNAs derived from ALG-3/4 class 26G the soma, including SAGO-1, SAGO-2, and NRDE-3 (36, 45). siRNA targets were reduced in mut-16(pk710) mutants. It is pos- mut-16– sible that mut-16 is only required for 22G siRNAs derived from NRDE-3 interacts with a subset of dependent somatic ALG-3/4 class 26G siRNA targets that are routed through the 22G 22G siRNAs to suppress pre-mRNA expression in the nucleus by siRNA pathway through an alternate ALG-3/4–independent cotranscriptional repression. Thus, mut-16 also plays an impor- mechanism. This is consistent with the relatively moderate re- tant role in the nuclear RNAi pathway. Given the broad roles of duction (∼70%) observed in 22G siRNAs derived from ALG-3/4 mut class genes in RNAi-related mechanisms, it will be impor- fi class 26G siRNA targets (Fig. 4A). tant to determine the speci c role of each gene in the various We showed that a larger proportion of transposons yield rel- siRNA pathways. Biochemical characterization of mut class genes may aid in our understanding of how certain transcripts atively high levels of siRNAs compared with other classes of C. GENETICS elegans genes, suggesting that they are hypersusceptible to rout- are routed into siRNA-mediated RNA silencing pathways and ing through RNA silencing pathways. Transcription levels of the mechanisms by which silencing occurs. transposons may be substantially higher on average than those of coding genes and pseudogenes, which could account for higher Methods median levels of siRNAs; however, this may not explain the high mg461 was mapped to mut-16 using standard mapping procedures. Strains propensity of transposons to siRNA generation. Tc3 transposons were genotyped for the mut-16(mg461) deletion by PCR and the following yield 21U RNAs that are required for mut-7–dependent Tc3 primers: mut-16 F1—CCCGCCGATACAGAAACTAA; mut-16 R1—AATATTC- — siRNA formation and subsequent transposon silencing (19). GATCGGCAAGCAG; mut-16 F2 CGACTTCCTATGTTTCTTTCCGTG; mut-16 — Thus, in some cases, 21U RNAs seem to trigger siRNA forma- R2 TCAAGTCGTGCACTGTTGCG. A PCR with primers mut-16 F1 and mut-16 tion. Here, we showed that the converse, siRNAs triggering 21U R1 will amplify an 824-bp product from the mut-16 WT allele and a 373-bp product from the mut-16(mg461) allele. A PCR with primers mut-16 F2 and RNA formation, might also occur. Tc1-derived 21U RNA levels mut-16 R2 will amplify a 2.3-kb product from the mut-16 WT allele and a 1.9- were reduced in mut-16(pk710) as well as in mut-2 and mut-7 kb product from the mut-16(mg461) allele. Additional strains used are de- small RNA deep-sequencing datasets, indicating a role for the scribed in SI Appendix, SI Text. Total RNA was isolated using RNA-Bee (Tel- endogenous siRNA machinery in the biogenesis of certain 21U Test). Northern blots were done as described (35, 46, 47). qRT-PCR for mRNAs RNAs. 21U RNAs derived from Tc1 are unique from other was done with iQ SYBR Green Supermix (Bio-Rad). Taqman qRT-PCR of 26G transposon-derived 21U RNAs in that they are processed from siRNAs was done as described (10). Small RNA cDNA amplicons for Illumina the inverted repeat sequences flanking the transposase coding sequencing were prepared as described (9). Sequences were mapped to the region. One possibility is that the siRNA machinery relaxes the C. elegans genome (Wormbase release WS203) using CASHX (version 2.0) presumably strong RNA secondary structure formed by the (48). Detailed methods are in SI Appendix, SI Text. Features used to classify inverted repeats, allowing access to the piRNA machinery. Tc3 endogenous siRNAs are listed in Dataset S7. Sequences of oligonucleotide and retr-1 transposon-derived 21U RNAs, which are not de- primers and probes are listed in Dataset S8. pendent on mut-16, are processed from the transposase sequen- ACKNOWLEDGMENTS. We thank Victor Ambros, Rene Ketting, and John ces that are less likely to form strong secondary RNA structures. Kim for helpful comments; Darryl Conte and Weifeng Gu in Craig Mello’s Transposition of Tc1 transposons was shown to be only slightly laboratory for providing protocols for generation of small RNA cDNA librar- elevated in Piwi mutants (19), indicating that 21U RNAs are not ies for deep sequencing; Allison Billi, Amanda Day, Ting Han, and John Kim essential for Tc1 silencing. This is not surprising given that siR- for protocols for 26G siRNA Taqman qRT-PCR and sharing preliminary data; Christian Daly and Jiangwen Zhang at Faculty of Arts and Sciences Center for NAs seem to act upstream of Tc1 21U RNA formation. In con- Systems Biology of Harvard University and Kaleena Shirley at Massachusetts trast, 21U RNAs derived from Tc3, which act upstream of General Hospital for Illumina deep sequencing; Mark Borowsky and Brad siRNAs, are required for Tc3 silencing (19). Chapman at Massachusetts General Hospital for computational assistance; Previous studies showed that eri-1 and other eri class genes are Leah Frater-Rubsam, Phil Anderson, Jennifer Whangbo, Aidan Porter, Craig required for both ERGO-1 and ALG-3/4 class 26G siRNA path- Hunter, and Andrew Fire for communicating data before publication; and – the Caenorhabditis Genetics Center (CGC) and Shohei Mitani of the Japanese ways (10 13). Both mut class and some eri class mutants show National Bioresources Project for stains. This work was supported by Na- a temperature-sensitive sterility phenotype at elevated temper- tional Institutes of Health Grant GM46419 (to G.R.) and Massachusetts Gen- atures. The temperature-sensitive sterility defect in eri class eral Hospital Executive Committee of Research Fund for Medical Discovery fellowship awards (to C.Z. and S.E.J.F.). Both T.A.M. and C.M.P. are supported mutants can be completely rescued by male mating, indicating by Damon Runyon Cancer Research Foundation Grants DRG-2029-09 and that it is caused entirely by defective sperm. Sperm defects in eri DRG1988-08. T.A.M. is a Damon Runyon Fellow, and C.M.P. is the Marion class mutants are likely caused by loss of ALG-3/4 class 26G Abbe Fellow of the Damon Runyon Cancer Research Foundation.

1. Fire A, et al. (1998) Potent and speciﬁc genetic interference by double-stranded RNA 2. Zamore PD, Tuschl T, Sharp PA, Bartel DP (2000) RNAi: Double-stranded RNA directs the in Caenorhabditis elegans. Nature 391:806–811. ATP-dependent cleavage of mRNA at 21 to 23 nucleotide intervals. Cell 101:25–33.

Zhang et al. PNAS | January 25, 2011 | vol. 108 | no. 4 | 1207 3. Bernstein E, Caudy AA, Hammond SM, Hannon GJ (2001) Role for a bidentate 25. Chung WJ, Okamura K, Martin R, Lai EC (2008) Endogenous RNA interference ribonuclease in the initiation step of RNA interference. Nature 409:363–366. provides a somatic defense against Drosophila transposons. Curr Biol 18:795–802. 4. Hutvágner G, et al. (2001) A cellular function for the RNA-interference enzyme Dicer 26. Tam OH, et al. (2008) Pseudogene-derived small interfering RNAs regulate gene in the maturation of the let-7 small temporal RNA. Science 293:834–838. expression in mouse oocytes. Nature 453:534–538. 5. Hutvagner G, Simard MJ (2008) Argonaute proteins: Key players in RNA silencing. Nat 27. Watanabe T, et al. (2008) Endogenous siRNAs from naturally formed dsRNAs regulate Rev Mol Cell Biol 9:22–32. transcripts in mouse oocytes. Nature 453:539–543. 6. Ghildiyal M, Zamore PD (2009) Small silencing RNAs: An expanding universe. Nat Rev 28. Collins J, Saari B, Anderson P (1987) Activation of a transposable element in the germ Genet 10:94–108. line but not the soma of Caenorhabditis elegans. Nature 328:726–728. 7. Claycomb JM, et al. (2009) The Argonaute CSR-1 and its 22G-RNA cofactors are 29. Vastenhouw NL, et al. (2003) A genome-wide screen identifies 27 genes involved in required for holocentric chromosome segregation. Cell 139:123–134. transposon silencing in C. elegans. Curr Biol 13:1311–1316. 8. van Wolfswinkel JC, et al. (2009) CDE-1 affects chromosome segregation through 30. Tijsterman M, Ketting RF, Okihara KL, Sijen T, Plasterk RH (2002) RNA helicase MUT- uridylation of CSR-1-bound siRNAs. Cell 139:135–148. 14-dependent gene silencing triggered in C. elegans by short antisense RNAs. Science – 9. Gu W, et al. (2009) Distinct argonaute-mediated 22G-RNA pathways direct genome 295:694 697. surveillance in the C. elegans germline. Mol Cell 36:231–244. 31. Chen CC, et al. (2005) A member of the polymerase beta nucleotidyltransferase – 10. Han T, et al. (2009) 26G endo-siRNAs regulate spermatogenic and zygotic gene superfamily is required for RNA interference in C. elegans. Curr Biol 15:378 383. expression in Caenorhabditis elegans. Proc Natl Acad Sci USA 106:18674–18679. 32. Tops BB, et al. (2005) RDE-2 interacts with MUT-7 to mediate RNA interference in – 11. Conine CC, et al. (2010) Argonautes ALG-3 and ALG-4 are required for spermatogenesis- Caenorhabditis elegans. Nucleic Acids Res 33:347 355. specific 26G-RNAs and thermotolerant sperm in Caenorhabditis elegans. Proc Natl Acad 33. Kim JK, et al. (2005) Functional genomic analysis of RNA interference in C. elegans. – Sci USA 107:3588–3593. Science 308:1164 1167. 12. Gent JI, et al. (2010) Distinct phases of siRNA synthesis in an endogenous RNAi 34. Perutz MF, Johnson T, Suzuki M, Finch JT (1994) Glutamine repeats as polar zippers: Their possible role in inherited neurodegenerative diseases. Proc Natl Acad Sci USA pathway in C. elegans soma. Mol Cell 37:679–689. 91:5355–5358. 13. Vasale JJ, et al. (2010) Sequential rounds of RNA-dependent RNA transcription drive 35. Duchaine TF, et al. (2006) Functional proteomics reveals the biochemical niche of C. endogenous small-RNA biogenesis in the ERGO-1/Argonaute pathway. Proc Natl Acad elegans DCR-1 in multiple small-RNA-mediated pathways. Cell 124:343–354. Sci USA 107:3582–3587. 36. Guang S, et al. (2008) An Argonaute transports siRNAs from the cytoplasm to the 14. Lee RC, Feinbaum RL, Ambros V (1993) The C. elegans heterochronic gene lin-4 nucleus. Science 321:537–541. encodes small RNAs with antisense complementarity to lin-14. Cell 75:843–854. 37. Guang S, et al. (2010) Small regulatory RNAs inhibit RNA polymerase II during the 15. Reinhart BJ, et al. (2000) The 21-nucleotide let-7 RNA regulates developmental timing elongation phase of transcription. Nature 465:1097–1101. in Caenorhabditis elegans. Nature 403:901–906. 38. Corrêa RL, Steiner FA, Berezikov E, Ketting RF (2010) MicroRNA-directed siRNA 16. Ambros V, Lee RC, Lavanway A, Williams PT, Jewell D (2003) MicroRNAs and other biogenesis in Caenorhabditis elegans. PLoS Genet 6:e1000903. tiny endogenous RNAs in C. elegans. Curr Biol 13:807–818. 39. Sijen T, Plasterk RH (2003) Transposon silencing in the Caenorhabditis elegans germ 17. Ruby JG, et al. (2006) Large-scale sequencing reveals 21U-RNAs and additional line by natural RNAi. Nature 426:310–314. – microRNAs and endogenous siRNAs in C. elegans. Cell 127:1193 1207. 40. Britten RJ (1995) Active gypsy/Ty3 retrotransposons or retroviruses in Caenorhabditis 18. Batista PJ, et al. (2008) PRG-1 and 21U-RNAs interact to form the piRNA complex elegans. Proc Natl Acad Sci USA 92:599–601. – required for fertility in C. elegans. Mol Cell 31:67 78. 41. Chen C, et al. (2005) Real-time quantification of microRNAs by stem-loop RT-PCR. 19. Das PP, et al. (2008) Piwi and piRNAs act upstream of an endogenous siRNA pathway Nucleic Acids Res 33:e179. to suppress Tc3 transposon mobility in the Caenorhabditis elegans germline. Mol Cell 42. Domeier ME, et al. (2000) A link between RNA interference and nonsense-mediated – 31:79 90. decay in Caenorhabditis elegans. Science 289:1928–1931. 20. Wang G, Reinke V (2008) A C. elegans Piwi, PRG-1, regulates 21U-RNAs during 43. Welker NC, et al. (2010) Dicer’s helicase domain is required for accumulation of some, – spermatogenesis. Curr Biol 18:861 867. but not all, C. elegans endogenous siRNAs. RNA 16:893–903. 21. Ketting RF, Haverkamp TH, van Luenen HG, Plasterk RH (1999) Mut-7 of C. elegans, 44. Kennedy S, Wang D, Ruvkun G (2004) A conserved siRNA-degrading RNase negatively required for transposon silencing and RNA interference, is a homolog of Werner regulates RNA interference in C. elegans. Nature 427:645–649. syndrome helicase and RNaseD. Cell 99:133–141. 45. Yigit E, et al. (2006) Analysis of the C. elegans Argonaute family reveals that distinct 22. Tabara H, et al. (1999) The rde-1 gene, RNA interference, and transposon silencing in Argonautes act sequentially during RNAi. Cell 127:747–757. C. elegans. Cell 99:123–132. 46. Lee RC, Ambros V (2001) An extensive class of small RNAs in Caenorhabditis elegans. 23. Yang N, Kazazian HH, Jr. (2006) L1 retrotransposition is suppressed by endogenously Science 294:862–864. encoded small interfering RNAs in human cultured cells. Nat Struct Mol Biol 13: 47. Fischer SE, Butler MD, Pan Q, Ruvkun G (2008) Trans-splicing in C. elegans generates 763–771. the negative RNAi regulator ERI-6/7. Nature 455:491–496. 24. Czech B, et al. (2008) An endogenous small interfering RNA pathway in Drosophila. 48. Fahlgren N, et al. (2009) Computational and analytical framework for small RNA Nature 453:798–802. profiling by high-throughput sequencing. RNA 15:992–1002.

1208 | www.pnas.org/cgi/doi/10.1073/pnas.1018695108 Zhang et al.