A Role for siRNA in X-Chromosome Dosage Compensation in Drosophila melanogaster

Abstract

Sex-chromosome dosage compensation requires selective identification of X chromatin. How this occurs is not fully understood. We show that small interfering RNA (siRNA) mutations enhance the lethality of Drosophila males deficient in X recognition and partially rescue females that inappropriately dosage-compensate. Our findings are consistent with a role for siRNA in selective recognition of X chromatin.

MALES of many species carry a euchromatic, gene-rich X chromosome and a gene-poor, heterochromatic Y chromosome (Charlesworth 1991). This creates a potentially lethal imbalance in the X to autosomal (X:A) ratio in one sex (Gupta et al. 2006; Nguyen and Disteche 2006; Deng et al. 2011). Dosage compensation is an essential process that equalizes X-linked gene expression between XY males and XX females, thereby maintaining a constant ratio of X:A gene products. Strategies to accomplish this differ between species, but share the need for coordinated regulation of an entire chromosome (Lucchesi et al. 2005). In flies, the male-specific lethal (MSL) complex, composed of five MSL proteins and noncoding roX (RNA on the X chromosome) RNA, binds with great selectivity to the X chromosome of males (Deng and Meller 2006a). The MSL complex directs H4K16 acetylation to the body of X-linked genes, increasing transcription by enhancing RNA polymerase II processivity (Smith et al. 2001; Larschan et al. 2011).

Recruitment of the MSL complex is postulated to occur at X-linked chromatin entry sites (CES) (Kelley et al. 1999; Alekseyenko et al. 2008; Straub et al. 2008). CES contain 21-bp MSL recognition elements (MREs), which are modestly enriched on the X chromosome (Alekseyenko et al. 2008). The MSL complex then spreads to nearby transcribed genes (Larschan et al. 2007; Sural et al. 2008). While this model elegantly describes the local distribution of the MSL complex, it fails to explain the exclusive recognition of X chromatin that is a hallmark of Drosophila dosage compensation.

The initiation of dosage compensation and hypertranscription of X-linked genes is dependent on roX RNA (Meller 2003; Deng and Meller 2006b). The X-linked roX genes, roX1 and roX2, are redundant for these functions (Meller and Rattner 2002). Mutation of a single roX gene is without phenotype, but simultaneous mutation of roX1 and roX2 reduces X-localization of the MSL complex, resulting in a reduction in X-linked gene expression and male-specific lethality (Meller and Rattner 2002; Deng and Meller 2006b).

Because the roX RNAs are necessary for exclusive X-localization of the MSL proteins, genetic modifiers of roX1 roX2 lethality may identify novel pathways that contribute to X-recognition. We previously reported that a maternally imprinted Y chromosome is a potent suppressor of roX1 roX2 lethality (Menon and Meller 2009). The expression of Y-linked protein-coding genes is restricted to the germline, making it unlikely that these genes influence the somatic process of dosage compensation. Furthermore, the Y chromosome itself is nonessential for dosage compensation (reviewed by Lucchesi 1973). We postulate that, in spite of the fact that Y-linked genes are unnecessary for dosage compensation, the Y-chromosome imprint modulates a pathway involved in this process.

Repetitive sequences, which are abundant on the Y chromosome, have been proposed to influence somatic gene expression (Lemos et al. 2008, 2010; Jiang et al. 2010; Piergentili 2010). Small RNA pathways are potential mediators of this effect. To pursue the idea that small RNA might play a role in dosage compensation, we conducted a directed screen of RNAi pathways. Mutations in the small interfering RNA (siRNA) pathway were found to enhance roX1 roX2 lethality. siRNA mutations disrupt localization of the MSL complex in roX1 roX2 mutants and partially rescue female flies that inappropriately dosage-compensate, leading to toxic overexpression of X-linked genes. Our findings are consistent with participation of siRNA in recognition of X chromatin.

Materials and Methods

Fly culture and genetics

Flies were maintained at 25° on standard cornmeal–agar fly food. Unless otherwise noted, mutations are described in Lindsley and Zimm (1992). roX1 mutations and a complex roX2 deletion (Df(1)52; [w⁺4Δ4.3]) have been described (Meller et al. 1997; Meller and Rattner 2002; Deng et al. 2005). A viable deletion of roX2 (roX2Δ) was accomplished by FLP-mediated recombination between CG11695^f01356 and nod^f04008. Description of dcr2^f06544, ago2^dop1, ago2⁴¹⁴, r2d2¹, D-elp1^c00296, loqs^f00791, ago1^k00281, spn-E¹, aub^QC42, aub^HN, and piwi⁰⁶⁸⁴³ can be found at http://flybase.org. ago2⁴¹⁴ was provided by R. Carthew, and all other mutations were provided by the Bloomington Drosophila Stock Center.

RNAi mutations were outcrossed for six generations to minimize genetic background effects. All stocks were constructed with the Y chromosome from the laboratory reference yw strain to eliminate confounding effects attributable to different Y chromosomes that we and others have observed (Lemos et al. 2008). After rebalancing, all mutations were confirmed by PCR or phenotype. Matings to determine the effect of RNAi pathway mutations on roX1^ex33 roX2Δ male and yw female survival are detailed in Figure S2.

qRT-PCR

Accumulation of roX1^ex40 transcript was measured by qRT-PCR as previously described (Deng et al. 2009). Briefly, RNA was prepared from three groups of 50 third instar male larvae. One microgram of RNA was reverse-transcribed using random hexamers and ImProm-II reverse transcriptase (Promega). Two technical replicates of each biological replicate were amplified with 300 nM of the primers TTTTTGTCCCACCCGAATAA and CCTTTTAATGCGTTTTCCGA. Expression of roX1^ex40 was normalized to autosomal Dmn, amplified with 300 nM of primers GACAAGTTGAGCCGCCTTAC and CTTGGTGCTTAGATGACGCA.

Results and Discussion

The roX1^ex33 roX2Δ X chromosome supports ∼20% eclosion of adult male escapers. roX1^ex33 roX2Δ females were mated to males heterozygous for mutations in the Piwi-interacting RNA, small interfering RNA (siRNA), and microRNA (miRNA) pathways (RNAi⁻/+). The survival of sons with reduced RNA interference (RNAi) function (roX1^ex33 roX2Δ ; RNAi⁻/+) was divided by that of their brothers with intact RNAi (roX1^ex33 roX2Δ ; +/+) to reveal enhancement or suppression of male lethality. Mutations in Dcr-2, Ago2, Loqs, and D-elp1 were found to lower the survival of roX1^ex33 roX2Δ males by 30, 55, 50, and 70%, respectively (Figure 1A). Dcr-2 and D-elp1 play a role in endogenous siRNA (endo-siRNA) production and transposon silencing, and Ago2 is a member of the RNAi-induced silencing complex (Carthew and Sontheimer 2009; Lipardi and Paterson 2009; Siomi and Siomi 2009). While loqs has a prominent role in miRNA biogenesis, an isoform of Loqs has been implicated in the biogenesis of endo-siRNA from structured loci and transposons (Okamura et al. 2008; Zhou et al. 2009; Marques et al. 2010). All of the candidate genes therefore affect siRNA production or function. Reduction of the canonical siRNA gene r2d2 did not enhance roX1 roX2 male lethality. R2D2 affects strand selection during loading of siRNA onto Ago2 (Liu et al. 2003; Tomari et al. 2004). It is possible that this is unnecessary for dosage compensation or that the level of R2D2 is not limiting when a single copy of the gene is mutated.

Figure 1 .

siRNA mutations enhance roX1 roX2 male lethality. (A) Eclosing roX1^ex33 roX2Δ males carrying RNAi mutations divided by their brothers with full RNAi function. SEM is represented by error bars. An asterisk indicates Student’s two-sample t-test significance of ≤0.05. (B) Ago2 reduction partially rescues the developmental delay of females expressing MSL2. Females carry the [H83M2]6I transgene and express MSL2. Solid bars represent females heterozygous for ago2⁴¹⁴; shaded bars represent females with wild-type ago2. (C) Ago2 reduction does not influence the eclosion of otherwise wild-type females. Solid bars depict females heterozygous for ago2⁴¹⁴; shaded bars are their sisters with wild-type ago2.

To confirm that siRNA selectively affects dosage compensation, we asked whether reduction of Ago2 rescued females that inappropriately deploy the dosage compensation machinery, leading to toxic overexpression of both X chromosomes. Ectopic expression of male-specific lethal 2 (msl2) induces dosage compensation in females (Kelley et al. 1995). MSL2 expression, driven by the [H83M2]6I transgene, reduces female survival and delays the peak of eclosion until day 6 (shaded bars, Figure 1B) (Kelley et al. 1995). In contrast, eclosion of sisters not expressing MSL2 peaks on day 2 (shaded bars, Figure 1C). Eclosion of [H83M2]6I females with one mutated ago2 allele is advanced by 2 days, peaking on day 4 (solid bars, Figure 1B). Reduction of Ago2 in otherwise wild-type females had no discernible effect on eclosion timing (Figure 1C). The enhancement of roX1 roX2 male lethality by siRNA mutations and partial rescue of MSL2-expressing females by reduction of Ago2 identifies a role for small RNA in Drosophila dosage compensation.

The roX1^ex40 internal deletion mutant supports full male survival, presumably because it retains essential 5′ and 3′ roX1 regions in a transcript of reduced size (Deng et al. 2005). Localization of the MSL complex on polytene chromosomes of roX1^ex40 roX2Δ males is similar to that observed in wild-type flies. roX1^ex40 therefore has a molecularly detectable but subphenotypic defect. Loss of Ago2 has no effect on male survival by itself, but when Ago2 is eliminated in roX1^ex40 roX2Δ males, survival is reduced to 8% (Figure 2A). Loss of Loqs reduces roX1^ex40 roX2Δ male survival by >50% (Figure 2B). roX1^ex40 roX2Δ males with reduced D-Elp1 levels have full viability, but D-elp1 lethality precludes homozygote testing. We took advantage of the synthetic lethality between roX1^ex40 roX2Δ and siRNA mutations to explore how siRNA contributes to dosage compensation.

Figure 2 .

roX1^ex40A roX2Δ is synthetic lethal with siRNA mutations. (A) Loss of Ago2 reduces the survival of roX1^ex40 roX2Δ adult males. The number of males recovered was the following: ago2⁴¹⁴, 245; roX1^ex40 roX2Δ, 274; roX1^ex40 roX2Δ; ago2⁴¹⁴/+, 1356; and roX1^ex40 roX2Δ; ago2⁴¹⁴, 45. (B) Loss of Loqs reduces roX1^ex40 roX2Δ adult male survival. The total number of males recovered was the following: loqs^f00791, 230; roX1^ex40 roX2Δ, 274; roX1^ex40 roX2Δ; loqs^f00791/+, 708; and roX1^ex40 roX2Δ; loqs^f00791, 166. Survival of roX1^ex40 roX2Δ; ago2⁴¹⁴ and roX1^ex40 roX2Δ; loqs^f00791 males was determined by mating roX1^ex40 roX2Δ; ago2⁴¹⁴ /TM3SbTb males and females or roX1^ex40 roX2Δ; loqs^f00791/In(2LR)Bc Gla males and females. Survival of ago2⁴¹⁴ and loqs^f00791 males was determined by observation of yw; ago2⁴¹⁴/TM3SbTb and yw; loqs^f00791/In(2LR)Bc Gla stocks.

To address the possibility that siRNA mutations act by modulating the level of roX RNA, quantitative RT-PCR (qRT-PCR) was used to measure roX1^ex40 transcript in ago2⁴¹⁴ or D-elp1^c00296/+ males. Accumulation of roX1^ex40 RNA was unaffected by these mutations (Supporting Information, Figure S1A). We also considered the possibility that siRNA indirectly influences the level of an MSL protein. Protein blotting revealed no reduction in core members of the MSL complex in males lacking Ago2 or with reduced D-elp1 (Figure S1, C–F). This conclusion is supported by whole-genome expression studies in S2 cells following Ago2 knockdown (Rehwinkel et al. 2006). As suggested by the lack of a male phenotype, the roX1^ex40 roX2Δ chromosome alone did not affect MSL protein levels (Figure S1, C–F). Disruption of dosage compensation in roX1 roX2 males with reduced siRNA therefore does not involve reduction in the core components of the MSL complex.

The synthetic lethality between roX1^ex40 roX2Δ and siRNA mutations suggested that siRNA could contribute to X-identification or to recruitment of the MSL complex to the X chromosome. If this is the case, loss of siRNA alone might disrupt MSL localization, which is exclusive to the X chromosome in wild-type males (Figure 3A). Reduction of D-Elp1 did not discernibly affect MSL1 localization to the polytene X chromosome of otherwise wild-type males (Figure 3B). A slight disruption of X-localization was detected in ago2 mutants, but this was only marginally higher than that observed in wild-type controls (Figure 3, B, C, and E; Table S1).

Figure 3 .

MSL1 localization is disrupted in roX1^ex40 roX2Δ males mutated for ago2 or D-elp1. (A) MSL1 localization is exclusive to the X chromosome in a polytene preparation from a wild-type male larva. (B) Percentage of nuclei of each genotype that display wild-type MSL1 recruitment to the X chromosome. (C) Percentage of nuclei with ectopic MSL1 binding at the chromocenter (compare arrowheads in A and D). (D) Minimal MSL1 recruitment to the X chromosome and strong chromocenter recruitment in a roX1^ex40 roX2Δ; ago2⁴¹⁴ male. (E) Percentage of nuclei with minimal or no MSL1 recruitment to the X chromosome (sum of categories “+” and “no MSL recruitment,” Table S1A). (F) Ectopic autosomal MSL1 binding in a roX1^ex40 roX2Δ; D-elp1/+ male. (G) Percentage of nuclei with four or more distinct autosomal MSL1-binding sites (arrowheads in F). (H) Percentage of nuclei with MSL1 recruitment to a telomere (arrows in F). Polytene chromosome preparations were immunostained for MSL1 as previously described (Kelley et al. 1999). MSL1 is detected by Texas Red, and DNA is detected by DAPI. One hundred fifty to 300 nuclei of each genotype were scored for MSL1 recruitment. Genotypes were obscured during scoring to eliminate bias. Full genotypes are the following: yw reference strain (wild type); ago2⁴¹⁴; D-elp1^c00296/+; roX1^ex40 roX2Δ; roX1^ex40 roX2Δ; and ago2⁴¹⁴ (open bars); and roX1^ex40 roX2Δ; D-elp1^c00296/+ (open bars). SEM is depicted by error bars. Categories of MSL1 recruitment are detailed in Table S1.

Ectopic MSL1 binding on the autosomes at the chromocenter and at the telomeres is a sensitive metric for disruption of MSL localization. Although MSL1 recruitment in roX1^ex40 roX2Δ males is superficially similar to wild type, examination of a large number of nuclei revealed a reduction of MSL recruitment to the X chromosome in some nuclei and elevated ectopic localization, particularly at the chromocenter (Figure 3, B and C; Table S1). This supports the idea that roX1^ex40 has a defect in function. However, mislocalization of MSL1 was notably more severe in chromosome preparations from roX1^ex40 roX2Δ; ago2⁴¹⁴ and roX1^ex40 roX2Δ; D-elp1^c00296/+ males. The number of nuclei exhibiting minimal or no recruitment of MSL1 to the X chromosome is enhanced over threefold by the loss or reduction of these siRNA proteins (Figure 3E). These same genotypes displayed a threefold increase in ectopic autosomal MSL1 localization (Figure 3, D, F, and G; Table S1). Despite increased mislocalization of the MSL complex, roX1^ex40 roX2Δ; D-elp1^c00296/+ male viability appears unaffected, and the viability of roX1^ex40 roX2Δ males with reduced levels of Ago2 or Loqs is also high (Figure 2, A and B). It is possible that this disparity is because the accumulation of mutated roX1 transcripts, including roX1^ex40, is lower in the salivary gland than in other tissues (Figure S1B; see figure 3 in Deng et al. 2005). In spite of reduced transcript in the salivary gland, the roX1^ex40A mutant directs considerable X-localization of the MSL complex, in accord with the ability of roX1^ex40 roX2Δ males to tolerate a partial, but not a complete, reduction in RNAi. Taken together, these studies reveal a role for siRNA in the process of dosage compensation in Drosophila. The genetic interaction between mutations affecting siRNA and roX1 roX2 chromosomes, as well as the enhancement of ectopic MSL mislocalization, suggests that siRNA contributes to X recognition or chromatin binding of the MSL complex.

Small RNA has been implicated in numerous chromatin-based processes, but the present study is the first to link small RNA to Drosophila dosage compensation. Small RNA typically acts through gene silencing (Pal-Bhadra et al. 2004; Verdel et al. 2004; Brower-Toland et al. 2007; Wang and Elgin 2011). For example, Ago2 and Dcr2 mutations suppress position-effect variegation in flies, suggesting a function in heterochromatic repression (Deshpande et al. 2005; Fagegaltier et al. 2009). Ago2 and Dcr2 exert a repressive effect on expression of euchromatic genes by modulating transcriptional elongation (Cernilogar et al. 2011). In contrast, dosage compensation selectively elevates transcription of a large portion of the fly genome. The siRNA mutations examined in this study dramatically enhance the male-specific lethality of roX1 roX2 chromosomes and promote delocalization of the MSL complex from the X chromosome. This suggests that siRNA modulates the stability of MSL binding or contributes to recognition of the X chromosome. While evidence that Ago2 or other siRNA factors directly activate gene expression is lacking, a few studies have demonstrated increased silencing at some loci upon loss of Ago2 and Piwi (Yin and Lin 2007; Moshkovich and Lei 2010). It is possible that siRNA influences dosage compensation not through direct action at compensated genes, but by contributing to interphase chromosome architecture or organization of the nucleus. This would be consistent with the role of RNAi at insulators (Lei and Corces 2006; Moshkovich et al. 2011). Intriguingly, the male X chromosome displays an interphase conformation distinct from that in females (Grimaud and Becker 2009).