Paramutation in Drosophila Requires Both Nuclear and Cytoplasmic Actors of the piRNA Pathway and Induces Cis-spreading of piRNA Production

Abstract

Transposable element activity is repressed in the germline in animals by PIWI-interacting RNAs (piRNAs), a class of small RNAs produced by genomic loci mostly composed of TE sequences. The mechanism of induction of piRNA production by these loci is still enigmatic. We have shown that, in Drosophila melanogaster, a cluster of tandemly repeated P-lacZ-white transgenes can be activated for piRNA production by maternal inheritance of a cytoplasm containing homologous piRNAs. This activated state is stably transmitted over generations and allows trans-silencing of a homologous transgenic target in the female germline. Such an epigenetic conversion displays the functional characteristics of a paramutation, i.e., a heritable epigenetic modification of one allele by the other. We report here that piRNA production and trans-silencing capacities of the paramutated cluster depend on the function of the rhino, cutoff, and zucchini genes involved in primary piRNA biogenesis in the germline, as well as on that of the aubergine gene implicated in the ping-pong piRNA amplification step. The 21-nt RNAs, which are produced by the paramutated cluster, in addition to 23- to 28-nt piRNAs are not necessary for paramutation to occur. Production of these 21-nt RNAs requires Dicer-2 but also all the piRNA genes tested. Moreover, cytoplasmic transmission of piRNAs homologous to only a subregion of the transgenic locus can generate a strong paramutated locus that produces piRNAs along the whole length of the transgenes. Finally, we observed that maternally inherited transgenic small RNAs can also impact transgene expression in the soma. In conclusion, paramutation involves both nuclear (Rhino, Cutoff) and cytoplasmic (Aubergine, Zucchini) actors of the piRNA pathway. In addition, since it is observed between nonfully homologous loci located on different chromosomes, paramutation may play a crucial role in epigenome shaping in Drosophila natural populations.

Keywords: Drosophila; gene regulation; mobile DNA; noncoding small RNAs; trans-generational epigenetics.

Figure 1

Effect of mutations affecting the piRNA or siRNA pathways on trans-silencing and small RNA production capacities of a BX2* cluster. Upper box: Mating scheme used to analyze the effect of mutations (mut) affecting aubergine, Dicer-2, rhino, cutoff, and zucchini on BX2* ovarian small RNA production and trans-silencing capacities. As the BX2 cluster, all genes are located on chromosome 2. (Left) Heterozygous BX2* aub⁻, BX2* Dcr-2⁻, BX2* rhi⁻, BX2* cuff ⁻, or BX2* zuc⁻ females were crossed with heterozygous males carrying a mutant allele of the same gene to generate either heterozygous control or loss-of-function females for deep sequencing of ovarian small RNAs. (Right) The same heterozygous females were crossed with heterozygous males carrying a mutant allele of the same gene plus a P-lacZ target transgene (BQ16) to measure the effect of the mutation tested on BX2* trans-silencing capacities (TSE %). TSE was quantified by determining the percentage of egg chambers with no lacZ expression in the germline. In the genotypes presented, maternally inherited chromosomes (chromosomes 2 and 3) are indicated above the bar. Cy is a balancer chromosome carrying a dominant phenotypic marker. (A–L) In each case, the genotype tested is indicated, and the percentage of TSE is given below the genotype with the total number of egg chambers assayed in parentheses. Nonparamutated BX2 (“naive”) and BX2* at generation 83 were analyzed as controls. Histograms show the length distributions of ovarian small RNAs matching the P{lacW} locus. Positive and negative values correspond to sense and antisense reads, respectively. Plots show the abundance of 19- to 29-nt small RNAs matching the P{lacW} locus. Analysis of the effect of aub, rhi, cuff, and zuc mutations shows that BX2* silencing capacities, ovarian P{lacW} 23- to 28-nt piRNA, but also 21-nt RNA production depend on the piRNA pathway. Dcr-2 loss of function strongly affects production of P{lacW} homologous 21-nt RNAs without affecting that of 23- to 28-nt RNAs or trans-silencing capacities. P{lacW} 21 nt are not necessary for the maintenance of the paramutated state and their biogenesis depends on both Dcr-2 and the piRNA pathway. TSE results for Dcr-2 mutants (E and F) are reprinted from de Vanssay et al. (2012).

Figure 1

Effect of mutations affecting the piRNA or siRNA pathways on trans-silencing and small RNA production capacities of a BX2* cluster. Upper box: Mating scheme used to analyze the effect of mutations (mut) affecting aubergine, Dicer-2, rhino, cutoff, and zucchini on BX2* ovarian small RNA production and trans-silencing capacities. As the BX2 cluster, all genes are located on chromosome 2. (Left) Heterozygous BX2* aub⁻, BX2* Dcr-2⁻, BX2* rhi⁻, BX2* cuff ⁻, or BX2* zuc⁻ females were crossed with heterozygous males carrying a mutant allele of the same gene to generate either heterozygous control or loss-of-function females for deep sequencing of ovarian small RNAs. (Right) The same heterozygous females were crossed with heterozygous males carrying a mutant allele of the same gene plus a P-lacZ target transgene (BQ16) to measure the effect of the mutation tested on BX2* trans-silencing capacities (TSE %). TSE was quantified by determining the percentage of egg chambers with no lacZ expression in the germline. In the genotypes presented, maternally inherited chromosomes (chromosomes 2 and 3) are indicated above the bar. Cy is a balancer chromosome carrying a dominant phenotypic marker. (A–L) In each case, the genotype tested is indicated, and the percentage of TSE is given below the genotype with the total number of egg chambers assayed in parentheses. Nonparamutated BX2 (“naive”) and BX2* at generation 83 were analyzed as controls. Histograms show the length distributions of ovarian small RNAs matching the P{lacW} locus. Positive and negative values correspond to sense and antisense reads, respectively. Plots show the abundance of 19- to 29-nt small RNAs matching the P{lacW} locus. Analysis of the effect of aub, rhi, cuff, and zuc mutations shows that BX2* silencing capacities, ovarian P{lacW} 23- to 28-nt piRNA, but also 21-nt RNA production depend on the piRNA pathway. Dcr-2 loss of function strongly affects production of P{lacW} homologous 21-nt RNAs without affecting that of 23- to 28-nt RNAs or trans-silencing capacities. P{lacW} 21 nt are not necessary for the maintenance of the paramutated state and their biogenesis depends on both Dcr-2 and the piRNA pathway. TSE results for Dcr-2 mutants (E and F) are reprinted from de Vanssay et al. (2012).

Figure 2

Paramutation by a cytoplasm devoid of siRNAs. (A) BX2* Dcr-2^−/− females were crossed to BX2 Dcr-2^+/+ (naive) males to generate G₁ females having paternally inherited a BX2 locus and maternally inherited piRNAs but no 21-nt siRNAs homologous to the P{lacW} cluster. In addition, these G₁ females paternally inherited a Dcr-2⁺ allele. These G₁ females were crossed with Dcr-2⁺ males to establish a BX2* Dcr-2⁺/Dcr-2⁺ line in which no maternal P{lacW} siRNAs have been initially introduced in G₀. Cy is a balancer chromosome carrying a dominant phenotypic marker. Maternal chromosomal complement is written above the bar. (B) At generations 0, 1, and 2, deep sequencing of ovarian small RNAs was performed and silencing capacity was controlled in parallel by crossing females with males carrying the BQ16 target transgene and scoring TSE in female progeny. The genotype tested is indicated and TSE percentages and ovarian 19- to 29-nt RNAs are presented as in Figure 1. As expected, G₁ and G₂ females show complete silencing capacities. The same was true for successive Gn generations (G₆, TSE = 100%, n = 450; G₁₀, TSE = 100%, n = 1100). Despite the lack of maternal P{lacW} siRNA maternal transmission in G₀, such ovarian small RNAs appear already in G₁ females and increase in G₂ in BX2* Dcr-2^+/+ females (dashed arrow). Therefore, maternal inheritance of P{lacW} homologous piRNAs, but no siRNAs, results in a strong and stable paramutated BX2* locus and in immediate production of both P{lacW} piRNAs and siRNAs in females carrying a Dcr2⁺ allele.

Figure 3

Paramutation of a BX2 locus by partially homologous piRNAs loci. BX2 males carrying a P{lacW} naive cluster were crossed to females carrying partially homologous P-transgenes inserted in telomeric piRNA-producing loci at the heterozygous state. Female progeny were recovered that carry the BX2 locus but did not carry the telomeric transgenes. These G₁ females inherited a cytoplasm carrying piRNA homologous to the telomeric transgenes, which therefore cover only partially the BX2 P{lacW} transgene sequence. Lines were established and the trans-silencing capacities of these putative BX2* lines were tested in subsequent generations, as in Figure 1 and Figure 2. In addition, deep sequencing of ovarian small RNAs was performed in G₃, G₅, and G₁₀. (A, left and middle) Structure of the P-1152 and RS3 telomeric P-transgenes used and of the BX2 P{lacW} transgenes. Colored thick lines below the transgenes indicate the sequences shared by the P{lacW} and telomeric transgenes. The P-1152 line carries two P-lacZ-rosy (P{lArB}) transgenes inserted in the telomeric associated sequences (TAS) of the X chromosome. RS3 carries a P-FRT-white transgene in the TAS of the 3R chromosomal arm. These two lines induce a strong TSE. (A, right) Ovarian piRNA production in females that carry a maternally inherited P-1152 or RS3 locus at the heterozygous state (G₁ females from crosses between P-1152 and RS3 females with males devoid of transgene) is mapped on the P{lacW} transgene. (B) Three types of replicate BX2* sublines were generated: Sublines that inherited in G₁ (1) a cytoplasm from females heterozygous for the P-1152 locus (^P-1152BX2 sublines, in orange); (2) a cytoplasm from females heterozygous for the RS3 locus (^RS3BX2 sublines, in green); and (3) a cytoplasm from females heterozygous for both the P-1152 and RS3 loci (^P-1152+RS3BX2 sublines, in purple). Trans-silencing capacities of various BX2* lines are shown with regard to generations (TSE %, n > 500 in all assays). Names of replicate sublines are given below the graphs. (C) Deep sequencing of ovarian small RNAs in various BX2* sublines at generations 3, 5, and 10. Plots show the abundance of 19- to 29-nt small RNAs matching P{lacW}. TSE assays show that silencing capacities vary among replicate BX2* lines but strong and stable paramutated lines can be recovered with the three types of G₀ cytoplasmic inheritance (B). Ovarian small RNAs were analyzed for a subset of the BX2* lines showing strong silencing capacities. The name of the line is indicated on the left of the graph. Deep sequencing analysis shows that piRNAs corresponding to the whole length of P{lacW} can be produced as early as generation 3 (C), showing that piRNA production by a paramutated locus can rapidly extend to the part of the locus that did not originally receive maternally inherited piRNAs.

Figure 4

Paramutagenicity of a partially homologous paramutation. BX2* lines having, at G₀, maternally inherited piRNA homologous to a part of the BX2 P{lacW} transgene sequence, and showing strong silencing capacities at the 20th generation (Figure 3), were tested for their ability to de novo paramutate a naive BX2 locus. (A) The name of the line is indicated beside the G₀ female genotype. Maternal chromosomal complement is written above the bar. Cy and CyRoi are balancer chromosomes devoid of transgenes and carrying different dominant markers. These ^P-1152BX2*, ^RS3BX2*, and ^P-1152+RS3BX2* females were crossed with males carrying a BX2 naive cluster and G₁ progeny having paternally, but not maternally, inherited a BX2 cluster were crossed to establish lines. These lines carry a second order paramutated locus. (B) These lines were subsequently tested for trans-silencing capacities at generations 1, 4, and 36. The TSE percentage is indicated with the number of egg chambers counted in parentheses. Strong silencing capacities are observed over generations showing that ^P-1152BX2*, ^RS3BX2*, and ^P-1152+RS3BX2* are strongly paramutagenic.

Figure 5

Somatic impact of maternally transmitted small RNAs. The effect of P{lacW} small RNA maternal inheritance on P{lacW} cluster expression in adult eyes was tested using two different transgene clusters, T-1 (A) and DX1 (B), as a read-out. T-1, but not DX1, has trans-silencing capacities, correlated with the presence of abundant ovarian piRNAs. The G₀ cross is indicated above the picture and was performed at 25°. Green female symbol indicates cytoplasmic inheritance containing P{lacW} piRNAs. The genotype (chromosome 2) of the G₁ females analyzed is given below the picture. Cy and Xa are chromosomes devoid of transgenes and carrying dominant markers. The maternally inherited chromosome is written above the bar. (A and B) Pigment levels (absorbance at 480 nm) are shown beside the pictures. Numbers in abscissa indicate the genotype analyzed (reported directly on the head picture). For pigment level dosages, Canton^y and w¹¹¹⁸ female heads served as WT and w null mutant controls, respectively. Control pictures are boxed in B. Bars indicate the standard error. Comparisons were performed using the Student’s t-test: *P < 0.05; **P < 0.01; ***P < 0001. Maternal inheritance of P{lacW} small RNAs produced by a T-1 female results in increased silencing of both T-1 and DX1 expression in the eye.