Chromatin-Associated Proteins HP1 and Mod(mdg4) Modify Y-Linked Regulatory Variation in the Drosophila Testis

Abstract

Chromatin remodeling is crucial for gene regulation. Remodeling is often mediated through chemical modifications of the DNA template, DNA-associated proteins, and RNA-mediated processes. Y-linked regulatory variation (YRV) refers to the quantitative effects that polymorphic tracts of Y-linked chromatin exert on gene expression of X-linked and autosomal genes. Here we show that naturally occurring polymorphisms in the Drosophila melanogaster Y chromosome contribute disproportionally to gene expression variation in the testis. The variation is dependent on wild-type expression levels of mod(mdg4) as well as Su(var)205; the latter gene codes for heterochromatin protein 1 (HP1) in Drosophila. Testis-specific YRV is abolished in genotypes with heterozygous loss-of-function mutations for mod(mdg4) and Su(var)205 but not in similar experiments with JIL-1. Furthermore, the Y chromosome differentially regulates several ubiquitously expressed genes. The results highlight the requirement for wild-type dosage of Su(var)205 and mod(mdg4) in enabling naturally occurring Y-linked regulatory variation in the testis. The phenotypes that emerge in the context of wild-type levels of the HP1 and Mod(mdg4) proteins might be part of an adaptive response to the environment.

CHROMATIN-REMODELING proteins are crucial in gene regulation and usually exert their function through chemical modifications of the DNA template, DNA-associated proteins, and RNA-mediated processes (Muller and Leutz 2001; Ng and Gurdon 2008; Brien and Bracken 2009). However, in fruit flies, DNA methylation is either absent or occurs at extremely low levels (Lyko et al. 2000; Phalke et al. 2009; Gou et al. 2010; Schaefer and Lyko 2010), such that the structure of chromatin might be entirely determined by protein modifications and RNA-mediated processes. The post-translational modifications of histones and DNA-binding proteins modulate the physical structure of DNA and ultimately regulate gene expression (Spencer and Davie 1999; Kouzarides 2007; Bonasio et al. 2010). Classical chromatin-remodeling proteins were classified based on their ability to modulate the spreading of heterochromatin at the boundaries between the domains of euchromatin and heterochromatin. Accordingly, genetic screens identified chromatin regulators in Drosophila using the white gene allele w(m4), in which the wild-type white gene is repositioned at a boundary between chromatin domains (position effect variegation, or PEV) (Ashburner et al. 2005). The studies uncovered >200 proteins with the ability to modulate chromatin states and enhance or suppress PEV (Henikoff 1990; Gelbart et al. 1997; Schotta et al. 2003; Ashburner et al. 2005; Tweedie et al. 2009). A small number of those modifiers have been mapped and functionally studied at the molecular level.

Heterochromatin protein 1 (HP1) is a classical nonhistone chromatin protein. HP1 is encoded by the Su(var)205 gene in Drosophila and localizes throughout the genome in euchromatic and heterochromatic sites. However, the protein is more abundant in segments of heterochromatin (de Wit et al. 2007; Yin et al. 2011). Recruitment of HP1 to chromatin is mediated by tri- or di-methylation of lysine 9 of histone 3 (H3K9m3 or H3K9me2), a chromatin mark established by the Su(var)3-9 protein (Grewal and Moazed 2003; Grewal and Jia 2007). The ability of HP1 knockdown to rescue silencing of the w(m4h) gene marker shows that HP1 is a key factor in chromatin structure and a strong suppressor of variegation (red-eye phenotype). Other modifiers of chromatin include Mod(mdg4) and JIL-1. Mod(mdg4) is a classical modifier of variegation. The locus is complex, with evidence of trans-splicing during messenger RNA (mRNA) processing (Cai and Levine 1997; Buchner et al. 2000; Mongelard et al. 2002). Experimental evidence indicates that Mod(mdg4) is involved in meiosis I and chromosome segregation in males (Soltani-Bejnood et al. 2007), in the regulation of chromatin assembly or disassembly (Gerasimova et al. 1995), oogenesis (Buchner et al. 2000), and germ-cell migration (Read et al. 2000). JIL-1 has histone kinase activity (Jin et al. 1999; Rath et al. 2006; Zhang et al. 2006) and is involved in the negative regulation of chromatin silencing and in chromosome organization. Deficiency of JIL-1 leads to strong enhancer of variegation (white-eye phenotype), counteracting even the suppressor-of-variegation effect of HP1 and Su(var)3-9 (Wang et al. 2011).

Y-linked regulatory variation (YRV) refers to the quantitative effects that polymorphic tracts of Y-linked chromatin exert on gene regulation of X-linked and autosomal genes (Lemos et al. 2008). One hypothesis is that copy-number polymorphism in tracts of sequence repeats underlies cryptic regulatory variation emerging from Y chromosomes (Lemos et al. 2008). This expectation rests on the lack of polymorphic variation in Y-linked protein-coding genes as well as the finding that Y-linked repeats are highly variable (Lyckegaard and Clark 1989; Zurovcova and Eanes 1999; Paredes et al. 2011; Larracuente and Clark 2013). The observation that YRV occurs in XXY female genotypes, in which Y-linked protein-coding genes are not expressed, gives further strength to the notion that cryptic Y-linked repeats might cause YRV (Lemos et al. 2010). Finally, repeat variation in the Y-linked ribosomal DNA (rDNA) locus is consequential to the specification of chromatin states and genome-wide gene expression in Drosophila (Paredes and Maggert 2009; Paredes et al. 2011).

Here we addressed whether the expression of YRV is dependent on Su(var)205 dosage. In addition, we tested the chromatin remodelers mod(mdg4) and JIL-1 for their ability to modify YRV in a Y-chromosome-dependent manner. We found that, independently of Y-chromosome background, Su(var)205 and mod(mdg4) strongly suppress PEV, whereas JIL-1 strongly enhances PEV. Furthermore, we addressed the contribution of these genes to the manifestation of testis-specific and somatic YRV. Specifically, we found that Y-ohio shows up-regulation of testis-specific gene expression in two wild-type backgrounds, a pattern that is abolished upon introduction of heterozygous loss-of-function (LoF) mutations for Su(var)205 and mod(mdg4). This highlights an attribute of YRV that is HP1- and Mod(mdg4)-dependent. Conversely, JIL-1 does not modify the pattern of YRV observed in the testis of wild-type genetic backgrounds. Taken together, the data suggest that the functions of HP1 and Mod(mdg4) contribute to Y-chromosome states that modify YRV in the Drosophila testis. The data also indicate that wild-type levels of both Su(var)205 and mod(mdg4) products are required for the suppression of YRV in somatic tissues. Altogether, our observations suggest possibly adaptive and deleterious manifestations of cryptic polymorphic variation on highly heterochromatic Drosophila Y chromosomes.

Materials and Methods

Drosophila strains

Fly strains were obtained from the Bloomington Stock Center. The chromosomes containing the gene deletion of mod(mdg4)^L3101 {y[1] w[1118]; P{w[+mC]=lacW}mod(mdg4)[L3101]/TM3, Ser[1]}, Su(var)205 {In(1)w[m4h]; Su(var)205[5]/In(2L)Cy,In(2R)Cy, Cy[1]}, mod(mdg4)^G16853 {w[1118]; P{w[+mC]=EP}mod(mdg4)[G16853]/TM6C, Sb[1]}, and JIL-1^Scim {y[1]; P{y[+mDint2] w[BR.E.BR]=SUPor-P}JIL-1[Scim] ry[506]} were introgressed into the genetic background y[1]; bw[1]; e[4]; ci[1] ey[R] (Supporting Information, Figure S1, Figure S2, and Figure S3). All females used in the introgressions were collected within 7 hr after eclosion. For gene expression analyses, flies were grown in incubators at 25°, 65% of relative humidity, and constant light. Newly emerged adults harboring the mutations (Figure S2 and Figure S3) were collected and aged for 2 days under the same rearing condition before they were flash-frozen in liquid nitrogen. Four replicas of each sample were collected and stored at −80°. To assess PEV, male genotypes (Y-ohio or Y-congo) with loss-of-function mutations were crossed with virgin In(1)w[m4h]; Su(var)3-10[2]/TM3, Sb[1] Ser[1] or In(1)w[m4h]; Su(var)2-4[01] pie[54]/CyO. Flies showing variegated eyes were collected on the same day, aged for 24 hr, and flash-frozen in liquid nitrogen before pictures were taken.

Gene expression analysis

Microarrays were ∼18,000-feature complementary DNA (cDNA) arrays spotted with Drosophila melanogaster cDNA PCR products. Total RNA was extracted from whole flies using TRIzol (Life Technologies). The synthesis of cDNA and its labeling with fluorescent dyes (Cy3 and Cy5) as well as hybridization conditions were carried out using 3DNA protocols and reagents (Genisphere). Four replicas of each sample were hybridized and scanned with Axon 400B scanner (Axon Instruments) and GenePix Pro 6.0 software. We used stringent quality control criteria to ensure reliability of foreground intensity reads for both Cy3 and Cy5 channels (Lemos et al. 2008). The procedure screens out poor quality arrays and ensures that the resulting data are minimally sensitive to the choice of normalization method. Foreground fluorescence of dye intensities was normalized by the Loess method in Bioconductor/Limma (Smyth and Speed 2003; Smyth 2005). The significance of variation in gene expression was assessed with the Bayesian analysis of gene expression levels (BAGEL) (Townsend and Hartl 2002). Results were checked for consistency using linear models in Limma. False discovery rates (FDRs) were empirically estimated by permutation of the data set. The microarray gene expression data used here can be obtained at the Gene Expression Omnibus database under accession nos. GSE46313 and GSE9457. For quantitative PCR (qPCR) analyses, three biological replicates of 10 adult flies each were flash-frozen in liquid nitrogen. Total RNA was extracted with TRIzol (Life Technologies), and cDNA was synthetized with a QuantiTect Reverse-Transcription Kit (Qiagen). Real-time qPCR analyses were carried out with the Fast Sybr Green Master Mix (Applied Biosystems) using 7900HT Fast Real-Time PCR (Applied Biosystems). Expression level of the Su(var)205 gene was analyzed with the software REST (Pfaffl et al. 2002).

Tissue expression

Tissue specificity of differentially expressed (DE) genes was assessed using the FlyAtlas (Chintapalli et al. 2007; downloaded on December 2011 ). We filtered the data to include only a nonredundant set of adult tissues: brain, eye, crop, midgut, hindgut, tubule, testis, accessory gland, adult fat body, heart, and trachea. For each Affymetrix probe in the FlyAtlas data set and for each tissue, we set expression level to 0 unless the probe was called as “present” in at least two of four arrays and then averaged over all probes and arrays for each FBgn to calculate the expression level for each gene in each tissue. Genes with expression counts below a threshold (20, 100, 250) were conservatively set to zero. We calculated a tissue prevalence index (TPi) for tissue i in a set of n tissues as:

$$TP\left((i)\right)=\frac{\text{Number of DE genes with expression in tissue}\left((\text{i})\right)}{\text{Total number of DE genes}}.$$

TP(i) of 1 means that all genes in the set of differentially expressed genes show evidence of expression in the focal tissue.

Results

Su(var)205 and mod(mdg4) are dominant suppressors of PEV regardless of Y-chromosome background

To address the interaction between Y chromosomes and two key suppressors of variegation, we introgressed Su(var)205 and mod(mdg4) LoF mutations into an isogenic genotype with variable Y chromosomes (Figure S1, Figure S2, and Figure S3). The two Y chromosomes assayed were chosen due to their contrasting contributions to w(m4h) variegation (Figure 1A; Lemos et al. 2010). We assayed for PEV in genotypes containing LoF mutations in Su(var)205 or mod(mdg4) and differing only in the origin of the Y chromosome. As expected, we observed that the mutation in Su(var)205 resulted in a wild-type eye phenotype in the background of both Y chromosomes (Figure 1B). Similarly, two LoF alleles of mod(mdg4) [alleles mod(mdg4)^L301and mod(mdg4)^G16853] resulted in a wild-type eye phenotype in the background of both Y chromosomes (Figure 1, C and D). The mod(mdg4)^L301 allele is a well-characterized allele with a P-element insertion in the third exon of the gene (Mongelard et al. 2002). Collectively, our results indicate that the enhancer-of-variegation strength of the Y-congo chromosome cannot counterbalance the suppression of variegation that is observed in genotypes that are heterozygous for a LoF mutation in Su(var)205 or mod(mdg4).

Figure 1.

Effects of chromatin modulators on white-mottled (wm4h) variegation in Drosophila genotypes with diverse Y chromosomes (A–E). Su(var)205 (B) and two distinct alleles of mod(mdg4) (C, D) are strong suppressors of chromatin spreading and overcome the enhancer-of-variegation effect of the Y-congo. JIL-1^Scim (E) is a strong enhancer of variegation, which is effective even in the background of the Y-ohio.

JIL-1 is a dominant enhancer of PEV regardless of Y-chromosome background

JIL-1 is a multidomain protein responsible for the phosphorylation of the histone H3, as well as maintenance of the structure of the polytene chromosomes in Drosophila. JIL-1 LoF alleles have a strong effect on PEV, and JIL-1; Su(var)205 and JIL-1; Su(var)3-9 double mutants show enhanced variegation (Wang et al. 2011). This indicates that reduced dosage of JIL-1 can counterbalance the typical suppressive effect of HP1 and Su(var)3-9 on PEV. We tested whether JIL-1 impacts PEV in the background of both Y-congo and Y-ohio genotypes. As in the experiment described above, the strains differed only in the origin of the Y chromosome and the presence of a chromosome containing a LoF JIL-1 allele. We observed that both Y-congo and Y-ohio show a markedly white-eye phenotype in the heterozygous JIL-1/+ genotype (Figure 1E). This indicates that loss of function of JIL-1 has an enhancer-of-variegation effect that is independent of the Y-chromosome genetic background. This occurs in spite of the fact that the JIL-1 allele tested is a weak hypomorphic allele (Zhang et al. 2003).

Su(var)205-dependent up-regulation of testis-specific gene expression in Y-ohio

HP1 is a major component of heterochromatin and one of the strongest known suppressors of variegation. Results presented thus far indicate that Su(var)205 has an overwhelming suppressive contribution to PEV, which might completely mask the contribution of naturally occurring Y-chromosome variants. The possibility remained, however, that HP1 might modulate gene expression in a Y-chromosome-dependent manner. Using the FlyAtlas (Chintapalli et al. 2007), we observed that, in a wild-type genetic background, the Y-ohio leads to a substantial up-regulation of testis-specific gene expression relative to the Y-congo (% testis-specific across the genome, 14.6; % testis-specific up-regulated by Y-ohio in the wild-type background, 41.2; Fisher’s exact test odds ratio, 2.82; P = 1.56E-15). Greater up-regulation of genes expressed in testis might be expected in Y-ohio because this Y-chromosome variant acts as a suppressor of variegation. However, when the whole genome is considered, a similar number of genes appear up-regulated and down-regulated in Y-ohio (Figure 2A).

Figure 2.

Y-linked regulatory variation in heterozygous genotypes with a loss-of-function Su(var)205 mutation. (A) Similar number of genes differentially expressed (BPP > 0.99; FDR < 0.10) between Y-congo and Y-ohio when assayed in wild-type and Su(var)205 loss-of-function genetic backgrounds. (B) qPCR indicates an ∼50–60% reduction in Su(var)205 mRNA abundance in Su(var)205/+ genotypes. (C) Volcano plot showing similarity between numbers and fold-changes of differentially expressed genes in the Su(var)205/+ background. Solid circles denote genes differentially expressed. Genes above and below zero are, respectively, up- and down-regulated in Y-ohio. (D) Agreement in fold-change estimates obtained with linear models in Limma and with Bayesian methods in BAGEL; data are for the contrast between Su(var)205/+; Y-congo and Su(var)205/+; Y-ohio.

To address the hypothesis that Y-ohio up-regulation of testis-specific expression might be dependent on HP1 function, we profiled Y-linked regulatory variation in flies that were heterozygous for a LoF allele of the Su(var)205 gene. The presence of one defective allele knocks down the expression of the Su(var)205 gene significantly (Figure 2B). Thus, we contrasted Su(var)205/+; Y-congo and Su(var)205/+; Y-ohio, which are genetically identical except for Y-chromosome origin. First, we observed hundreds of genes significantly DE in this comparison, with the Y-congo and Y-ohio inducing similar numbers of genes in both the wild-type and Su(var)205/+ genotypes (Figure 2, A and C). We observed 473 DE genes at Bayesian posterior probability (BPP) > 0.99 (FDR < 0.10), with a comparable number of genes up- and down-regulated (Figure 2, A and C). We observed a high correlation between P-values obtained with linear models in Limma and BPP obtained in BAGEL (ρ = 0.96; P < 10E-16), with nearly perfect correspondence between the fold-change estimates by the two methods (ρ = 0.99; P < E-16; Figure 2D). Finally, we investigated the consequences of the Su(var)205 LoF allele on the patterns of testis-specific up-regulation that we observed in the Y-ohio background. Surprisingly, the Y-chromosome-dependent up-regulation of testis-specific genes in Y-ohio vanishes in the Su(var)205/+ genotype (Figure 3). Indeed, the number of DE testis-specific genes in Su(var)205/+ genotypes with variable Y chromosomes is actually significantly smaller than that expected from the genomic background (testis-specific genes genome-wide, 14.6%; testis-specific genes up-regulated by Y-ohio in the Su(var)205/+ background, 6.2%; Fisher’s exact test odds ratio, 0.41; P = 0.0006). Altogether, these data indicate that wild-type HP1 levels are required for the manifestation of testis-specific gene expression differences that are exerted by cryptic Y-chromosome polymorphisms.

Figure 3.

Wild-type level of HP1 is required for Y chromosome-dependent up-regulation of testis-specific genes. Heat maps showing the genes (vertical columns) that are up-regulated in Y-ohio relative to Y-congo, with their expression values across 11 tissues (horizontal lines). Data are shown for genes that are up-regulated in Y-ohio relative to Y-congo in the +/+ and Su(var)205/+ genetic backgrounds. The scale color of the heat map was normalized for each gene with green and red denoting high and low transcript abundance, respectively. The first line in each of the heat maps depicts expression levels in the testis tissue. The evident green line, highlighted by the black bar on the top of the heat maps, indicates testis-specific genes that are up-regulated in Y-ohio.

YRV in genetic backgrounds with loss-of-function mutations in mod(mdg4) and JIL-1

Our analyses suggested a complex architecture of Y-chromosome-by-background interaction, with genomic background exerting a strong influence on YRV. To address the robustness in the pattern of testis-specific up-regulation in Y-ohio, we performed genome-wide expression analysis in additional backgrounds. As expected, in a second wild-type strain, Y-ohio showed the same pattern and a similar magnitude of testis-specific gene up-regulation as observed before (41.2% of the DE genes in background 1 are testis specific vs. 45.1% in background 2). To extrapolate the results obtained with Su(var)205/+ to other chromatin regulators, we assessed the gene expression profile in mod(mdg4)/+ and JIL-1/+ heterozygous. Testis-specific up-regulation in Y-ohio is insensitive to the presence of a JIL-1 LoF allele, with 49% of the genes up-regulated in JIL-1/+; Y-ohio showing testis specificity relative to 14.6% across the genome (Fisher’s exact test odds ratio, 3.42; P = 7.897E-10; Figure 4). Indeed, the proportion of genes up-regulated in Y-ohio that are testis specific is indistinguishable between the two wild-type backgrounds and the JIL-1/+ genotype (Fisher’s exact test, P > 0.10 in all comparisons between these three genotypes). Notably, this stability occurs in spite of the substantial effect of the JIL-1 allele on PEV, which is indistinguishable between Y-ohio and Y-congo when these chromosomes are in the presence of a JIL-1 LoF mutation (Figure 1E). On the other hand, results with mod(mdg4)/+ recapitulate those observed with Su(var)205/+; namely, the pattern of Y-ohio testis-specific gene up-regulation vanishes (Figure 4).

Figure 4.

Genetic interaction between the Y chromosome and mod(mdg4) modulates testis-specific gene expression. Heat maps showing the genes (vertical columns) that are up-regulated in Y-ohio relative to Y-congo in the +/+ (background 2), mod(mdg4)/+, and JIL-1/+ genetic backgrounds with their expression values across 11 tissues (horizontal lines). The scale color of the heat map was normalized for each gene with green and red denoting high and low transcript abundance, respectively. The first line in each of the heat maps depicts expression levels in the testis tissue. The black bar on the top of the heat maps indicates testis-specific or testis-biased genes that are up-regulated in Y-ohio.

Ubiquitously expressed genes are disrupted in Su(var)205-dependent and mod(mdg4)-dependent YRV

Genetic interactions between Su(var)205 and Y-ohio and mod(mdg4) and Y-ohio are required to promote the Y-ohio-dependent up-regulation of gene expression in the testis (Figure 5). To further address the consequences of the breakdown of genetic interactions in Su(var)205/+ genotypes, we investigated patterns of tissue up-regulation using the FlyAtlas data set (Chintapalli et al. 2007). FlyAtlas is a unique resource with tissue expression data collected across several tissues. To gauge the contribution of specific tissues to YRV, we devised a tissue prevalence index (TPi) to estimate the contribution of genes expressed in each tissue to the total number of DE genes. Figure 6A shows that the genes up-regulated in two wild-type backgrounds and the JIL-1/+ genotype display a stable and coherent pattern: 44–58% of genes up-regulated in Y-ohio show evidence of testis expression above the detectable threshold. On the other hand, <18% of the genes up-regulated in Y-ohio show evidence of expression in any of the other tissues. Remarkably, the introduction of the Su(var)205 or mod(mdg4) LoF allele causes the differential expression of several genes that are expressed in somatic tissues while it masks the original differences detected in the testis (Figure 6B). Indeed, this is in agreement with our observations that genes differentially expressed in the Su(var)205 background show a significant depletion in targets with testis expression (Figure 5). For the assessment of tissue prevalence, we used average expression counts >100 units across the four replicates in the FlyAtlas set to be an indication that the gene is expressed. Using more stringent (200 expression units) or less stringent (20 expression units) cutoff thresholds for ascertaining expression yielded nearly identical results. Hence, we conclude that the wild-type functions of Su(var)205 and mod(mdg4) might be required to mitigate the impact of Y-chromosome polymorphisms on gene expression in somatic tissues.

Figure 5.

Proportion of testis-specific genes in the set of genes up-regulated by Y-ohio in the presence/absence of loss-of-function alleles. Y-ohio up-regulation of testis-specific expression is Su(var)205-dependent and mod(mdg4)-dependent. Two +/+ backgrounds were used (see Figure S1, Figure S2, and Figure S3 for details on backgrounds). P-values (based on Fisher’s exact tests) relative to the proportion of testis-specific genes across the genome (14.6%) is <0.001 for all cases.

Figure 6.

Tissue prevalence of genes that are modulated by genetic interactions between Y chromosomes and Su(var)205, mod(mdg4), and JIL-1. (A) The contrasts between Y-congo and Y-ohio in two control wild-type +/+ genotypes and in the JIL-1/+ genotype indicate that the majority of the genes up-regulated by Y-ohio show evidence for expression in the testis. (B) The contrasts between Y-congo and Y-ohio in the Su(var)205/+ and mod(mdg4)/+ genotypes indicate that the genes up-regulated in Y-ohio show a prevalence for expression in somatic tissues. For reference, a wild type +/+ is also displayed in B.

Discussion

Chromatin modifications are central to genomic regulation, with euchromatin and heterochromatin as the most clearly distinguishable domains. Chromatin states are defined by post-transcriptional marks on histones that guide nonhistone proteins to bind specifically to the DNA in active or inactive segments of DNA. In Drosophila, heterochromatin represents 30–35% of the genome of most somatic cells (Girton and Johansen 2008), and the proteins Su(var)3-9, Su(var)3-7, and HP1 are abundant components of heterochromatin. The Y chromosome represents a major heterochromatin segment in fruit flies, and its content modulates PEV and gene expression (Dimitri and Pisano 1989; Ashburner et al. 2005; Lemos et al. 2008; Jiang et al. 2010; Lemos et al. 2010).

Interactions between Su(var)205 and YRV

HP1 is a structural nonhistone chromosomal protein that is conserved from human to yeast (Saunders et al. 1993; Allshire et al. 1995; Pak et al. 1997). It functions in heterochromatin-mediated gene silencing, as well as in the regulation of active euchromatic genes (Eissenberg et al. 1990; de Wit et al. 2007; Kwon and Workman 2011; Yin et al. 2011). The epigenetic mark for HP1 recruitment and heterochromatin formation is H3K9 methylation, exerted by the enzyme Su(var)3-9 (Rea et al. 2000; Peters et al. 2003). H3K9 methylation marks generate affinity sites for HP1, and interactions between HP1 and Su(var)3-9 and between HP1 and Su(var)3-7 are critical for the establishment of heterochromatin (Cleard et al. 1997; Bannister et al. 2001; Lachner et al. 2001; Delattre et al. 2004; Vermaak and Malik 2009). In Drosophila, the amount of HP1 correlates with suppression of the PEV phenotype (Eissenberg et al. 1990). As expected, heterozygous genotypes containing a LoF allele of Su(var)205 show an ∼50% decrease in mRNA abundance, leading to a strong suppression of heterochromatin spreading that shifts the PEV phenotype to wild type. We observed that the ability of HP1 to suppress PEV is independent of the presence of a strong naturally occurring enhancer-of-variegation Y-congo chromosome in the genetic background. On the other hand, gene expression analysis uncovered genetic interactions between HP1 dosage and polymorphic variation in Y chromosomes. Accordingly, the results indicate that polymorphic Y-chromosome-dependent modulation of testis-specific gene expression is modified by HP1 dosage. The affected pathways are unclear because the vast majority of testis-specific genes are yet to be functionally characterized (McQuilton et al. 2012). Nevertheless, testis-specific genes are enriched in “black” chromatin (Filion et al. 2010), which is the prevalent type of repressive chromatin in the Drosophila embryonic cell line Kc167. Black chromatin is substantially marked by histone H1 and chromosomal proteins IAL, D1, and SUUR, whereas Su(Hw), LAM, and EFF are also frequently present (Filion et al. 2010). Finally, we emphasize that the effects described here are not attributable to the Y chromosome or to Su(var)205 alone, but are instead dependent on the interaction between the Su(var)205 locus and cryptic polymorphic factors on the Y chromosome.

Interactions between mod(mdg4) and YRV

Modifier of mdg4 [Mod(mdg4)] is a multifunctional protein that encodes a BTB/POZ domain transcription factor (Read et al. 2000). Transcripts of this locus can be differentially processed and generate numerous splice variants, with a 2.2-kb transcript being the most abundant isoform (Gerasimova et al. 1995; Buchner et al. 2000; Labrador et al. 2001). The mod(mdg4) LoF allele used in this work [mod(mdg4)^L3101] is a recessive homozygous lethal in the larval stage. It is also a strong suppressor of cycE (Read et al. 2000), which may play an important role in the regulation and entry into the S phase of the cell cycle. Furthermore, mutations in mod(mdg4) disrupt X–Y segregation more severely than autosomal segregation (Soltani-Bejnood et al. 2007) and display a recessive-lethal interaction with the Y chromosome. Altogether, these results suggest that mod(mdg4) interacts with the Y chromosome. Here we observed that mod(mdg4) deficiency specifically disturbs the expression of testis-specific genes in Y-ohio. As noted, these genes are abundant in black chromatin, which is frequently marked by the chromosomal proteins D1 and Su(Hw) (Filion et al. 2010). D1 and Su(Hw) proteins physically interact with Mod(mdg4) (Guruharsha et al. 2011), and Su(Hw) is essential for Mod(mdg4) function (Georgiev and Kozycina 1996; Gdula and Corces 1997). These observations raise the possibility that the pathways through which Mod(mdg4) interacts with the Y chromosome intersect those mediated by HP1, Su(Hw), and D1. Specifically, D1 binds to satellite DNA repeats (Blattes et al. 2006). Furthermore, the Su(Hw) protein is required for the gypsy transposon insulator to block enhancer–promoter interactions (Gause et al. 2001). Upon DNA binding, Su(Hw) recruits Mod(mdg4), and Mod(mdg4) determines how gypsy insertions alter gene expression with insulator and repressor functions (Gerasimova et al. 1995; Georgiev and Kozycina 1996; Cai and Levine 1997). Together, our results indicate that the interaction between Mod(mdg4) and the Y chromosome is required for testis-specific gene up-regulation in Y-ohio. The interactions might be mediated by cryptic factors that are polymorphic in the Y chromosome and affect spermatogenesis. The data also suggest that wild-type mod(mdg4) functions might be required to suppress the somatic effects of cryptic Y-linked polymorphisms.

Interactions between JIL-1 and YRV

Many chromatin-remodeling proteins might be able to interact directly or indirectly with the Y chromosome and modulate PEV. Here we addressed whether reductions in JIL-1 dosage due to heterozygous LoF mutations might be important modifiers of YRV. JIL-1, a kinase responsible for H3S10 phosphorylation, can reduce the suppression-of-PEV phenotype of Su(var)3-9, Su(var)3-7, and Su(var)205 LoF mutations (Deng et al. 2010; Wang et al. 2011). We observed this same result when we introgressed Y-congo and Y-ohio into the JIL-1^Scim LoF allele. The variegated phenotype was abolished even for the Y-ohio, which shows a strong suppressor-of-variegation capacity. This occurs despite the fact that JIL-1^Scim express nearly 58% of native protein, while other alleles abolish a larger fraction of the protein pool (Zhang et al. 2003). We cannot exclude the possibility that stronger JIL-1 LoF could disrupt testis-specific up-regulation in Y-ohio. It is also possible that over evolutionary timescales, subtle interactions between JIL-1 alleles and variable Y chromosomes play important roles in the dynamics of chromatin states in natural populations.

Evolution of Y-chromosome-by-autosome interactions

A key puzzle is the interplay between dosage variation in chromatin components and variable Y-chromosome chromatin. Polymorphic Y chromosomes might interact with naturally occurring variation in chromatin dosage across individual genotypes and tissue types. Our findings point to interactions between the sources of YRV in Y chromosomes and the dosage of HP1 and Mod(mdg4). Disruption in backgrounds with mutations in Su(var)205 or mod(mdg4) leads to the tissue-wide consequences of YRV. One hypothesis is that YRV in somatic tissues might be deleterious, with limited fitness benefits accrued from the interaction between Y chromosomes and variable chromatin dosage. Indeed, we expect that tissue-wide disruption of gene expression due to the interaction between Su(var)205 and mod(mdg4) with polymorphic Y chromosomes might be deleterious over evolutionary timescales. On the other hand, testis-specific YRV might be conditionally beneficial and driven by adaptive Y-chromosome evolution and positive natural selection of cryptic factors responding to HP1 levels. In this notion, the higher expression of testis-specific genes that is driven by the Y chromosome in specific backgrounds might result in adaptive variation in reproductive fitness. Accordingly, evidence that polymorphic Y chromosomes interact with the autosomal background and impact reproductive phenotypes and fitness has been documented (Chippindale and Rice 2001). Interestingly, the higher rDNA copy number in Y-congo relative to Y-ohio (Paredes et al. 2011) might suggest a role for rDNA–HP1 and rDNA–Mod(mdg4) interactions in fitness. We predict that the balance between somatic and testis-specific consequences of YRV in variable genomic backgrounds encountered in natural populations will prove to be a major determinant of how natural selection shapes the evolution of the Y chromosome and its repeats.