A Single Mechanism of Biogenesis, Initiated and Directed by PIWI Proteins, Explains piRNA Production in Most Animals

Abstract

In animals, PIWI-interacting RNAs (piRNAs) guide PIWI proteins to silence transposons and regulate gene expression. The mechanisms for making piRNAs have been proposed to differ among cell types, tissues, and animals. Our data instead suggest a single model that explains piRNA production in most animals. piRNAs initiate piRNA production by guiding PIWI proteins to slice precursor transcripts. Next, PIWI proteins direct the stepwise fragmentation of the sliced precursor transcripts, yielding tail-to-head strings of phased precursor piRNAs (pre-piRNAs). Our analyses detect evidence for this piRNA biogenesis strategy across an evolutionarily broad range of animals, including humans. Thus, PIWI proteins initiate and sustain piRNA biogenesis by the same mechanism in species whose last common ancestor predates the branching of most animal lineages. The unified model places PIWI-clade Argonautes at the center of piRNA biology and suggests that the ancestral animal-the Urmetazoan-used PIWI proteins both to generate piRNA guides and to execute piRNA function.

Keywords: Argonaute; PIWI; Piwi-interacting RNA; flies; mice; piRNA; small RNA evolution; small silencing RNA; spermatogenesis.

Figure 1.. The Current Model for Primary piRNA Biogenesis in Post-Natal Mouse Testis and the Unified Model for piRNA Biogenesis in Animals

Figure 2.. Evidence for Ping-Pong and Phased Pre-piRNAs among Animals

Cladogram of a representative set of 35 animal species showing the approximate time of each divergence. piRNA length profiles are shown for all sequencing reads taking into account their abundance; distance probability analyses are for ≥24-nt sequencing reads without taking into account abundance. Data for piRNA 5′-to- 5′ distance was processed using non-parametric regression (LOWESS), and the first peak of the smoothed data was used to estimate pre-piRNA length. The significance of ping-pong (peak at 10 for 5′-to-5′ distance on opposite genomic strands) and phasing (peak at 0 for 3′-to-5′ distance on the same genomic strand) signatures was assessed using Z-scores. Red: species with Z0 > 1.96 (p < 0.05) for 3′-to-5′ distance. Orange: species for which autocorrelation analysis detected periodic peaks of 5′-to- 5′ piRNA distances. All data are from wild-type animals, except Mus musculus (Pnldc1^−/−).

Figure 3.. Mature piRNAs are Replaced by Untrimmed Pre-piRNAs in Pnldc1^−/− Mice

(A) Male germ cell small RNA length profiles and, for ≥24-nt small RNAs, the probability of 3′-to-5′ distances on the same genomic strand and nucleotide composition of the genomic neighborhood of small RNA 3′ ends in wild-type and Pnldc1^−/− mice. Data are from a single representative biological sample excluding reads with non-templated 3′ nucleotides. C, cell chromosome content; N, ploidy. (B) At left, mean (± standard deviation; n = 3) steady-state molecular abundance of Mili, Miwi, and Pnldc1 mRNAs in male germ cells purified from wild-type and Pnldc1^−/− mice. Spg, spermatogonia; SpI, primary spermatocytes; SpII, secondary spermatocytes; RS, round spermatids. At right, representative Western blot images and relative mean (± standard deviation; n = 3) steady-state abundance of MILI and MIWI proteins. Each lane contained lysate from ~11,000 cells. Figures S2C and S2D show uncropped Western blot images.

Figure 4.. MILI and MIWI Participate in Phased Pre-piRNA Production

(A) Length profiles of MILI- and MIWI-bound piRNAs in wild-type and pre-piRNAs in Pnldc1^−/− primary spermatocytes. Abundance was normalized to all genome mapping reads and reported in parts per million (ppm). Data are from a single representative biological sample; reads with non-templated 3′ nucleotides were excluded. (B) Probability of distances between the 5′ ends of MILI- and MIWI-bound piRNAs. Numbers indicate the total frequency of 5′ or 3′ ends of MILI-bound small RNAs residing before, after, or coinciding with the 5′ or 3′ ends of the MIWI-bound small RNAs. Data are from a single representative biological sample. (C) Distance between the most frequent 3′ end of the MILI-bound pre-RNA group and the most frequent 3′ end of the corresponding paired MIWI-bound pre-RNA group. Data are from a single representative biological sample.

Figure 5.. PIWI Protein Identity and the Availability of Uridines Dictate the Position of Mouse Pre-piRNA 3′ Ends

At left, a comparison of the most frequent 3′ ends of individual MILI- and MIWI- bound pre-piRNAs sharing the same 5′ end in Pnldc1^−/− primary spermatocytes. Data are in parts per million (ppm) for pre-piRNAs derived from pachytene piRNA loci. At right, the nucleotide bias of the genomic neighborhood around the most frequent 3′ ends of paired MILI- and MIWI-bound pre-piRNA groups in Pnldc1^−/− primary spermatocytes. Data are from a single representative biological sample.

Figure 6.. PIWI Protein Identity and the Availability of Uridines Dictate the Position of Drosophila melanogaster Pre-piRNA 3′ Ends

Nucleotide bias of the genomic neighborhood around the most frequent 3′ ends of paired Piwi- and Aub-bound piRNA groups from wild-type (A), and pre-piRNA groups in papi^−/− (B) and nibbler^−/− (C) D. melanogaster ovaries. Data are for all unambiguously mapping RNAs from a single representative biological sample.

Figure 7.. piRNA-directed, PIWI Protein-Catalyzed Slicing Initiates piRNA Biogenesis in Most Animals

(A) Strategy for the identification of putative pre-pre-piRNAs among ≥150-nt long, 5′ monophosphorylated RNAs derived from pachytene piRNA loci from wild-type mouse primary spermatocytes. RNAs sharing 5′ ends with MILI- or MIWI-bound mature piRNAs were classified as putative pre-pre-piRNAs; the remaining long RNAs served as the control group. Data in (B) and (C) are from a single representative biological sample. (B) Nucleotide bias of the 5′ ends of putative pre-pre-piRNAs compared to the control group. (C) Probability of distances between the 5′ ends of MILI- or MIWI-bound piRNAs and the 5′ ends of putative pre-pre-piRNAs or control RNAs complementary to nucleotides 2 to 10 of the piRNA (g2–g10). (D) Percent of piRNAs explaining the 5′ ends of either putative pre-pre-piRNAs or control RNAs. Nucleotides 2 to 10 of guide piRNAs (g2–g10) were required to be complementary to the target long RNAs. Data are for long RNAs and MILI- or MIWI- bound piRNAs derived from pachytene piRNA loci. All possible pairwise combinations of data sets from two biological replicates were used to calculate medians. Whiskers correspond to minimum and maximum values. Wilcoxon rank- sum test was used to assess statistical significance. (E) Strategy to identify putative secondary and primary piRNAs, and to measure the probability of distances from the 5′ ends of putative secondary piRNAs to the 5′ ends of putative primary piRNAs. Distance probability analyses are for ≥24-nt sequencing reads without taking into account abundance. Data was processed using non- parametric regression (LOWESS).