Retrotransposons are a major component of the genomes of higher eukaryotes and have been identified as a significant source of loss-of-function and regulatory mutations 1. Over evolutionary time host genomes have developed mechanisms to mitigate the mutational potential of retrotransposons by transcriptionally silencing or otherwise blocking their transpositional activity 2. One of the primary mechanisms by which retrotransposons are transcriptionally silenced is by methylation and/or other epigenetic mechanisms. Indeed it has been hypothesized that most, if not all, epigenetic mechanisms originally evolved as a defense against retrotransposons and have subsequently been co-opted for other essential cellular functions 34.

Approximately 10% of the Drosophila melanogaster genome is comprised of retrotransposons, the majority of which are LTR retrotransposons. LTR retrotransposon insertions are a major source of mutations in D. melanogaster and are believed to have contributed significantly to genome evolution 5. While histone acetylation and other epigenetic mechanisms are believed to play an essential role in dosage compensation and other vital functions in D. melanogaster, little is known about the role of these mechanisms in the regulation of retrotransposons in this species [e.g., 678]. In this paper, we present genetic and molecular evidence that the histone acetyltransferase, MOF, is involved in the transcriptional repression of the copia and perhaps other families of LTR retrotransposons in Drosophila. Our findings are consistent with the hypothesis that MOF may have originally functioned to silence retrotransposons and was subsequently co-opted for its role in dosage compensation.

A series of Drosophila melanogaster strains transformed with a copia LTR-CAT (chloramphenicol acetyltransferase) reporter construct (Figure 1) was monitored for CAT activity in larvae and adult flies carrying the construct in the hemizygous vs. the homozygous state. Inconsistent with a simple additive model, CAT activity was found, on average, to be ~7-fold higher in strains homozygous vs. hemizygous for the construct (Table 1). In contrast, CAT activity of strains in which two copies of the hemizygous construct were present at non-homologous locations was consistent with an additive model (Table 2 and 3). Thus, the observed non-additive effect between hemizygous and homozygous copia elements is cis-dependent.

Previous studies in other experimental systems have demonstrated that transposable elements are often partially or totally repressed due to host-mediated modifications in chromatin structure [e.g., 91011]. To determine if the non-additive effect we observed in copia expression is also associated with alterations in chromatin structure, we tested the ability of the ApaI restriction enzyme to access and digest a site in the copia LTR-CAT construct in larval chromatin extracts 12. The results indicate that hemizygous copies of the copia LTR-CAT construct are being down regulated due, at least in part, to a chromatin mediated effect (Figure 2).

A number of genes have been shown to regulate chromatin structure in Drosophila [e.g., 671314]. To determine if members of this class of genes are involved in repressing the expression of hemizygous copies of the copia LTR-CAT construct, we monitored CAT activity in larvae or adult flies hemizygous for the construct within a series of genetic backgrounds mutant for genes known to affect chromatin structure in Drosophila 1314 [LOW/+, HDAC 1³²⁶/+, HDAC 1³²⁸/+, E(z)²⁸/+, E(z)⁶¹/+, Psc25/+, Sxl^fl /+, and mof¹ (Table 4, see Additional files 12, Tables 1, 2, 3, 4), presence of extra Y chromosome or absence of Y chromosome (see Additional file 2, Table 5)]. With the exception of mof¹, each of the alleles tested display a dominant mutant phenotype in heterozygotes and are (embryonic) lethal when homozygous. Males hemizygous for the mutant (X-linked) mof¹allele develop to the 3^rd instar stage but fail to metamorphose and hatch. In contrast, the phenotype and viability of females homozygous for the mutant mof¹ allele is unaffected.

No significant effect on LTR-CAT expression was detected in genetic backgrounds heterozygous for the dominant mutant LOW, E(z)⁶¹ and Sxl ^fl alleles (Table 4; see Additional file 2, Tables 1,3 and 4). Flies heterozygous for the dominant mutant HDAC 1³²⁶, HDAC 1³²⁸, E(z)²⁸ and Psc²⁵ alleles and for the presence of an extra Y chromosome or absence of a Y chromosome displayed a slight but significant (p < 0.01) decrease in CAT activity relative to controls (Table 4, see Additional file 2, Tables 2, 3 and 5). In contrast, a highly significant (p < 0.001) increase in the expression of the copia LTR-CAT constructs was observed in male larvae mutant for the recessive (hemizygous) X-linked mof ¹ allele (males absent on first) (Table 5). This increase in expression in the mutant mof ¹ background was observed in all of the independent copia LTR-CAT transformants examined (Table 5). Thus, the effect is not dependent upon the chromosomal location of the construct.

The protein product of the mof gene (MOF) is a member of the MYST family of histone acetyltransferases and, as part of the Drosophila MSL (male specific lethal) complex 1516, has been shown to play an essential role in dosage compensation [e.g., 171819]. Our results are consistent with the effect of mof on copia LTR-CAT expression being chromatin-mediated (Figure 2).

Since most active transposable elements (including LTR retrotransposons) are considered to be hemizygous in wild collected populations, we tested the effect of the mutant mof ¹ allele on global expression levels of copia and other families of LTR retrotransposons in a strain established from a natural population (Athens, GA). The results of a series of RT-PCR and Northern analyses presented in Figure 3 demonstrate that males carrying the mutant mof ¹ allele display a consistent and significant increase in the expression of all families of LTR retrotransposons examined. This suggests that the association of mof with down regulated expression of hemizygous copies of copia may extend generally to other families of LTR retrotransposons in Drosophila.

While MOF is known to bind predominately to the male X-chromosome, weak binding also occurs at multiple locations on autosomes of Drosophila melanogaster 19 (Figure 4a). We observed a substantially reduced level of autosomal binding of MOF in Drosophila simulans, a species closely related to melanogaster but known to have significantly fewer numbers of LTR retrotransposons (Figure 4b) 20. This suggests that at least some of the autosomal binding of MOF may be to LTR retrotransposon sequences.

To directly determine if MOF can bind to copia elements, we performed chromatin immunoprecipitation (ChIP) analyses using the Drosophila melanogaster Kc167 (female) cell line and previously described Drosophila MOF, MSL-1 and MSL-3 antibodies 21. In males, these three proteins combine with MSL-2 (not expressed in females) and are sequestered to the X-chromosome as part of the MSL complex 21. In females, MOF, MSL-1 and MSL-3 have been shown to be associated with all chromosomes at a reduced but significant level 19. Since copia elements are known to be located on the X-chromosome, we chose to conduct our ChIP assays with the female Kc167 cell line in order not to confound copia specific binding by MOF with the more generalized X-chromosome binding that occurs in males. Spt4 is an autosomal gene previously shown not to be subject to dosage compensation 22. Rox is a well-established binding site for MOF within the context of the MSL complex 23. Since it has been shown that MSL-1, MSL-3 and MOF bind with reduced affinity to the X-chromosome in females, we selected rox as a putative positive control.

The results presented in Figure 5 indicate that all three proteins bind to copia elements in vivo (and to a lesser extent to rox) but presumably not within the context of the MSL complex since MSL-2 is not expressed in females (Figure 5). Previous studies have demonstrated that the function of MOF within the context of the MSL complex is to acetylate lysine 16 on histone H4 (H4K16ac) 24, resulting in chromatin decondensation and increased transcription. In contrast, our results indicated that MOF is associated with reduced transcription of hemizygous copies of copia and perhaps other Drosophila LTR retrotransposons. Although the acetylation of histones is generally associated with transcriptional activation, it can also be involved in transcriptional repression 252627. Further studies will be required to determine the molecular basis of these findings.

A determination of the mechanistic basis of MOF mediated silencing of copia remains to be determined. Nevertheless, the fact that hemizygous copies of copia and perhaps other families of Drosophila LTR retrotransposons are the target of epigenetic repression, appears to be analogous to MSUD (meiotic silencing by unpaired DNA) in Neurospora where DNA unpaired in meiosis causes silencing of the unpaired sequence and all DNA homologous to it 2829. Similar phenomena have been reported in C. elegans 3031 and mice 32, and have been associated with epigenetically mediated modifications in chromatin structure. Our results indicate that the repression of hemizygous copies of copia and other Drosophila LTR retrotransposons is also mediated by changes in chromatin structure. The fact that this repression appears to be mediated by MOF implies a relationship between retrotransposons and the evolution of dosage compensation in Drosophila.

While it is generally acknowledged that MOF plays an essential role in equalizing the expression of X-linked and autosomal genes in Drosophila males, the mechanism(s) by which this is achieved remains controversial 33. One model postulates that MOF, in association with other members of the MSL complex, binds to the hemizygous X-chromosome resulting in a two-fold increase in X-linked gene expression 3435. According to this model, MOF acetylates H4 on lysine 16 [H4Ac16] resulting in a relaxed chromatin configuration and a consequent elevation in gene expression. While our data indicate that MOF interacts with hemizygous copies of copia, the consequence is repression rather than elevation in expression. This suggests that the regulatory functions of Drosophila MOF may be context dependent.

The alternative "inverse regulator model" of dosage compensation postulates that hemizygosity of the male X-chromosome results is a genome-wide elevation in gene expression 1936. The significance of the sequestering of the MSL complex to the male X-chromosome in this model is two-fold. First removal of the MOF acetyltransferase from autosomes is postulated to attenuate the increased level of autosomal gene expression due to the X dosage affect. Secondly, the binding of the MSL complex prevents X-linked genes from responding to the elevated levels H4Ac16. The mechanism(s) underlying this second proposed function is unknown, although it has been demonstrated that the expression of at least some X-linked and autosomal genes are significantly elevated in male mof mutants 1936. This suggests that MOF alone or within the context of the MSL complex can act to repress expression of at least some X-linked genes. This finding is consistent with our observation that MOF represses copia expression in the hemizygous condition.

Since retrotransposons significantly predate the evolution of dosage compensation, our results suggest that the original function of MOF and perhaps other members of the MSL complex was to silence retrotransposons and that these functions were later co-opted in the evolution of dosage compensation in Drosophila.

A growing body of evidence in both plants and animals indicates that epigenetic mechanisms originally evolved as a defense against transposable elements and were subsequently co-opted for a variety of cellular functions 373839. Our findings are consistent with this model and suggest that at least some of the mechanisms underlying dosage compensation in Drosophila may have their origins in processes originally evolved to defend against the mutagenic potential of transposable elements.

LVM helped conceive the study, conducted the experiments and contributed to the data analysis; NJB contributed to the data analysis; JFMcD conceived the study, contributed to the data analysis and wrote the paper.