Environmentally-induced epigenetic conversion of a piRNA cluster

Abstract

Transposable element (TE) activity is repressed in animal gonads by PIWI-interacting RNAs (piRNAs) produced by piRNA clusters. Current models in flies propose that germinal piRNA clusters are functionally defined by the maternal inheritance of piRNAs produced during the previous generation. Taking advantage of an inactive, but ready to go, cluster of P-element derived transgene insertions in Drosophila melanogaster, we show here that raising flies at high temperature (29°C) instead of 25°C triggers the stable conversion of this locus from inactive into actively producing functional piRNAs. The increase of antisense transcripts from the cluster at 29°C combined with the requirement of transcription of euchromatic homologous sequences, suggests a role of double stranded RNA in the production of de novo piRNAs. This report describes the first case of the establishment of an active piRNA cluster by environmental changes in the absence of maternal inheritance of homologous piRNAs.

Editorial note: This article has been through an editorial process in which the authors decide how to respond to the issues raised during peer review. The Reviewing Editor's assessment is that all the issues have been addressed (see decision letter).

Keywords: D. melanogaster; epigenetics; genetics; genomics; germline; heterochromatin; non coding small RNA; temperature; transposable element.

Figure 1.. Functional assay of the BX2 epigenetic state.

Females carrying either one of the BX2 epialleles and P(TARGET)^GS were analyzed. (A) When BX2 is OFF for production of piRNAs (BX2^OFF), no repression of P(TARGET)^GS occurs, allowing expression of ß-Galactosidase in both germline and in somatic lineages in ovaries (ß-Galactosidase staining). BX2^OFF is illustrated by a blue fly. (B) When BX2 is ON for production of piRNAs (BX2^ON), repression of P(TARGET)^GS occurs only in the germline lineage. BX2^ON is illustrated by a light brown fly. (C) Drawing of an intermediate egg chamber showing germ cells (nurse cells and oocyte in blue) surrounded by somatic follicular cells (in pink), adapted from Figure 1A from Frydman and Spradling (2001). (D) At 25°C, BX2^OFF and BX2^ON are stable over generations. (E) At 29°C, BX2^OFF can be converted into BX2^ON, while BX2^ON is stable over generations.

Figure 1—figure supplement 1.. Schematic maps, drawn at scale, of the different transgenes used in this study.

BX2 is a cluster of seven P(lac)W insertion in tandem. P(TARGET) corresponds to euchromatic transgenes used as the reporter. The common transcribed sequence between BX2 and P(TARGET) transgenes is 3547 bp long, from the transcription start at position 87 in P5’ to the end of lacZ sequence. Hsp70 sequences corresponds to transcription termination signal and are different between P(lac)W and P(TARGET) transgenes. A very few small RNAs matched these Hsp70 sequences, and their number is not affected by temperature. Black arrows indicate transcription orientation. Upper scale is in kb.

Figure 1—figure supplement 2.. Maternal (A) and paramutagenic (B) effect of BX2^ON lines.

(A) Reciprocal crosses were performed at 25°C between BX2^ON, P(TARGET)^GS flies (either BX2^Θ or BX2*) and flies carrying a balancer of the second chromosome (Cy). Lines were established with G1 individuals of maternal or paternal inheritance (named MI or PI, respectively) that carry the BX2, P(TARGET)^GS chromosome, associated with white⁺ phenotype conferred by BX2 transgenes, over Cy and maintained at 25°C. Silencing capacities of these lines were tested over generations by ovarian ß-Galactosidase staining (Supplementary file 3). (B) The capacity of maternally inherited BX2^ON piRNAs to activate a BX2^OFF cluster was tested as shown on the mating scheme. BX2^ON, P(TARGET)^GS females (either BX2^Θ or BX2*) were crossed at 25°C with BX2^OFF, P(TARGET)^GS males and lines were established with G1 individuals which maternally inherited piRNAs and paternally inherited the BX2 cluster. Lines were distinguished thanks to two distinct balancers, Cy and CyRoi that are devoid of transgenes and carrying different dominant markers. These lines were maintained at 25°C and their silencing capacities were tested over generations by intra-strain ovarian ß-Galactosidase staining (Supplementary file 4).

Figure 1—figure supplement 3.. Dynamic of conversion across generations at 29°C.

(A) Fraction of repression across generations for eight independent BX2, P(TARGET)^GS lines incubated at 29°C. The fraction of repressed egg chambers was determined at each generation (Supplementary file 6). BX2 conversion occurred in each line tested with various dynamics: one of them appears to be homogeneous and stable from the 12^th generation (1B) while others still fluctuate. 2A subline was lost at the 18th generation. (B) Mean repression and 95% confidence interval of repression frequency for the eight independent lines. Frequencies were modified (arcsine square root) prior to calculate means and confidence intervals. Data were then back transformed (sine-squared) for graphical representation.

Figure 2.. BX2^Θ and BX2* produce piRNAs and are enriched in H3K9me3.

(A) Size distribution of ovarian small RNAs matching BX2 transgene sequences reveals that both BX2^Θ and BX2* but not BX2^OFF produce 21 nt siRNAs and 23–29-nt piRNAs (upper panels, pBR = backbone plasmid pBR322). Positive and negative values correspond to sense (red) and antisense (blue) reads, respectively. Unique 23–29 nt mappers are shown on the BX2 transgene sequences (lower panels). (B) Percentage of 23–29 nt small RNAs from BX2^Θ and BX2* matching transgene sequence with a U at the first position are shown. n.d.: not determined due to low number of reads. (C) Relative frequency (z-score) of overlapping sense-antisense 23–29 nt RNA pairs reveals an enrichment of 10 nt overlapping corresponding to the ping-pong signature. (D) H3K9me3 and (E) Rhino binding on the BX2 transgene in ovaries of BX2^OFF, BX2^Θ and BX2* strains revealed by chromatin immunoprecipitation (ChIP) followed by quantitative PCR (qPCR) on specific white sequences. In both ‘ON’ strains, BX2^Θ and BX2*, H3K9me3 and Rhino levels over the transgene are very similar and higher than in the BX2^OFF strain (unpaired t-test was used to calculate significance of the differences (p<0.05, n = 5).

Figure 2—figure supplement 1.. 42AB piRNA production is not altered in different BX2 epigenetic contexts.

(A) The size distribution and alignments of unique reads that match the 42AB piRNA cluster show that all genotypes produce similar profile of small RNAs corresponding to 42AB sequence (upper panels). Unique 23–29 nt mappers are shown on the 42AB sequence (lower panels). Positive and negative values correspond to sense (red) and antisense (blue) reads, respectively. (B) Percentage of 1U-containing small RNAs from each genotype that match 42AB sequence. (C) Relative frequency (z-score) of overlapping sense-antisense 23–29 nt RNA pairs reveals an enrichment of 10-nucleotides overlapping small RNAs, which corresponds to the ping-pong signature.

Figure 2—figure supplement 2.. The AGO1 gene region does not produce piRNA by itself.

Alignments of unique 23–29 nt reads that match the AGO1 gene region (from nucleotide 13943387 to 13958089 on the 2R chromosome) show that all genotypes produce no significant levels of piRNAs. Compare this result to that of piRNAs matching BX2 sequences (Figure 2, same y-axis scale). Positive and negative values correspond to sense (red) and antisense (blue) reads, respectively. This observation demonstrates that the AGO1 gene region is not a natural piRNA producing locus.

Figure 2—figure supplement 3.. Mapping the genome for regions stably producing new piRNAs.

(A) Unique piRNAs matching Drosophila chromosomes were kept only if they correspond to specific positions in BX2^Θ compared to BX2^OFF. The number of reads per specific positions was then resampled per windows of 50 kilobases. Background noise was reduced by considering only regions that match more than five reads per kilobase on average on both libraries. Log2 ratio of the number of specific reads from BX2^Θ as compared to those from BX2^OFF was plotted along chromosomes (50 kb window). Only one region on the X chromosome presents a high increase (log2 ratio >8.5, marked by an asterisk). (B) These small RNAs match exons of the white gene and correspond to piRNAs produced by P(lacW) transgenes corresponding to the BX2^Θ locus. No other regions than BX2 stably produce new piRNAs.

Figure 3.. BX2 conversion at 29°C occurs in one generation at a low rate.

(A) G0 females carrying the P(TARGET)^GS reporter and BX2^OFF laid eggs at 25°C during three days. The BX2^OFF state of these females was confirmed after the three days at 25°C by ß-Galactosidase staining (number of egg chambers ≥ 1200). (B) Their eggs were allowed to develop at 29°C until emergence of the next generation. G1 females (n = 181) were individually mated with two siblings and left to lay for three days at 25°C. G1 females were then individually stained for ß-Galactosidase expression. Strikingly, 130 females (71.8%) show ß-Galactosidase repression in some egg chambers (586 among ≈21700 - estimation of the total egg chamber number among 181 females). The BX2^OFF into BX2^ON conversion frequency is ≈2.7%. (C) Analysis of each G1 female progeny developed at 25°C by ß-Galactosidase staining. The progeny of the 51 G1 females that did not present repression maintained BX2^OFF state (n flies = 342). Most of the progeny of the 130 G1 females presenting conversion show no repression (97.2%, n flies = 1488) while 41 females present partial (n = 24) or complete (n = 17) repression of the germline expression of ß-Galactosidase.

Figure 3—figure supplement 1.. No BX2 conversion at 29°C without the P(TARGET)^GS reporter transgene.

This experiment is similar to the one in Figure 3, except that P(TARGET)^GS is absent in these flies incubated at 29°C. (A) The BX2^OFF state of G0 females was confirmed by staining the progeny of a cross between females from this strain and P(TARGET)^GS males. G0 females maintained at 25°C lay eggs at 25°C during three days. (B) Eggs were then transferred at 29°C until emergence of the next generation. 157 G1 females were individually mated with two siblings and left to lay for three days at 25°C. It was not possible to know the epigenetic state of BX2 in these females because of the absence of a reporter transgene. (C) Analysis of the progeny of G1 females reveals that only one female among 1137 presents a partial repression (only one ovariole). Complete repression was never observed.

Figure 3—figure supplement 2.. No BX2 conversion at 29°C with a transgene that is not expressed in the germline.

This experiment is similar to the one in Figure 3, except that P(TARGET)^GS has been replaced by P(TARGET)^S expressed only in the somatic cells surrounding the germ cells. (A) The BX2^OFF state of G0 females was confirmed by staining the progeny of a cross between females from this strain and P(TARGET)^GS males. G0 females maintained at 25°C lay eggs at 25°C during three days. (B) Eggs were then transferred at 29°C until emergence of the next generation. 163 G1 females were mated en masse with males bearing P(TARGET)^GS and left to lay for three days at 25°C. It was not possible to know the epigenetic state of BX2 in these females because of the absence of a germinally-expressed reporter transgene. (C) Analysis of the progeny of G1 females reveals that no repression was observed among 784 analyzed flies (≈94080 estimated egg chambers).

Figure 3—figure supplement 3.. BX2^ON silencing depends on moonshiner.

ß-Galactosidase staining of ovaries from females harboring BX2^ON, P(TARGET)^G, the nosGAL4 driver and a RNAi against the white gene as a control (A) or a RNAi against moonshiner (moon^PA61 and moon^PA62, B and C, respectively). Both of the moon knockdown lines affect the silencing induces by BX2^ON showing that moonshiner is required for the BX2^ON state.

Figure 4.. BX2^OFF antisense RNA increase at 29°C.

(A) RT-qPCR experiments revealed that the steady-state level of ovarian lacZ RNAs from BX2 is more abundant at 29°C (n = 5) compared to 25°C (n = 6). (B) Sense-specific RT-qPCR experiments revealed that only antisense transcripts from BX2, corresponding to antisense lacZ transcripts, are increased (25°C n = 6, 29°C n = 4). Significant p-values are given (bilateral Student's t-test). ns: not significant. (C) Map of the BX2 locus containing seven P(lacW) transgenes inserted into the AGO1 gene. P(lacW) and AGO1 are drawn to scale. The lacZ gene contained in P(lacW) and AGO1 are transcriptionally in opposite directions. Black arrows show lacZ primers used for (A) and (B) experiments. White arrows show primers used for sense-specific reverse transcription in (B) experiment. Grey arrows show AGO1 primers used for (D) experiment: these primers are specific for AGO1 transcripts (RA, RC and RD) that originate from promoters located upstream the BX2 insertion point and, thus, are potentially convergent to BX2. Red arrows show primers used to measure steady-state of all AGO1 isoforms (see Figure 4—figure supplement 1D). (D) RT-qPCR experiments performed on flies devoid of P(lacW) transgenes (w¹¹¹⁸ context) revealed that the steady-state level of ovarian AGO1-RA, -RC and -RD RNA isoforms is more abundant at 29°C (n = 5) compared to 25°C (n = 6). (E–I) To compare small RNAs at 25 versus 29°C, total RNAs were extracted from BX2^OFF ovaries dissected from adults incubated at 25°C or 29°C. Three samples were tested for each temperature. Small RNAs from 18 to 30 nt were deep sequenced. For each library, normalization has been performed for 1 million reads matching the Drosophila genome (rpm, Supplementary file 7). Size distributions of unique reads that match reference sequences are given. (E) Small RNAs matching the Drosophila genome present similar profiles in both temperatures except for 22 nt RNAs that are more represented at 25°C. (F) The 21 to 25 nt reads matching the 42AB piRNA cluster that range from 21 to 25 nt are slightly more abundant at 25°C. (G) Strikingly, almost only 21 nt RNAs match BX2 sequence. They are equally distributed among sense (H) and antisense (I) sequences at both temperatures. (J) No small RNAs corresponding to the AGO1 gene can be detected whatever the temperature. *=p < 0.05, bilateral Student's t-test.

Figure 4—figure supplement 1.. Complementary RNA steady-state measurements by quantitative RT-PCR.

(A, B) Measurement of AGO1 RNA at 25°C and 29°C in different genetic contexts. (A) BX2^OFF flies at 25°C (n = 6) and 29°C (n = 4). As in the w¹¹¹⁸ background experiment (shown in Figure 4D), the amount of AGO1 transcripts is increased at 29°C. (B) Same as (A) except that RNA was extracted from flies carrying BX2^OFF and P(TARGET)^GS. Again, the amount of AGO1 transcripts is increased at 29°C (n = 4) compared to 25°C (n = 4). (C) The amount of P(TARGET)^GS transcripts is decreased at 29°C compared to 25°C. RT-qPCR experiments performed on total RNA extracted from ovaries of flies containing only P(TARGET)^GS transgene in w¹¹¹⁸ context revealed that the steady-state level of lacZ RNA from P(TARGET)^GS is less abundant at 29°C (n = 6) compared to 25°C (n = 3). (D) Total steady-state AGO1 RNA was measured in w¹¹¹⁸ context at 25°C (n = 6) and 29°C (n = 5) using a set of primers located downstream the BX2 insertion point (represented as red arrows on Figure 4C). A slight but non significant difference is observed. p-value are given (bilateral Student's t-test).

Figure 5.. Model of BX2 activation at 29°C.

(A) At 25°C, a low bidirectional transcription of the BX2 cluster leads to a small production of 21 nt RNAs. BX2^OFF is stable at 25°C, no conversion event is observed and the lacZ reporter gene from P(TARGET)^GS remains active throughout generations. (B) At 29°C, a specific increase in antisense transcription occurs, presumably due to a higher activity of the promoter of the AGO1 gene (orange box). This excess of BX2 antisense RNA could interact with sense P(TARGET)^GS transcripts to produce double stranded RNA. Through a yet unknown mechanism, such dsRNA could lead to the formation of de novo BX2 piRNAs. These piRNAs could in turn trigger the conversion of BX2^OFF into an active piRNA cluster, a phenomenon observed based on the repression of the lacZ reporter gene of P(TARGET)^GS. The BX2 conversion is a rare event (≈2% per generation) but once achieved, BX2^ON remains active throughout generations due to the maternal inheritance of homologous piRNAs and the paramutation of the paternal BX2^OFF allele.