Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2015 Dec 4.
Published in final edited form as: Mol Cell. 2014 Nov 20;56(5):708–716. doi: 10.1016/j.molcel.2014.10.016

The Initial Uridine of Primary piRNAs Does not Create the Tenth Adenine That is the Hallmark of Secondary piRNAs

Wei Wang 1,2,4, Mayu Yoshikawa 3,4, Bo W Han 1, Natsuko Izumi 3, Yukihide Tomari 3,*, Zhiping Weng 2,*, Phillip D Zamore 1,*
PMCID: PMC4337030  NIHMSID: NIHMS663586  PMID: 25453759

SUMMARY

PIWI-interacting RNAs (piRNAs) silence transposons in animal germ cells. PIWI proteins bind and amplify piRNAs via the “Ping-Pong” pathway. Because PIWI proteins cleave RNAs between target nucleotides t10 and t11—the nucleotides paired to piRNA guide positions g10 and g11—the first ten nucleotides of piRNAs participating in the Ping-Pong amplification cycle are complementary. Drosophila piRNAs bound to the PIWI protein Aubergine typically begin with uridine (1U), while piRNAs bound to Argonaute3, which are produced by Ping-Pong amplification, often have adenine at position 10 (10A). The Ping-Pong model proposes that the 10A is a consequence of 1U. We find that 10A is not caused by 1U. Instead, fly Aubergine as well as its homologs, Siwi in silkmoth and MILI in mice, have an intrinsic preference for adenine at the t1 position of their target RNAs; during Ping-Pong amplification, this t1A subsequently becomes the g10A of a piRNA bound to Argonaute3.

INTRODUCTION

PIWI-interacting RNAs (piRNAs) protect the genome of animal germ cells by silencing transposons and repetitive sequences (Aravin et al., 2007; Grimson et al., 2008; Houwing et al., 2008; Armisen et al., 2009; Friedlander et al., 2009; Kawaoka et al., 2009; Lau et al., 2009). Ranging from 23–35 nt, piRNAs were discovered in the Drosophila melanogaster testis (Aravin et al., 2001) and ovary (Aravin et al., 2003). PIWI proteins, which form an animal-specific clade of the Argonaute family of small RNA- or DNA-guided proteins found in all three domains of life (Cenik and Zamore, 2011).

The domain architecture of Argonaute proteins allows them to bind target RNAs or DNAs via sequence complementarity to a 6–8 nt subsequence of the guide RNA or DNA, the seed sequence. Argonaute proteins prepay the entropic cost of hybridization by displaying the seed sequence in a conformation that facilitates target binding (Lewis et al., 2003; Ma et al., 2005; Parker et al., 2005; Bartel, 2009; Parker et al., 2009; Wang et al., 2009; Frank et al., 2010; Boland et al., 2011; Lambert et al., 2011; Frank et al., 2012; Cora et al., 2014). PIWI and other Argonaute proteins can catalyze endonucleolytic cleavage of their targets. Target cleavage requires extensive complementarity between the guide and the target to enable a conformational change that brings the target closer to the Argonaute active site (Wang et al., 2008; Wang et al., 2009). Anchoring of the 5′ phosphate of the guide in a phosphate-binding pocket ensures that the guide remains tightly bound through many rounds of target cleavage (Parker et al., 2005; Yuan et al., 2005). Anchoring the 5′ phosphate in the pocket also helps position the target in the cleavage site: the scissile phosphate always lies between positions t10 and t11 of the target (Elbashir et al., 2001a; Elbashir et al., 2001b; Rivas et al., 2005).

In flies, repetitive, transposon-rich genomic regions, “piRNA clusters,” produce precursor transcripts that are processed into primary piRNAs (Brennecke et al., 2007). piRNA clusters contain numerous, often nested, transposition events that record the past transposon exposure of the animal. Both strands of most fly germline piRNA clusters are transcribed. The transposon orientations in piRNA clusters appear to be random, yet most piRNAs are antisense to transposon mRNAs.

The Ping-Pong model of piRNA biogenesis (Brennecke et al., 2007; Gunawardane et al., 2007) attempts to explain how the initial piRNA population is amplified and how the antisense bias is acquired, thereby increasing the number of piRNAs that can silence transposon mRNAs. The model proposes that primary or maternally deposited antisense piRNAs are loaded into the PIWI protein Aubergine (Aub), directing it to bind and cleave complementary targets, including both piRNA cluster transcripts and transposon mRNAs (Brennecke et al., 2008). The model further proposes that the PIWI protein Argonaute3 (Ago3) binds the 5′ end of the 3′ cleavage product made by Aub, initiating production of an Ago3-bound secondary piRNA that can guide Ago3 to bind and cleave another piRNA cluster transcript. The resulting 3′ cleavage product is then loaded into Aub, generating a new, Aub-bound secondary piRNA that can initiate another cycle of Ping-Pong amplification.

The Ping-Pong model seeks to explain three remarkable features of piRNAs: (1) Aub-bound piRNAs are typically antisense to transposon mRNAs, while Ago3-bound piRNAs are usually sense; (2) the first 10 nucleotides of Aub-bound piRNAs are often complementary to the first 10 nucleotides of Ago3-bound piRNAs, a relationship termed the “Ping-Pong signature”; and (3) Aub-bound piRNAs often begin with a uracil (g1U), whereas Ago3-bound piRNAs show no first nucleotide bias (g1N), but tend to have an adenosine as their tenth nucleotide (g10A). The model proposes that the g1U of an Aub-bound piRNA selects a target RNA with a corresponding t1A nucleotide. When the 3′ cleavage product generated by Aub is loaded into Ago3, the t1 position becomes the g10 position, generating the g10A signature characteristic of Ago3-bound piRNAs.

Here, we report that fly Aub, Bombyx mori Siwi, and mouse MILI select targets bearing a t1A nucleotide irrespective of the identity of the g1 nucleotide of their piRNA guide. Consequently, the g10A of Ago3-bound piRNAs arises not through g1U:t1A base pairing, but rather reflects an intrinsic property of the protein. A similar preference for t1A was noted earlier for mammalian AGO proteins guided by microRNAs (miRNAs) (Lewis et al., 2005). In contrast, fly and silkmoth Ago3 proteins do not have the nucleotide preference at the t1 position. We propose that mammalian AGO proteins and a subset of mammalian and insect PIWI proteins contain a binding pocket that best accommodates adenine at the t1 position of their targets. Thus, the tendency of Aub-bound piRNAs to begin with a U is, at least in part, a consequence, not a cause, of the g10A of Ago3-bound piRNAs.

RESULTS

Cause and Effect in the Ping-Pong Model

The Ping-Pong model proposes that an antisense primary piRNA guides Aub to cut a transposon mRNA between nucleotides t10 and t11, the target nucleotides that pair with nucleotides g10 and g11 of the piRNA guide. The resulting 3′ cleavage product is loaded into Ago3. The model envisions that the 5′ g1U of the guide piRNA bound to Aub dictates the t1A of the cleaved target. After the target is converted to a secondary piRNA loaded into Ago3, t1A becomes g10A—the hallmark of Ago3-bound piRNAs (Figure 1A, Model I) (Brennecke et al., 2007; Gunawardane et al., 2007).

Figure 1. Models for the origin of the Adenine at the Tenth Position of Ago3 piRNAs.

Figure 1

Open in a new tab

(A) Model I posits that g1U causes t1A by Watson and Crick pairing. Model II accommodates the interactions between the 5′ phosphate of guides and the phosphate-binding pocket in Argonaute proteins.

(B) Nucleotide composition of g1 from guide piRNAs and t1 (g10) from their targets inferred from Ping-Pong analysis in w1 ovaries.

(C) Numbers of cis Ping-Pong pairs.

(D) Aub requires g2–g16 complementarity to cleave a target RNA. The frequency of pairing for the individual positions g11–g16 was higher in the biological data than in the shuffled controls (left). When examining the extent of contiguous pairing rather than the frequency of pairing at individual positions, the pairing frequency for the biological data was only higher than the pairing frequency in the shuffled controls when base pairing extended from position g2 to at least g13. Error bars (mean ± 2 × S.D., n = 10) indicate the paired frequency for the shuffled controls.

See also Figure S1.

This idea conflicts with an evolutionarily conserved property of Argonaute proteins: the first nucleotide of a small RNA or DNA guide bound to a eubacterial, archael, or eukaryotic Argonaute lies in a 5′ phosphate-binding pocket that precludes the first guide base (g1) from pairing with a complementary target base (t1) (Ma et al., 2005; Parker et al., 2005; Wang et al., 2009; Frank et al., 2010; Boland et al., 2011; Elkayam et al., 2012; Frank et al., 2012; Cora et al., 2014). In fact, a first position mismatch does not impair Argonaute-directed target cleavage (Haley and Zamore, 2004; Haley and Zamore, 2004; Reuter et al., 2011; Wee et al., 2012).

The Ping-Pong model is not the first to grapple with the origin of a t1A across from a g1U in an Argonaute-bound guide. Most mammalian miRNAs begin with uridine, and high-confidence miRNA-binding sites in vertebrates typically bear a t1A across from the g1U position. Yet the t1A persists even when the miRNA begins with A, C, or G, rather than changing to the complementary nucleotide (Lewis et al., 2005; Grimson et al., 2007). This unexpected observation likely reflects the presence of a t1 adenosine-binding site in Argonaute. Might the g10A of Ago3-bound piRNAs reflect a propensity of Aub to bind targets bearing adenosine at the t1 position, rather than a consequence of the g1U of Aub-bound piRNAs (Figure 1A, Model II)?

Cis- and trans-Targets

Ago3-bound piRNAs result from the target cleavage by Aub-bound piRNAs, so the targets of Aub-bound piRNAs can be inferred from Ago3-bound piRNAs. We immunoprecipitated Ago3 and Aub from w1 fly ovaries and sequenced the piRNAs bound to each protein. We used piRNAs uniquely bound to Ago3 (60% of species and 17% of reads of all Ago3-bound piRNAs) to infer the targets of piRNAs uniquely bound to Aub (76% of species and 22% of reads of all Aub-bound piRNAs). Such targets fall into two categories: cis-targets, which overlap their piRNAs in genomic coordinates, and trans-targets, which do not. A cis-target and its piRNA guide, for example, could correspond to the two precursor transcripts from the opposite genomic strands of a dual-strand piRNA cluster. In contrast, trans-targets could correspond to mRNAs transcribed from euchromatic transposon insertions.

For Aub-bound piRNAs and their cis-Targets, only g1U:t1A Plays Ping-Pong

In its simplest form, Model I predicts that the g1 nucleotide of an Aub-bound piRNA should always pair with the t1 nucleotide of its cis target. Nearly half of Aub-bound piRNAs do not begin with U (Figure 1B). To test whether Aub-bound piRNAs beginning with A, C, or G participate in the Ping-Pong cycle, we analyzed the 5′-to-5′ distance between each Aub-bound piRNA and its overlapping Ago3-bound piRNA from the opposite genomic strand (Figure 1C). Non-Ping-Pong overlaps (1–9 and 11–16) were used as the background distribution to compute Ping-Pong Z-scores (i.e., 10 nt overlap). A small number of piRNA species could dominate Z-score, so we performed the analysis for species and reads. As anticipated, the number of pairs of Aub-bound g1U piRNAs with t1A target partners was significantly greater than the background attributable to chance (Z-score =16.8 for reads and 7.34 for species). Yet the other three complementary pairs g1A:t1U, g1C:t1G, and g1G:t1C were not significant (Figure 1C).

Similarly, mouse MILI showed significant g1U:t1A Ping-Pong with its cis-targets inferred from MIWI2-bound piRNAs; the three other complementary g1:t1 pairs did not (Figure S1A). In other words, g1V (i.e., not U) piRNAs bound to fly Aub or mouse MILI are not significantly amplified by the Ping-Pong cycle. We can imagine two explanations for this surprising observation: (1) efficient cleavage catalyzed by Aub or MILI requires a g1U (Figure 1A; Model I) or (2) Aub and MILI prefer to bind and cleave target RNAs that bear a t1A (Figure 1A; Model II).

Trans-Targets of Aub Reveal Its Preference for t1A Targets

To distinguish between these two explanations, we analyzed the trans-targets of Aub-bound piRNAs. Standard Ping-Pong analyses measure the distance between the 5′ ends of piRNAs that overlap in genomic space (Brennecke et al., 2007; Klattenhoff et al., 2009; Li et al., 2009). Such analyses only detect piRNA:cis-target pairs. Cis-targets are uninformative for our purposes: the g1 nucleotide of a piRNA is always complementary to the t1 of a cis-target, because the two strands of genomic DNA are always complementary.

The biochemical properties of Argonaute proteins suggest that piRNA-directed cleavage requires extensive, but not complete complementarity between piRNA and target (Ameres et al., 2007; Reuter et al., 2011; Wee et al., 2012). To test the extent of complementarity required for trans pairing, we examined the length of complementarity between Aub-bound piRNAs and their potential targets—inferred from Ago3-bound piRNAs—requiring complementary to positions g2–g10. We compared the observed frequency of piRNA:target pairing at each position g11–g23 to the frequency expected by chance. We also computed the frequency of contiguous piRNA:target base pairing from position g2 to each position g11–g23. The data suggest that authentic trans-targets of Aub-bound piRNA are generally complementary to their guides from g2 until at least position g16 (Figure 1D). Similarly, our analysis suggests that Ago3 can slice a target when piRNA:target complementarity extends from g2 to at least g14 (Figure S1B). Although our results for contiguous pairing are consistent with Aub and Ago3 tolerating one or more mismatches within regions of complementarity, we restricted our analysis of predicted trans-targets to those with complete complementarity from positions g2–g16 (Figure 2A and Figure S1C). Using this metric, we could infer cis- or trans-targets from the Ago3 piRNA population for 35% of piRNAs uniquely bound to Aub. Conversely, we could infer cis or trans targets from Aub-bound piRNAs for 52% of piRNAs uniquely bound to Ago3. Among all possible Aub-bound piRNA:target pairs, 33% were unambiguously identified as trans-targets.

Figure 2. Trans Ping-Pong Analysis Reveals a t1A Preference for Aub.

Figure 2

Open in a new tab

(A) Schema for detecting trans-targets.

(B) Ping-Pong profile for each guide:target trans pair with at least 16 nt complementarity.

See also Figures S2 and S3.

To test whether Aub requires g1U to cleave targets, we analyzed the trans-targets of Aub-bound piRNAs beginning with A, C, or G. For those Aub-bound piRNAs with Ping-Pong partners, ~32% begin with A, C, or G. Thus, g1V piRNAs bound to Aub function as guides for trans-targets.

To test whether Aub prefers to bind and cleave target RNAs that bear a t1A, we asked whether these g1V, Aub-bound piRNAs were complementary to their trans-targets at the t1 position. Among g1V piRNAs, the most significant t1 nucleotide for trans-targets was t1A: the mismatched pairs g1A:t1A, g1C:t1A, g1G:t1A were far more likely than expected by chance, while the frequency of the complementary pairs, g1A:t1U, g1C:t1G, and g1G:t1C, were nearly indistinguishable from background (Figure 2B). Other mismatched g1V:t1B (B = not A) pairs were non-significant (Z-score = 2.96 corresponds to a Bonferroni corrected, two-tailed p-value of 0.05) or significant but much closer to the background distribution than g1V:t1A.

Our analysis removes the most abundant piRNA species, because abundant piRNAs are typically present in both the Aub and Ago3 immunoprecipitates. (Only piRNAs mapping perfectly to the genome were analyzed, reducing the likelihood that differences between the g1 position of otherwise identical piRNA species reflect sequencing errors.) Among 2,050,844 Aub-bound piRNA species, read 20,938,987 times in total, only 11,577 (0.6%) species (1,645,988 reads; 8%) differ from at least one other Aub-bound piRNA solely at the g1 position. Among this subset, about half (~53% of species; ~60% of reads) were either (1) present in both the Ago3 and Aub immunoprecipitates or (2) contain a sibling piRNA uniquely assigned to Aub or Ago3 that does not participate in Ping-Pong. These two classes of piRNAs do not contribute to our analysis. However, a third class of abundant piRNAs could, in theory, bias our results: sibling piRNAs that differ solely at their g1 position and for which one, typically less abundant, sibling participates in Ping-Pong with piRNAs uniquely bound to a single PIWI protein. To assess the impact of such piRNAs on our results, we removed this third type of piRNA and repeated the analysis. Our conclusions were unaltered: a g1U piRNA bound to Aub does not cause the t1A of the target RNA.

To include more of the most abundant piRNA species, we developed a strategy to assign piRNAs to Aub or Ago3 without requiring the piRNA to be uniquely present in one immunoprecipitate. Instead, a piRNA was required to be fivefold more abundant for one PIWI protein than the other and have a p-value from a χ2 test ≤ 10−40. (Our conclusions remained the same using p-values > 0.005, 10−10, or 10−20 with the fivefold ratio or using a p-value > 0.005 with a 20-fold ratio.)

This approach identified 4,633 additional species for Aub and 6,360 for Ago3, but many additional reads: 5,095,308 reads for Aub and 6,908,635 for Ago3. We added these piRNAs back to the original, more stringently assigned set and repeated our analysis. To avoid having a small number of highly abundant piRNA pairs dominate, we performed the new analysis with respect to species. As before, only g1X:t1A Aub:Ago3 trans pairs were significant (Figure S2C). Taken together, the results of our analyses suggest that an Aub prefers a t1A irrespective of the identity of the g1 nucleotide.

Selection of a t1A nucleotide is likely also a property of mammalian PIWI proteins. We detected a preference for t1A, irrespective of the identity of the g1 nucleotide, for piRNAs uniquely bound to MILI in the embryonic mouse testis, using targets inferred from piRNAs uniquely bound to MIWI2 (Figure S2B, left). We also detected a weaker preference for t1G, suggesting that MILI selects targets with a t1 purine. Requiring 19-nt (g2–g20) rather than 16-nt complementarity did not change our conclusions (Figure S3).

Catalytically Inactive Ago3ADH Accumulates t1A Secondary piRNAs

To further test whether Aub prefers t1A targets irrespective of the identity of the g1 nucleotide of its piRNA, we engineered flies in which the Ping-Pong cycle was abridged by expressing catalytically inactive Ago3 protein (Ago3ADH) in an ago3t2/ago3t3 mutant background. Ago3ADH is predicted not to cleave RNA; Ago3AAH can not cleave RNA in vitro (Liu et al., 2004; Huang et al., 2014). These flies are expected not to sustain the Ping-Pong amplification cycle: Aub-bound piRNAs will only be primary and maternal piRNAs and Ago3ADH bound piRNAs will correspond to the first cycle of secondary piRNAs produced by Aub-catalyzed cleavage; no Aub-bound secondary piRNAs can be produced by Ago3ADH.

In flies lacking Ago3, Aub:Aub homotypic Ping-Pong replaces normal, productive heterotypic Aub:Ago3 Ping-Pong (Li et al., 2009; Zhang et al., 2011). In contrast, the presence of Ago3ADH suppresses homotypic Aub:Aub Ping-Pong because Ago3ADH can accept cleavage products from Aub. Indeed, we observed weaker overall Ping-Pong in the ago3ADH ovaries (Z-score = 11.7) than in ovaries lacking Ago3 altogether (Z-score = 50.2). cis Ping-Pong between piRNAs uniquely bound to Aub and piRNAs uniquely bound to Ago3ADH (Z-score = 3.30) was far weaker than cis Ping-Pong between piRNAs uniquely bound to Aub and piRNAs uniquely bound to wild-type Ago3 (Z-score = 16.8; Figure 1C). Using piRNAs uniquely bound to Ago3ADH to infer trans-targets, g1U:t1A, g1A:t1A and g1C:t1A were the only significant piRNA:trans-target pairs (Figure 3).

Figure 3. The t1A Preference for Aub is Reinforced in Aub:Ago3ADH and Aub:degradome guide:target pairs.

Figure 3

Open in a new tab

(A) Nucleotide composition of position g1 of piRNAs and position t1 of their targets as inferred from piRNAs bound to Ago3ADH or detected directly in the RNA degradome of w1 ovaries.

(B) Guide piRNAs and their inferred targets were determined by sequencing the piRNAs in immunoprecipitates of Aub or Ago3ADH.

In these analyses, trans-targets were inferred from the piRNAs bound to wild-type or mutant Ago3. Of course, not all trans-targets cleaved by Aub are expected to escape destruction and generate an Ago3-bound secondary piRNA. To more accurately capture the targets cleaved by Aub, we sequenced 5′ monophosphorylated RNAs >200 nt from w1 ovaries. We obtained 4,019,873 such “degradome” reads matching the fly genome; most (~64%) corresponded to mRNA fragments. An additional 2.3% mapped to transposons. Among these, ~25% showed the 10-nt 5′-5′ overlap with Aub-bound piRNAs expected for Aub-catalyzed, piRNA-directed 3′ cleavage fragments. Of these, 34% were trans-targets.

For these putative Aub-catalyzed, piRNA-directed 3′ cleavage products, g1U:t1A (Z-score = 22.5), g1A:t1A (Z-score = 15.5), g1C:t1A (Z-score = 14.7) and g1G:t1A (Z-score = 5.61) piRNA:trans-target pairs were both significant and more abundant than other types of pairings. One additional pair, g1C:t1U (Z-score = 5.45) was also significant, but was only 23% as abundant as the g1C:t1A pairs. Thus, when we inferred (1) the trans-targets of Aub-bound piRNAs from the secondary piRNAs-bound to wild-type or catalytically inactive Ago3 or (2) identified targets directly from degradome sequencing, Aub appears to select targets bearing a t1A nucleotide irrespective of the identity of its g1 piRNA nucleotide (Figure 3).

Ago3 has no t1A Preference

Ago3 appears to have no t1A preference (Figure 1B). We inferred the trans-targets of Ago3 from the piRNAs bound to Aub. Ago3-bound piRNAs showed no t1 nucleotide preference: all 16 g1:t1 Ago3:trans-target pairs were significant (Figure 2B, right). We conclude that Ago3 is impartial to the identity of the t1 target nucleotide that faces the g1 piRNA nucleotide and requires only sufficient complementarity to bind and cleave its RNA targets.

MIWI2, unlike Ago3, Prefers a Purine

In insects and mammals, the Ping-Pong pathway amplifies transposon-silencing piRNAs. In mice, homotypic MILI:MILI Ping-Pong dominates piRNA amplification, perhaps explaining why mouse transposon-silencing piRNAs are sense biased (Aravin et al., 2008; De Fazio et al., 2011). Heterotypic MILI:MIWI2 Ping-Pong does take place, but in vivo the process is not reciprocal (De Fazio et al., 2011).

We used publicly available data to infer the trans-targets of MIWI2 piRNAs from the piRNAs bound to MILI. Unlike Ago3, but like MILI, MIWI2 favors a purine at the t1 position (Figure S2A). Consistent with this preference, MILI-bound piRNAs favor a purine at their g10 position, the nucleotide that serves as t1 for MIWI2 (Figure S2B, right); no such g10 bias was detected for Aub-bound piRNAs in fly ovaries (Figure 1B and 2B, right).

Bombyx mori Siwi, like Drosophila Aub, Prefers to Cleave a t1A Target

Our analyses suggest that fly Aub and mouse MILI, but not fly Ago3, have an intrinsic preference for t1A targets, regardless of the identity of the facing g1 nucleotide. Immortalized Bombyx mori BmN4 cells currently provide the only cell-free extracts for analyzing the germline piRNA pathway in insects. We used BmN4 extract to examine the t1 preference of Siwi, the silkmoth ortholog of Aub. We measured the cleavage efficiency of PIWI proteins immunopurified from BmN4 cells expressing FLAG-tagged Siwi or FLAG-tagged BmAgo3 (Figure 4A).

Figure 4. Siwi, the Silkmoth Ortholog of Aub, Prefers t1A Targets.

Figure 4

Open in a new tab

(A) FLAG-Siwi or FLAG-BmAgo3 were immunopurified from BmN4 lysates and incubated with target RNAs bearing t1A, U, G or C but otherwise fully complementary to an abundant g1U or g1C piRNA. Representative data is shown.

(B) Quantification of the experiments in (A). p-values of one-way ANOVA and Tukey’s multiple comparison test are shown (n = 3).

See also Figure S4.

Cross-contamination of one PIWI protein by the other was small (Figure S4). We used high-throughput sequencing data to identify abundant g1U and g1C piRNAs bound to FLAG-tagged Siwi or BmAgo3 (Izumi et al., 2013). For each piRNA, we constructed a set of fully complementary target RNAs bearing A, U, G or C at the t1 position. When guided by the g1U or the g1C piRNA, Siwi cleaved the t1A target significantly better than t1U, t1G or t1C (for all pairwise comparisons, the p-value by Tukey’s multiple comparison test was ≤1 × 10−3; Figure 4B). In contrast, BmAgo3 showed little or no t1 nucleotide preference among the four targets (Figure 4B). Thus, like Drosophila Aub, Bombyx mori Siwi selects targets bearing a t1A nucleotide, regardless of the g1 nucleotide. Conversely, BmAgo3, like fly Ago3, has little or no t1 nucleotide preference.

DISCUSSION

To date, models for secondary piRNA biogenesis have assumed that the g10A characteristic of Ago3-bound secondary piRNAs is generated by a requirement for the g1U of an Aub-bound piRNA to pair with the t1 position of its cleavage targets. Our analyses demonstrate that the t1A preference is maintained even when g1:t1 cannot pair. Taken together, our computational and biochemical experiments suggest a revision to the standard Ping-Pong model for piRNA amplification: the g10A hallmark of secondary piRNAs reflects a preference of Aub for a t1A and is not a consequence of g1U:t1A base pairing. This revision helps unify the piRNA and miRNA target selection mechanisms: mammalian Argonaute proteins, and likely those in other animals and plants, display a preference for t1A that is determined by the protein and not base pairing. We note that for both pathways, a protein preference for t1A eliminates the paradox that the g1U base was proposed to base pair despite its appearing unavailable for base pairing in multiple three-dimensional structures.

What purpose does the preference of Aub, Siwi, and mouse MILI for a t1 adenine serve in the Ping-Pong cycle? One consequence of the t1A preference of Aub is that Ago3-bound piRNAs will bear a g10A even when the Aub-bound piRNA does not begin with U. Subsequently, this Ago3-bound, g10A piRNA will cleave only targets that bear a t10U, because base pairing of the central region of a small RNA guide is required for target cleavage. Because the t10 nucleotide becomes position g1 of the piRNA generated by Ago3-catalyzed target cleavage, the process ensures that secondary piRNAs loaded into Aub (or Siwi or MILI) begin with uracil. Biochemical experiments demonstrate that Siwi prefers g1U guides (Kawaoka et al., 2011), and the MID domain of mouse MIWI specifically recognizes U (Cora et al., 2014). Current evidence is consistent with Aub also preferring g1U. Thus, the t1A preference of Aub serves to produce optimal piRNA precursors for loading into Aub and Piwi, the PIWI proteins that mediate transposon silencing in flies. In contrast, the absence of a t1 nucleotide preference for Ago3 expands the pool of sequences that can become piRNAs: if Ago3 also preferred a t1A, piRNA production would be restricted to genomic regions with an A exactly 10 nucleotides from a U. Understanding how Aub, Siwi, and MILI select for t1A remains a challenge for future studies of these proteins.

EXPERIMENTAL PROCEDURES

Cis piRNA Pairs

To detect Aub:Ago3 Ping-Pong pairs, we identified Aub-bound guides immunoprecipitated with anti-Aub and inferred targets from the Ago3-bound piRNAs immunoprecipitated with anti-Ago3. We removed those piRNAs that present in both immunoprecipitates to avoid ambiguity in assignment.

Putative Aub:Ago3 Ping-Pong pairs were required to overlap by 10 nt with perfect complementarity. To identify cis-pairs, we mapped all piRNAs to the reference fly genome without allowing mismatches. When a piRNA mapped to more than one genomic location, if any combinations of the genome mapping locations of the two piRNAs lead to a 10-nt 5′-5′ overlap, the pair was classified as a cis-pair.

Estimation of the Extent of Complementarity Required for Trans Ping-Pong

To estimate the extent of complementarity required, we analyzed all putative trans pairs for their pairing status at positions g11–g23. The t11–t23 sequence of a putative trans-target was inferred from the genomic sequence immediately upstream of the piRNA (Figure S1C). The read counts of the guide and target were multiplied to yield pair abundance. For each position g11–g23, we summed the abundance of pairs exhibiting complementarity. Normalization across the 13 positions gave the pairing frequency.

To assess the significance of the pairing frequency of putative trans-pairs, we randomly shuffled the t11–t23 subsequences among putative trans-targets while maintaining their t2–t10 sequences and pairing relationship with their guides. Ten sets of shuffled sequences gave the control pairing frequencies. The pair abundance of contiguous complementarity from g2 until X were normalized across g11–g23 to give pairing frequency; controls using shuffled data were normalized separately.

Trans Ping-Pong Pair Analyses

Trans-pairs were required to be perfectly complementary from g2–g16. When multiple guides paired with the same trans-target, reads were apportioned evenly among the pairs. On the other hand, when one guide paired with multiple trans-targets, we did not apportion the reads of the guide (apportioning the guide but not the target would have reached the same result).

Supplementary Material

Supplemental Data
Supplemental Information

Acknowledgments

We thank members of the Zamore, Weng and Tomari laboratories for helpful discussions; Cindy Tipping for help with fly husbandry; Jia Xu for her pioneering efforts constructing the Ping-Pong algorithm; and Chengjian Li for advice on library construction. Supported in part by National Institutes of Health grants HG007000 to Z.W. and GM62862 and GM65236 to P.D.Z., a Grant-in-Aid for Scientific Research on Innovative Areas (21115002 and 26113007) from the Japan Ministry of Education, Culture, Sports, Science and Technology to Y.T., and a Honjo International Scholarships for Graduate Students to M.Y.

Footnotes

ACCESSION NUMBERS

Small RNA deep sequencing data have been deposited in the NCBI Short Read Archive (www.ncbi.nlm.nih.gov/sites/sra) with accession number SRP043482.

SUPPLEMENTAL INFORMATION

Supplemental information includes Supplemental Experimental Procedures, four figures and one table, and can be found with this article online.

AUTHOR CONTRIBUTIONS

W.W., M.Y., Y.T, Z.W., and P.D.Z. conceived and designed the experiments. M.Y. and N.I. performed biochemical experiments and M.Y., N.I., and Y.T. analyzed the results. B.W.H. performed the immunoprecipitation, small RNA cloning and degradome cloning. W.W. implemented the algorithm and analyzed the deep sequencing data. W.W., M.Y., Y.T, Z.W., and P.D.Z. wrote the manuscript.

References

  1. Ameres SL, Martinez J, Schroeder R. Molecular basis for target RNA recognition and cleavage by human RISC. Cell. 2007;130:101–112. doi: 10.1016/j.cell.2007.04.037. [DOI] [PubMed] [Google Scholar]
  2. Aravin AA, Lagos-Quintana M, Yalcin A, Zavolan M, Marks D, Snyder B, Gaasterland T, Meyer J, Tuschl T. The small RNA profile during Drosophila melanogaster development. Dev Cell. 2003;5:337–350. doi: 10.1016/s1534-5807(03)00228-4. [DOI] [PubMed] [Google Scholar]
  3. Aravin AA, Sachidanandam R, Bourc’his D, Schaefer C, Pezic D, Toth KF, Bestor T, Hannon GJ. A piRNA pathway primed by individual transposons is linked to de novo DNA methylation in mice. Mol Cell. 2008;31:785–799. doi: 10.1016/j.molcel.2008.09.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Aravin AA, Sachidanandam R, Girard A, Fejes-Toth K, Hannon GJ. Developmentally regulated piRNA clusters implicate MILI in transposon control. Science. 2007;316:744–747. doi: 10.1126/science.1142612. [DOI] [PubMed] [Google Scholar]
  5. Aravin AA, Naumova NM, Tulin AV, Vagin VV, Rozovsky YM, Gvozdev VA. Double-stranded RNA-mediated silencing of genomic tandem repeats and transposable elements in the D. melanogaster germline. Curr Biol. 2001;11:1017–1027. doi: 10.1016/s0960-9822(01)00299-8. [DOI] [PubMed] [Google Scholar]
  6. Armisen J, Gilchrist MJ, Wilczynska A, Standart N, Miska EA. Abundant and dynamically expressed miRNAs, piRNAs, and other small RNAs in the vertebrate Xenopus tropicalis. Genome Res. 2009;19:1766–1775. doi: 10.1101/gr.093054.109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bartel DP. MicroRNAs: target recognition and regulatory functions. Cell. 2009;136:215–233. doi: 10.1016/j.cell.2009.01.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Boland A, Huntzinger E, Schmidt S, Izaurralde E, Weichenrieder O. Crystal structure of the MID-PIWI lobe of a eukaryotic Argonaute protein. Proc Natl Acad Sci U S A. 2011;108:10466–10471. doi: 10.1073/pnas.1103946108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Brennecke J, Aravin AA, Stark A, Dus M, Kellis M, Sachidanandam R, Hannon GJ. Discrete small RNA-generating loci as master regulators of transposon activity in Drosophila. Cell. 2007;128:1089–1103. doi: 10.1016/j.cell.2007.01.043. [DOI] [PubMed] [Google Scholar]
  10. Brennecke J, Malone CD, Aravin AA, Sachidanandam R, Stark A, Hannon GJ. An epigenetic role for maternally inherited piRNAs in transposon silencing. Science. 2008;322:1387–1392. doi: 10.1126/science.1165171. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Cenik ES, Zamore PD. Argonaute proteins. Curr Biol. 2011;21:R446–R449. doi: 10.1016/j.cub.2011.05.020. [DOI] [PubMed] [Google Scholar]
  12. Cora E, Pandey RR, Xiol J, Taylor J, Sachidanandam R, McCarthy AA, Pillai RS. The MID-PIWI module of Piwi proteins specifies nucleotide- and strand-biases of piRNAs. RNA. 2014;20:773–781. doi: 10.1261/rna.044701.114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. De Fazio S, Bartonicek N, Di Giacomo M, Abreu-Goodger C, Sankar A, Funaya C, Antony C, Moreira PN, Enright AJ, O’Carroll D. The endonuclease activity of Mili fuels piRNA amplification that silences LINE1 elements. Nature. 2011;480:259–263. doi: 10.1038/nature10547. [DOI] [PubMed] [Google Scholar]
  14. Elbashir SM, Lendeckel W, Tuschl T. RNA interference is mediated by 21- and 22-nucleotide RNAs. Genes Dev. 2001a;15:188–200. doi: 10.1101/gad.862301. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Elbashir SM, Martinez J, Patkaniowska A, Lendeckel W, Tuschl T. Functional anatomy of siRNAs for mediating efficient RNAi in Drosophila melanogaster embryo lysate. EMBO J. 2001b;20:6877–6888. doi: 10.1093/emboj/20.23.6877. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Elkayam E, Kuhn CD, Tocilj A, Haase AD, Greene EM, Hannon GJ, Joshua-Tor L. The structure of human argonaute-2 in complex with miR-20a. Cell. 2012;150:100–110. doi: 10.1016/j.cell.2012.05.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Frank F, Hauver J, Sonenberg N, Nagar B. Arabidopsis Argonaute MID domains use their nucleotide specificity loop to sort small RNAs. EMBO J. 2012;31:3588–3595. doi: 10.1038/emboj.2012.204. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Frank F, Sonenberg N, Nagar B. Structural basis for 5′-nucleotide base-specific recognition of guide RNA by human AGO2. Nature. 2010;465:818–822. doi: 10.1038/nature09039. [DOI] [PubMed] [Google Scholar]
  19. Friedlander MR, Adamidi C, Han T, Lebedeva S, Isenbarger TA, Hirst M, Marra M, Nusbaum C, Lee WL, Jenkin JC, Sanchez Alvarado A, Kim JK, Rajewsky N. High-resolution profiling and discovery of planarian small RNAs. Proc Natl Acad Sci U S A. 2009;106:11546–11551. doi: 10.1073/pnas.0905222106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Grimson A, Farh KK, Johnston WK, Garrett-Engele P, Lim LP, Bartel DP. MicroRNA targeting specificity in mammals: determinants beyond seed pairing. Mol Cell. 2007;27:91–105. doi: 10.1016/j.molcel.2007.06.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Grimson A, Srivastava M, Fahey B, Woodcroft BJ, Chiang HR, King N, Degnan BM, Rokhsar DS, Bartel DP. Early origins and evolution of microRNAs and Piwi-interacting RNAs in animals. Nature. 2008;455:1193–1197. doi: 10.1038/nature07415. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Gunawardane LS, Saito K, Nishida KM, Miyoshi K, Kawamura Y, Nagami T, Siomi H, Siomi MC. A slicer-mediated mechanism for repeat-associated siRNA 5′ end formation in Drosophila. Science. 2007;315:1587–1590. doi: 10.1126/science.1140494. [DOI] [PubMed] [Google Scholar]
  23. Haley B, Zamore PD. Kinetic analysis of the RNAi enzyme complex. Nat Struct Mol Biol. 2004;11:599–606. doi: 10.1038/nsmb780. [DOI] [PubMed] [Google Scholar]
  24. Houwing S, Berezikov E, Ketting RF. Zili is required for germ cell differentiation and meiosis in zebrafish. EMBO J. 2008;27:2702–2711. doi: 10.1038/emboj.2008.204. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Huang H, Li Y, Szulwach KE, Zhang G, Jin P, Chen D. AGO3 Slicer activity regulates mitochondria-nuage localization of Armitage and piRNA amplification. J Cell Biol. 2014;206:217–230. doi: 10.1083/jcb.201401002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Izumi N, Kawaoka S, Yasuhara S, Suzuki Y, Sugano S, Katsuma S, Tomari Y. Hsp90 facilitates accurate loading of precursor piRNAs into PIWI proteins. RNA. 2013;19:896–901. doi: 10.1261/rna.037200.112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Kawaoka S, Hayashi N, Suzuki Y, Abe H, Sugano S, Tomari Y, Shimada T, Katsuma S. The Bombyx ovary-derived cell line endogenously expresses PIWI/PIWI-interacting RNA complexes. RNA. 2009;15:1258–1264. doi: 10.1261/rna.1452209. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Kawaoka S, Izumi N, Katsuma S, Tomari Y. 3′ end formation of PIWI-interacting RNAs in vitro. Mol Cell. 2011;43:1015–1022. doi: 10.1016/j.molcel.2011.07.029. [DOI] [PubMed] [Google Scholar]
  29. Klattenhoff C, Xi H, Li C, Lee S, Xu J, Khurana JS, Zhang F, Schultz N, Koppetsch BS, Nowosielska A, Seitz H, Zamore PD, Weng Z, Theurkauf WE. The Drosophila HP1 homolog Rhino is required for transposon silencing and piRNA production by dual-strand clusters. Cell. 2009;138:1137–1149. doi: 10.1016/j.cell.2009.07.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Lambert NJ, Gu SG, Zahler AM. The conformation of microRNA seed regions in native microRNPs is prearranged for presentation to mRNA targets. Nucleic Acids Res. 2011;39:4827–4835. doi: 10.1093/nar/gkr077. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Lau NC, Ohsumi T, Borowsky M, Kingston RE, Blower MD. Systematic and single cell analysis of Xenopus Piwi-interacting RNAs and Xiwi. EMBO J. 2009;28:2945–2958. doi: 10.1038/emboj.2009.237. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Lewis BP, Burge CB, Bartel DP. Conserved seed pairing, often flanked by adenosines, indicates that thousands of human genes are microRNA targets. Cell. 2005;120:15–20. doi: 10.1016/j.cell.2004.12.035. [DOI] [PubMed] [Google Scholar]
  33. Lewis BP, Shih IH, Jones-Rhoades MW, Bartel DP, Burge CB. Prediction of mammalian microRNA targets. Cell. 2003;115:787–798. doi: 10.1016/s0092-8674(03)01018-3. [DOI] [PubMed] [Google Scholar]
  34. Li C, Vagin VV, Lee S, Xu J, Ma S, Xi H, Seitz H, Horwich MD, Syrzycka M, Honda BM, Kittler EL, Zapp ML, Klattenhoff C, Schulz N, Theurkauf WE, Weng Z, Zamore PD. Collapse of germline piRNAs in the absence of Argonaute3 reveals somatic piRNAs in flies. Cell. 2009;137:509–521. doi: 10.1016/j.cell.2009.04.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Liu J, Carmell MA, Rivas FV, Marsden CG, Thomson JM, Song JJ, Hammond SM, Joshua-Tor L, Hannon GJ. Argonaute2 is the catalytic engine of mammalian RNAi. Science. 2004;305:1437–1441. doi: 10.1126/science.1102513. [DOI] [PubMed] [Google Scholar]
  36. Ma JB, Yuan YR, Meister G, Pei Y, Tuschl T, Patel DJ. Structural basis for 5′-end-specific recognition of guide RNA by the A. fulgidus Piwi protein. Nature. 2005;434:666–670. doi: 10.1038/nature03514. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Parker JS, Parizotto EA, Wang M, Roe SM, Barford D. Enhancement of the seed-target recognition step in RNA silencing by a PIWI/MID domain protein. Mol Cell. 2009;33:204–214. doi: 10.1016/j.molcel.2008.12.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Parker JS, Roe SM, Barford D. Structural insights into mRNA recognition from a PIWI domain-siRNA guide complex. Nature. 2005;434:663–666. doi: 10.1038/nature03462. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Reuter M, Berninger P, Chuma S, Shah H, Hosokawa M, Funaya C, Antony C, Sachidanandam R, Pillai RS. Miwi catalysis is required for piRNA amplification-independent LINE1 transposon silencing. Nature. 2011;480:264–267. doi: 10.1038/nature10672. [DOI] [PubMed] [Google Scholar]
  40. Rivas FV, Tolia NH, Song JJ, Aragon JP, Liu J, Hannon GJ, Joshua-Tor L. Purified Argonaute2 and an siRNA form recombinant human RISC. Nat Struct Mol Biol. 2005;12:340–349. doi: 10.1038/nsmb918. [DOI] [PubMed] [Google Scholar]
  41. Wang Y, Juranek S, Li H, Sheng G, Wardle GS, Tuschl T, Patel DJ. Nucleation, propagation and cleavage of target RNAs in Ago silencing complexes. Nature. 2009;461:754–761. doi: 10.1038/nature08434. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Wang Y, Sheng G, Juranek S, Tuschl T, Patel DJ. Structure of the guide-strand-containing argonaute silencing complex. Nature. 2008;456:209–213. doi: 10.1038/nature07315. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Wee LM, Flores-Jasso CF, Salomon WE, Zamore PD. Argonaute divides its RNA guide into domains with distinct functions and RNA-binding properties. Cell. 2012;151:1055–1067. doi: 10.1016/j.cell.2012.10.036. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Yuan YR, Pei Y, Ma JB, Kuryavyi V, Zhadina M, Meister G, Chen HY, Dauter Z, Tuschl T, Patel DJ. Crystal structure of A. aeolicus argonaute, a site-specific DNA-guided endoribonuclease, provides insights into RISC-mediated mRNA cleavage. Mol Cell. 2005;19:405–419. doi: 10.1016/j.molcel.2005.07.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Zhang Z, Xu J, Koppetsch BS, Wang J, Tipping C, Ma S, Weng Z, Theurkauf WE, Zamore PD. Heterotypic piRNA Ping-Pong requires qin, a protein with both E3 ligase and Tudor domains. Mol Cell. 2011;44:572–584. doi: 10.1016/j.molcel.2011.10.011. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Data
NIHMS663586-supplement-Supplemental_Data.pdf (3.1MB, pdf)
Supplemental Information
NIHMS663586-supplement-Supplemental_Information.docx (67.5KB, docx)

RESOURCES