The Drosophila GAGA Factor Is Required for ... - Semantic Scholar

Copyright  2004 by the Genetics Society of America

The Drosophila GAGA Factor Is Required for Dosage Compensation in Males and for the Formation of the Male-Specific-Lethal Complex Chromatin Entry Site at 12DE Anthony J. Greenberg,1 Judith L. Yanowitz2 and Paul Schedl3 Department of Molecular Biology, Princeton University, Princeton, New Jersey 08544 Manuscript received March 13, 2003 Accepted for publication October 6, 2003 ABSTRACT Drosophila melanogaster males have one X chromosome, while females have two. To compensate for the resulting disparity in X-linked gene expression between the two sexes, most genes from the male X chromosome are hyperactivated by a special dosage compensation system. Dosage compensation is achieved by a complex of at least six proteins and two noncoding RNAs that specifically associate with the male X. A central question is how the X chromosome is recognized. According to a current model, complexes initially assemble at ⵑ35 chromatin entry sites on the X and then spread bidirectionally along the chromosome where they occupy hundreds of sites. Here, we report that mutations in Trithorax-like (Trl) lead to the loss of a single chromatin entry site on the X, male lethality, and mislocalization of dosage compensation complexes.

ROSOPHILA melanogaster males have one X chromosome, while females have two. To compensate for this disparity in gene dose, there are mechanisms to ensure that X-linked genes are expressed at the same level in the two sexes. One of these mechanisms is the hyperactivation of genes on the male X chromosome by a special dosage compensation system (reviewed in Lucchesi and Manning 1987; Baker et al. 1994; Cline and Meyer 1996; Kelley and Kuroda 2000; Pannuti and Lucchesi 2000). Hyperactivation of X-linked genes in males is mediated by the male-specific-lethal (MSL) complex, which consists of at least six proteins, Msl1, Msl2, and Msl3, maleless (Mle; reviewed in Lucchesi and Manning 1987; Baker et al. 1994), males absent on the first (Mof; Hilfiker et al. 1997), and JIL-1 (Jin et al. 2000), as well as two noncoding RNAs, roX1 and roX2 (Franke and Baker 1999; Meller et al. 2000). Assembly of the dosage compensation complex on the X chromosome results in preferential association of an acetylated form of histone H4 with the X (Bone et al. 1994) and hyperactivation of most X-linked genes (Lucchesi and Manning 1987; Baker et al. 1994). In females, the Sex-lethal (Sxl) gene turns off the male dosage compensation system by repressing msl2 translation (Bashaw and Baker 1997; Kelley et al. 1997). In the absence of Msl2, Msl1 (Kelley et al. 1997) and Msl3 (Gorman et

D

1 Present address: Department of Ecology and Evolution, University of Chicago, Chicago, IL 60637. 2 Present address: Department of Embryology, Carnegie Institution of Washington, Baltimore, MD 21210. 3 Corresponding author: Department of Molecular Biology, Lewis Thomas Labs, Washington Rd., Princeton University, Princeton, NJ 08544. E-mail: [email protected]

Genetics 166: 279–289 ( January 2004)

al. 1995) are degraded and the dosage compensation complex is not assembled on the X chromosomes of females. Msl2 can be ectopically expressed in females by deleting the Sxl-binding sites in the msl2 mRNA untranslated regions (Kelley et al. 1995). MSL complexes are then assembled on the X, resulting in a partial loss of female viability (Kelley et al. 1995). The lethal effects of ectopic Msl2 can be suppressed by reducing the dose of msl1 (Kelley et al. 1995; Lyman et al. 1997) or mof (Hilfiker et al. 1997). A similar effect can be achieved by eliminating mle or msl3 activity (Kelley et al. 1995). While a good deal is known about the assembly and mechanism of action of the MSL dosage compensation complex, it is not fully understood how the complex is specifically targeted to the X chromosome. The current model is based on the observation that in the absence of msl3, mle, or mof, Msl1 and Msl2 associate with ⵑ35 special, high-affinity sites on the X chromosome (Palmer et al. 1993; Gorman et al. 1995; Lyman et al. 1997; Gu et al. 1998). Two of these sites correspond to the roX1 and roX2 genes (Kelley et al. 1999). When moved to the autosomes, transgenes bearing copies of roX1 and roX2 initiate binding and, in some cases, spreading of the dosage compensation complex into adjacent chromatin (Kelley et al. 1999; Meller et al. 2000; Henry et al. 2001; Kageyama et al. 2001). These data led to a proposal (Kelley et al. 1999) that the high-affinity sites, termed “chromatin entry sites,” serve as assembly points of the dosage compensation complex from which it spreads into neighboring regions of the X chromosome. With the exception of JIL-1 (Wang et al. 2001), the protein components of the MSL complex are encoded by genes in which loss-of-function mutations cause malespecific lethality (Baker et al. 1994; Hilfiker et al. 1997).

280

A. J. Greenberg, J. L. Yanowitz and P. Schedl

Genes that function not only in dosage compensation but also in processes vital to both sexes would not be identified by these criteria. However, genes that have this type of dual function could disproportionately affect male viability when their activity is compromised but not completely eliminated. This is the case for JIL-1, a tandem kinase that phosphorylates the histone H3 (Wang et al. 2001). It was first characterized as a protein enriched on the X chromosome ( Jin et al. 1999) and turned out to be part of the MSL complex ( Jin et al. 2000). Null mutations in jil-1 are sex nonspecifically lethal. However, males homozygous for a hypomorphic allele of the gene are less viable than their sibling females, although the females also display reduced viability (Wang et al. 2001). Another gene that appears to function in both dosage compensation and sex-nonspecific vital processes is Trithorax-like (Trl; Farkas et al. 1994). Trl encodes the Drosophila GAGA factor (Farkas et al. 1994; reviewed in Granok et al. 1995; Wilkins and Lis 1997). The GAGA factor functions in the formation/maintenance of nucleosome-free regions of chromatin both in vitro (Tsukiyama et al. 1994) and in vivo and is required for proper expression of many different genes (Wilkins and Lis 1997). Although Trl is an essential gene, a hypomorphic allele, Trl13C, can produce viable adults at a low frequency (Farkas et al. 1994; Bhat et al. 1996). In the course of our characterization of this Trl mutant, we found that certain combinations of this allele with strong loss-of-function alleles affect viability of males much more than that of females, as has been observed for jil-1 [H. Gyurkovics and J. Gausz (personal communication) have made similar observations]. Since a reduction in Trl activity is more deleterious to males than to females, we hypothesized that this gene may be involved in MSL-mediated dosage compensation. In the study reported here, we found that, in addition to male lethality, a reduction in Trl function results in the loss of a single X chromosome chromatin entry site and the redistribution of the MSL complex to autosomes. These results provide further evidence in support of the entry site hypothesis and point toward possible ways of identifying new genes involved in entry site formation. MATERIALS AND METHODS Genetic crosses and fly stocks: All crosses were performed at 22⬚ in an incubator. Flies were grown on standard media. Trl13C and Trl 62 are independent P-element-induced alleles (Farkas et al. 1994), while Trl 2.3 is an EMS-induced allele (A. J. Greenberg, unpublished results). All three alleles were obtained in independent screens. Trl 62/Trl 2.3 is a larval lethal combination of alleles. Trl13C/Trl13C, Trl13C/Trl 62, and Trl13C/ Trl 2.3 are partially adult viable. All msl mutants and transgene lines were a generous gift of R. Kelley and M. Kuroda. The male-to-female ratio of flies carrying adult-viable combinations of Trl alleles described above was determined by combining data from crosses in several genetic backgrounds (i.e., different balancers and different directions of the cross). For

each combination of alleles the ratio was not significantly different between genetic backgrounds (data not shown). To assess the effects of msl mutants on the viability of males with incomplete Trl function, the following cross was performed, w ; msl/CyO; Trl13C/ TM6b,Sb ⫻ w ; ⫹/⫹; Trl ⫺/ TM6b,Sb, where msl denotes alleles of msl1, msl2, mle and their combination that were used as described in Figure 2a and Table 1. Trl ⫺ is either Trl 62 (for crosses described in Figure 2a and the top of Table 1) or Trl13C for the cross described in the bottom of Table 1 (for the latter cross, TM3,Ser balancer chromosome was used). To ensure that any effects seen in the experiment outlined above were not due to differences in genetic background, the following control cross was performed: w ; ⫹/CyO; Trl13C/TM6b,Sb ⫻ w ; ⫹/⫹; Trl 62/TM6b,Sb. The male:female ratio of w ; ⫹/⫹; Trl13C/Trl 62 flies obtained in this cross was then compared to the ratios obtained in the experimental crosses. The control for the cross involving the triple msl mutant combination in the Trl13C homozygous background was w ; ⫹/CyO; Trl13C/ TM6b,Sb ⫻ w ; ⫹/⫹; ⫹/⫹. The male:female ratios were compared as shown in Figure 1c. The following crossing scheme was implemented to test the effects of the H83M2 transgene on the Trl mutant phenotype, w/ Y; H83M2 Trl13C/Trl13C ⫻ w ; Trl ⫺/ TM6b,Sb, where Trl ⫺ denotes Trl13C, Trl 62, or Trl 2.3. Since fertility of Trl13C homozygous males is limited, we used several parallel crosses involving two or three different recombinant isolates between Trl13C and the H83M2-61 or H83M2-87A line to obtain enough progeny to score. The results for each Trl genotype and H83M2 line were then pooled. These crosses produced Trl mutant siblings that differed only in whether or not they carried the H83M2 transgene, thus minimizing the effects of genetic background. Western blots: Extract from ⵑ2.5 third instar larvae was loaded on an 8% polyacrylamide gel (PAGE). Western blotting was performed as described in Deshpande et al. (1995). Blots were probed with 1:3000 dilution of Msl1 (Palmer et al. 1994) antibody as described in Gorman et al. (1995). For a loading control, equal amounts of extract were loaded on a 12% PAGE and probed with 1:15 dilution of anti-snf monoclonal antibody 4G3 (Deshpande et al. 1996) in 5% milk in PBS with 0.1% Triton X-100. HRP-conjugated secondary antibodies were used. Signal was detected using the Lumi-Light (Boehringer Mannheim, Indianapolis) chemiluminescence system. Staining of polytene chromosomes: Larvae were grown at 22⬚ on standard media regularly supplemented with water and yeast. Sex of larvae was determined either by size of gonads or by crossing y⫹ males with y⫺ females. Salivary glands were dissected in PBS with 1% Tween-20 (Sigma, St. Louis) and fixed for 2 min in a drop of solution containing 50% acetic acid, 3.7% formaldehyde, and 1% Tween-20. Chromosomes were spread and stored in PBS with 0.05% Tween-20 at 4⬚. Slides were blocked for 30 min in 5% BSA [for rabbit anti-Msl1 (Palmer et al. 1994) or Msl2 (Kelley et al. 1995) antibodies; dilution of 1:100] or 5% milk ⫹ 5% normal horse serum [for goat anti-Msl2 antibody (Lyman et al. 1997), used for double staining; dilution of 1:50]. Affinity-purified rabbit anti-GAGA581 (Benyajati et al. 1997) at a dilution of 1:50 was used to detect Trl protein. Alexa-488 and -546 secondary antibodies (Molecular Probes, Eugene, OR) were used at a dilution of 1:1000 in milk or BSA with 5% of appropriate normal serum.

GAGA Factor and Dosage Compensation DNA was detected with Hoechst (0.5 ␮g/ml in water). To quantitate the number of autosomal sites, all chromosome spreads with all four major autosomal arms intact were counted on each slide. Four to six slides of each genotype were scored, with two pairs of salivary glands per slide. Chromatin entry sites were visualized using the anti-Msl1 antibody. Salivary glands from w ; Trl13Cmsl3H83M2-6I/⫹ and w ; Trl13Cmsl3H83M2-6I/Trl 2.3msl3 larvae were used to assess the effect of Trl on the distribution of chromatin entry sites. The X chromosomes in both of these classes of larvae came from the same w1 stock. Images were collected using the Zeiss LSM 510 confocal microscope. Statistical tests: For comparison within crosses, such as those illustrated in Figure 2b, the ␹2 test was used with Microsoft Excel. Two-tailed Fisher’s exact test was used to compare male: female ratios between crosses (such as in Figure 2, a and c, and Table 1), computed using a web-based program available at http://www.matforsk.no/ola/fisher.htm. Confidence intervals in Figure 1a were computed using the normal approximation of the binomial distribution.

RESULTS

Trl mutations interfere with MSL-dependent dosage compensation: Trl 13C is the weakest Trl allele. A little more than half of the homozygotes survive to the adult stage (Farkas et al. 1994; Bhat et al. 1996) and these display a variety of other phenotypes. These phenotypes become markedly more severe when Trl13C is trans to the stronger loss-of-function alleles Trl 62 or Trl 2.3. In addition, while there is little difference between male and female viability of Trl13C homozygotes, this is not the case when Trl13C is combined with one of these two stronger Trl alleles. As illustrated in Figure 1a, the viability of Trl13C/Trl 62 females is nearly the same as that of the Trl 13C homozygotes, while there is a significant drop in the viability of Trl 13C/Trl 62 males. An even more pronounced reduction in male viability is observed when the stronger Trl allele Trl 2.3 is combined with Trl 13C. Since sex-specific lethality in flies is usually associated with upsets in X chromosome dosage compensation, one possible explanation for the male lethal effects of these two Trl mutant combinations is that the reduction in Trl activity somehow perturbs the functioning of the msl dosage compensation system. If this is the case, one would predict that the male lethal effects seen in these two Trl mutant combinations would be exacerbated by reducing the dose of the msl genes. In an otherwise wild-type background, mutations in msl1 and msl2 are not haploinsufficient, and heterozygous mutant males show little or no reduction in viability (not shown). However, as indicated in Figure 1b and Table 1, a reduction in the dose of either msl1 or msl2 substantially enhances the male lethality of the Trl13C/Trl 62 mutant combination. By contrast, these msl mutations have little, if any, effect on the viability of Trl13C/Trl 62 females. Similarly, reducing the dose of msl1, msl2, and mle together substantially increases the male lethal effects of the Trl13C/Trl 62 mutant combination (Table 1). Whereas the viability of Trl13C/Trl 62 mutant females carrying (or not carrying)

281

the three MSL-complex mutants is only twofold less than that of sibling females heterozygous for Trl, the viability of Trl13C/Trl 62 males carrying the triple MSL-complex mutant combination is reduced ⬎10-fold. Since four independent chromosomes carrying various combinations of msl mutants were used, the enhancement we observed is unlikely to be due to second-site mutations unrelated to the dosage compensation system, although the precise magnitude of the effect is probably influenced by the genetic background, as seen from the different male viability resulting from the introduction of two independent msl1 alleles (Table 1). Additional support for the idea that a reduction in Trl activity may compromise the dosage compensation system comes from the effects of mutations in the msl complex on Trl13C homozygous males. Normally, Trl 13C homozygous males are as viable as their sibling females (Figure 1a); however, when the dose of msl1, msl2, and mle is simultaneously halved, the viability of the Trl13C homozygous males is significantly reduced compared to their sibling females (Figure 1c). While single mutations in msl1 and msl2 exacerbated the male lethal effects of the Trl13C/Trl 62 mutant combination, no effects were observed for a mutation in mle (Figure 1b). This is most likely due to the fact that the Mle protein is present in excess, while Msl1 and Msl2 are thought to be limiting components of the MSL complex in males (Kelley et al. 1995; Chang and Kuroda 1998). If the male lethality of Trl mutations is enhanced by reducing the amount of the MSL complex, we reasoned that it might be possible to suppress lethality by increasing the amount of complex. For this purpose, we used a transgene that constitutively expresses the Msl2 protein under the control of the hsp83 promoter (referred to as H83M2 transgene; Kelley et al. 1995). Sibling Trl mutant males with or without the transgene were compared to ensure uniformity of genetic background (see materials and methods). We found that the introduction of the H83M2 transgene into the Trl mutant backgrounds improves male viability (Figure 1d). Although small, the effects on male viability were consistent. Moreover, we observed an increase in male viability in two different combinations of Trl alleles with two independent lines of the H83M2 transgene (Figure 1d). Trl mutant males possess a functional MSL complex: One explanation for these gene dose effects is that Trl is required for the expression of a limiting component(s) of the MSL complex. To test this hypothesis, we compared Msl1 and Msl2 protein levels in wild-type and Trl mutant males. As illustrated for Msl1 in Figure 2a, we found that Trl mutations did not detectably alter the levels of either Msl1 or Msl2 proteins. It could be argued that Trl mutations lead to the production of defective MSL complexes either because the expression of some other, perhaps unknown component of the dosage compensation system is substantially

282


Figure 1.—Male viability is compromised by reduced Trl function. (a) Male-to-female ratio in flies carrying adult-viable combinations of Trl alleles. Error bars represent 95% confidence intervals. Numbers above bars represent viability of females compared to Trl heterozygotes. More than 400 individuals of each Trl mutant genotype were scored. The data presented are a compilation of reciprocal crosses in which the two different Trl alleles were inherited from either the mother or the father. Since the same effects on viability were observed when the crosses were performed in either direction (data not shown), the data were pooled. (b) Viability of males wild type or heterozygous for msl1, msl2, or mle in Trl 13C/Trl 62 background relative to female siblings of the same genotype. Additional data are provided in Table 1. The control (msl ⫹) flies were generated from parents carrying the CyO balancer to control for any background effects (see materials and methods for cross details). To ensure statistical significance, at least 100 flies of the appropriate Trl and msl genotype were scored. P values were calculated using the two-tailed Fisher’s exact test. (c) msl1 msl2 mle triple-mutant chromosome induces male-specific lethality in Trl13C homozygous males. Genotypes of flies scored: 1, msl11 msl21 mle1/⫹; Trl 13C/⫹, and 2, msl11 msl21 mle1/⫹; Trl 13C/Trl 13C. The details of the cross are presented in materials and methods. A total of 800 Trl 13C heterozygous and 141 Trl 13C homozygous flies were scored. The P value (shown above the right bar) was calculated using the two-tailed Fisher’s exact test. (d) Viability of Trl 13C/ Trl 62 and Trl 13C/Trl 2,3 males with (solid bars) and without (gradient bars) H83M2 transgene compared to female siblings of the same Trl genotype without the transgene. H83M2-87A and H83M2-6I are different lines of the same transgene (Kelley et al. 1995). Comparison was performed on sibling males (see materials and methods). To ensure that the results are statistically significant, at least 100 Trl mutant flies were scored for each cross. P values were calculated using the ␹2 test.

reduced or because Trl is required for the assembly of functional complexes. If either of these scenarios were correct, then mutations in Trl would be expected to suppress the lethal effects of ectopically expressing Msl2 protein in females from the H83M2 transgene. Contrary to this prediction, we found that Trl mutations enhance rather than suppress the lethal effects of the H83M2 transgene. Under the conditions of the experiment shown in

Figure 2b (growth at 22⬚), females carrying a single copy of the H83M2 transgene that are either wild type or heterozygous for Trl13C are almost fully viable (see legend to Figure 2). In contrast, when they are homozygous for Trl13C, there is a marked reduction in viability. For the H83M2-87A transgene line, viability is reduced to ⬍50% of the Trl13C females lacking the transgene, while it is ⬍10% for the H83M2-6I line. Even more striking reductions in viability are evident in H83M2 females

GAGA Factor and Dosage Compensation

283

TABLE 1 Mutations in msl genes reduce viability of Trl mutant males

Genotype w; w; w; w; w; w;

Male/female ratio

P valuea

Female viability (%)b

0.44 0.22 0.03 0.21 0.52 0.15

NA 0.005 3.3 ⫻ 10⫺10 0.002 0.38 0.00006

61 70 53 42 57 56

⫹/⫹; Trl 13C/Trl 62 msl11/⫹; Trl 13C/Trl 62 msl1L60/1⫹; Trl 13C/Trl 62 msl21/⫹; Trl 13C/Trl 62 mle1/⫹; Trl 13C/Trl 62 msl1⫺ msl2⫺ mle⫺/⫹; Trl 13C/Trl 62

NA, not applicable. Compared to w ; ⫹/⫹; Trl 13C/Trl 62, using two-tailed Fisher’s exact test. To ensure statistical significance, at least 100 Trl mutant flies were scored. For cross details, see materials and methods. b Compared to heterozygous (over TM6b balancer) females; see materials and methods. a

trans-heterozygous for Trl13C and for either Trl 62 or Trl 23. In both of these cases, Trl mutant females carrying the transgene are never observed. The effects on female viability are unlikely to be due to some nonspecific interaction of the transgene with Trl, since female lethality can be rescued by reducing the dose of msl1 by half or by eliminating the function of either msl1 or mle (data not shown). These findings argue that the level and at least the global activity of the MSL complexes is not likely to be compromised by loss of Trl function. Increased number of MSL-complex-bound autosomal sites in Trl mutant males: Another hypothesis is that Trl may be required for association of the MSL complex with a subset of the ⵑ200 sites on the male X chromosome. To explore this possibility, we first compared the distribution of GAGA (Trl) protein and MSL complexes on polytene chromosomes from wild-type males. Double labeling experiments revealed that only 6 of the ⵑ50 X chromosomal sites bound by GAGA also have MSL complexes (Figure 3a). We next compared the distribution of MSL complexes on polytene chromosomes from wild-type and Trl mutant males. For these experiments we used two different Trl mutant combinations. In the first combination, Trl 62/Trl 2.3, we attempted to reduce the level of functional GAGA factor as much as possible. The Trl 62/Trl 2.3 mutant combination is lethal and the animals die during the third instar larval stage. Since the distribution of the MSL complex might be altered nonspecifically in these dying animals, we also used a second Trl mutant combination, Trl 13C/Trl 2.3. Trl 13C/Trl 2.3 third instar larvae are viable and develop much like wild type. Side-by-side comparison of polytenes from the two Trl mutant combinations and wild-type larvae failed to reveal any obvious gaps or striking discontinuities in the distribution of MSL complexes on the mutant X chromosome even at the sites that contain GAGA protein (Figure 3b). While there are no regions of the X chromosome in the Trl mutants in which MSL com-

plexes are entirely missing, our assay is not sufficiently quantitative to detect modest reductions (two- to fourfold) in the amount of MSL complexes associated with the X chromosome or even with specific regions of the X chromosome. That the loading of the MSL complexes onto the X chromosome is, in fact, perturbed in Trl mutant males is suggested by the finding that the number of autosomal sites bound by MSL complexes is increased. This is illustrated in Figure 4, a and b. In wildtype males the number of autosomal sites for the Msl1 protein is 7 ⫾ 1 and for the Msl2 protein is 5 ⫾ 1. In the strong Trl mutant combination, Trl 62/Trl 2.3, the number of autosomal sites for Msl1 is 15 ⫾ 1 and for Msl2 is 11 ⫾ 1. This increase in the number of autosomal sites does not seem to be due to some nonspecific lethal effects in this Trl mutant combination because a similar increase is evident in the Trl13C/Trl 2.3 mutant combination (see Figure 4b). Additional evidence for abnormalities in the distribution of the MSL complexes comes from the presence of MSL complexes at a site in the 3C region of the X chromosome of Trl mutant males that is unoccupied in wild-type males (Figure 3b, arrows). Similar results were obtained when we compared the distribution of MSL complexes in wild-type and Trl ⫺ females that ectopically express Msl2 protein from the H83M2 transgene (Figure 4c). As found in Trl males, the number of autosomal sites is increased. The presence of ectopic MSL sites on the autosomes in Trl mutants raises the possibility that male lethality arises from the hyperactivation of nearby genes. However, this suggestion seems unlikely since Trl13C/Trl 2.3 males rescued by the H83M2 transgene have, if anything, an even larger number of ectopic autosomal sites for Msl2 (15 ⫾ 1) than do Trl13C/Trl 2.3 males lacking the transgene (13 ⫾ 1, Figure 4b). One chromatin entry site is missing in Trl mutant males: Although Trl ⫺ males displayed no striking irregularities in the distribution of MSL complexes on the X chromosome, the presence of complexes at ectopic sites

284


Figure 2.—Trl mutants possess an intact MSL complex. (a) Western blot showing equal levels of Msl1 protein in wild-type and Trl mutant males. The lower part of the filter was cut off and probed with anti-snf antibody (Deshpande et al. 1996). (b) Viability of females of different Trl genotypes carrying either the 87A line (open bars) or the 6I line (solid bars) of the H83M2 transgene. Viability was compared to that of sibling females of the same Trl genotype but without the transgene. At least 100 Trl mutant flies were scored for each cross. All P values were ⬍10⫺3 by Fisher’s exact test. Under our conditions, females that carry a single copy of the H83M2 transgene lines and are either wild type or heterozygous for Trl13C are almost completely viable. There are several reasons why our results differ from those described in Kelley et al. (1995). First, we reared our flies at 22⬚ (see materials and methods), whereas Kelley et al. performed their crosses at 25⬚. When we repeated our experiments at 25⬚, female viability was significantly reduced (data not shown). In addition, Kelley et al. stopped scoring their crosses 6 days after eclosion (see Figure 6e of Kelley et al. 1995), while we continued scoring for 11 days.

on the autosomes and X chromosome suggests that the distribution is abnormal and that some regions of the X chromosome may not have wild-type levels of the complex. This could arise, for example, if there were a defect in the loading of MSL complexes at one or more chromatin entry sites on the X chromosome. To investigate this possibility, we first compared the X chromosome distribution of GAGA and MSL-complex entry sites (Figure 5). Entry sites can be visualized because they are bound by Msl1 and Msl2 proteins in mle, msl3, or mof mutant backgrounds (Gu et al. 1998; Kelley et al.

1999; Meller et al. 2000). Since msl⫺ males produce poor polytene chromosomes, females that express the Msl2 protein in the appropriate msl mutant background are usually used to visualize entry sites. It is thought that the location of the entry sites in the two sexes is the same. Two of the ⵑ35 chromatin entry sites on the X chromosome correspond to the genes encoding roX1 (position 3F) and roX2 (position 10C) RNAs, while the identity of the remaining sites is unknown. Both roX1 and roX2 have several (GA)n/(CT)n motifs that should be recognized by the GAGA factor. Moreover, these motifs are thought to be important for entry site function and are in hypersensitive sites in chromatin from male flies (Park et al. 2003). Thus we expected to observe colocalization of GAGA and Msl complexes at the roX1 and roX2 entry sites. Contrary to these expectations, we could not detect GAGA in the vicinity of roX2. Although GAGA protein could be detected very near to the roX1 entry site, the GAGA signal did not appear to directly overlap with the Msl signal. This would suggest that the GAGA factor does not colocalize with the Msl complex at the roX1 entry site. Of the remaining ⵑ33 entry sites, GAGA protein and the Msl complex appeared to colocalize only at one site, which is at 12DE. In unstretched chromosomes, GAGA protein and the MSL entry site at 12DE overlap; however, in favorable preparations (see Figure 5a) the GAGA band resolves into a tight “doublet” with the MSL complex in between. We next examined the association of the MSL complex with X chromosome entry sites in Trl mutants. We analyzed four chromosomes from four different slides for each genotype. The w1; msl3 83M2-6I Trl13C/msl3 Trl 2.3 female larvae that looked healthy and were of a comparable developmental stage to the w1; msl3 83M2-6I/msl3 Trl 2.3 control females were used. We were able to find a significant number of larvae to perform polytene squashes, presumably because mutations in msl3 rescue the effects of Msl2 expression in females. We found that the binding of MSL complexes to X chromosome entry sites in Trl ⫺ was identical to that in wild type except for the entry site at 12DE, the one that is flanked by GAGA protein. As illustrated in Figure 5b, the 12DE entry site is consistently absent in Trl ⫺ polytene chromosomes. This effect is not due to a polymorphism at this site, since the strains were isogenic for their X chromosomes (see materials and methods). DISCUSSION

Balancing gene expression from the sex chromosomes and autosomes is a critical process in organisms with sex differences in the number of sex chromosomes. In the fruit fly D. melanogaster, one of the mechanisms used to equalize expression levels in the two sexes is the hyperactivation of X-linked genes in male animals. The male dosage compensation system upregulates transcription by modifying the chromatin structure of X-linked genes.


285

Figure 3.—No defects of Msl2 association with the male X chromosome are detectable in Trl mutants. (a) Colocalization of Msl2 (red) and Trl (green) on the male X chromosome. Overlapping bands appear yellow. Examination of well-stretched chromosomes revealed that only the sites indicated by arrows are truly overlapping. These regions are enlarged and shown as insets, with the Trl and Msl2 staining separated. DNA is in blue. (b) Sites that are occupied by both Msl2 and Trl are not affected in Trl mutant males. Two examples of chromosomes from each genotype are presented. Colors are as in a. Arrows indicate the position of the extra site at 3C that appears on X chromosomes from Trl mutant males.

This is accomplished by a special multi-component complex that preferentially localizes to the X chromosome. Complex assembly depends most critically upon Msl1 and Msl2 and these two MSL proteins may provide some type of scaffold for recruiting/stabilizing other components of the complex. These other components include the two noncoding RNAs, roX1 and roX2, which have been implicated in targeting the complex to entry sites on the X chromosome. In addition, the chromatin-modifying enzymes themselves, Mof-1, a histone acetylase, and JIL-1, a tandem histone kinase, the putative helicase, Mle, and the Msl3 protein, are believed to associate with the complex through specific interactions. All but one of these factors appears to function primarily, if not exclusively, in the MSL dosage compensation system. The exception is the JIL-1 kinase that is required not only for proper dosage compensation, but also for other aspects of transcriptional regulation that are vital to both sexes. It would be reasonable to anticipate that other factors like JIL-1 that have crucial activities in dosage compensation while at the same time functioning in other processes that are important for both sexes will be identified. This seems to be true for the GAGA factor that is encoded by the Trl gene. The importance of the GAGA factor in many aspects of gene regulation and chromatin dynamics has been extensively documented in previous studies (Biggin and Tjian 1988; Croston et al. 1991; Kerrigan et al. 1991; Lu et al. 1993; Farkas et al. 1994; Tsukiyama et al. 1994; Bhat et al. 1996; Greenberg and Schedl 2001). Evidence presented here indicates that GAGA also has a more specialized role in X chromosome dosage compensation. First, heteroallelic combinations of weak and strong Trl mutations have much greater effects on male

than on female viability. In fact, the differences in the viability of the sexes in the two heteroallelic Trl mutant combinations tested here are equivalent to, if not more pronounced than, those reported for mutations in jil-1 (Wang et al. 2001). Second, as might be expected if the functioning of the dosage compensation system is compromised when Trl function is impaired, we find that the male lethal effects of the two heteroallelic Trl mutant combinations are exacerbated by reductions in the dose of either msl1 or msl2. Moreover, although the hypomorphic Trl13C allele by itself exhibits no sex-specific lethality when homozygous, male-specific lethality can be induced by reducing the dose of the MSL complex. Conversely, increasing the level of the Msl2 gene product using an hsp83 promoter to drive the expression of an msl2 cDNA partially rescues males carrying Trl mutations. Although Trl appears to function primarily in chromatin remodeling rather than as a dedicated component of the transcriptional machinery, it is involved in the expression of a very large and diverse array of genes. Since males must upregulate transcription of X-linked genes to achieve the same level of expression as females, it would be reasonable to suppose that males are likely to be much more dependent upon the proper functioning of the general transcriptional machinery than are females. Accordingly, any reduction in the activity of a factor critical for transcription would be expected to have considerably more deleterious effects on males than on females. If this idea is correct, then the malespecific lethality of Trl mutations could simply be due to a decline in the overall activity or efficiency of the transcriptional machinery rather than to an effect spe-

286


Figure 4.—Extra autosomal sites occupied by the MSL complex in the Trl mutant background. (a) Msl2 staining (in red; DNA in blue) of polytene chromosomes from wild-type (left) and Trl 62/Trl 2.3 (right) males. Arrows indicate positions of autosomal sites. (b) Numbers of autosomal sites bound by Msl1 (solid bars) and Msl2 (open bars) in wild-type and Trl mutant males. Error bars represent 95% confidence intervals. All P values for the difference in the number of sites between wild type and mutants were ⬍10⫺9. (c) Numbers of autosomal sites bound by Msl2 in Trl 13C heterozygous and Trl 13C/Trl 2.3 females that carry the H83M2-6I line. Slides were scored as in b. Error bars represent 95% confidence intervals.

cific to the process of dosage compensation itself. This “impaired transcription” model would also explain why the male-specific lethal effects of Trl mutations are enhanced by a reduction in the dose of the msl genes and suppressed by increasing the dose of Msl2. Of course, this model predicts that hypomorphic mutations in components of the transcriptional apparatus should also exhibit male-specific lethality like mutations in Trl. Although some alleles of TAF250 do cause preferential male lethality (D. Wassarman, personal communication), there is no evidence that compromising the activity of other general transcription factors affects males more than females. In fact, none of the many hypomorphic mutations in the gene coding for the 140-kD subunit of RNA polymerase II give rise to male lethality (Parkhurst and Ish-Horowicz 1991; M. A. Mortin, S. M. Parkhurst and D. Ish-Horowitz, personal communication), and neither do mutations in the small subunit of TFIIA (Zeidler et al. 1996; M. Mlodzik, personal communication). An additional problem with the “impaired transcription” model is that it would not account for the finding that Trl mutations enhance the female lethal effects of the hsp83:msl2 transgene. The opposite result would be expected, namely that Trl mutations would suppress the female lethal effects of ectopic Msl2 protein. An alternative, and we believe more plausible, explanation for the effects of Trl mutations on male viability

is that the GAGA factor plays some important role in the functioning or activity of the msl-dependent dosage compensation system. In addition to accounting for both the male-specific lethality of Trl mutations and the genetic interactions between Trl and msl-complex genes, this suggestion would help explain two other findings. First, we observed abnormalities in the distribution of the MSL complexes in polytene chromosomes from Trl mutant males. These abnormalities include the presence of at least one ectopic site on the X chromosome and an increase in the number of autosomal sites. This redistribution of MSL complexes argues that the GAGA factor is important for correctly targeting the dosage compensation machinery to the X chromosome. A similar although more dramatic MSL redistribution was observed by Meller and Rattner (2002) when roX1 and roX2 were simultaneously deleted in males. Second, we found that one of the ⵑ35 chromatin entry sites on the X chromosome, at 12DE, is missing in Trl mutants. Unlike any of the other chromatin entry sites observed in polytene chromosomes, GAGA is localized to the 12DE site. This finding argues that the GAGA factor is important in the formation/maintenance of this particular chromatin entry site. Neither of these effects on the chromosomal association of MSL complexes would be explained by a model in which the male-specific lethality of Trl mutations is due to some general reduction in the activity of the transcriptional machinery.


287

Figure 5.—Trl loss of function leads to loss of a single chromatin entry site. (a) Colocalization of Trl (green) and Msl2 (red) on the X chromosome of a female that carries two copies of H83M2-6I and is homozygous for msl3 to visualize the entry sites (Kelley et al. 1999). (b) Absence of the entry site at 12DE in Trl mutants. Entry sites were visualized as in a, except that only one copy of the transgene was present. Only the Trl genotype was different between the lines as indicated on the panel. Msl2 is in red, DNA in blue.

While it would be reasonable to propose that there is a direct connection between the defects in the chromosomal association of Msl complexes and male lethality, the precise mechanism is not entirely clear. One possibility is that male lethality is due to the loss of the 12DE entry site. Supporting this idea, Kelley et al. (1999) have found that MSL complexes formed at ectopic entry sites on the autosomes usually spread only limited distances. This also appears to be true on the X chromosome (Oh et al. 2003). It is thus possible that the two entry sites flanking 12DE, at 12C and 12F, would be unable to compensate completely for the loss of the 12DE site. As a consequence, insufficient levels of the MSL complex would be recruited into the 12C–12F interval in Trl mutant males to fully upregulate gene expression, and this might result in male lethality. Although we believe that the loss of the 12DE entry could significantly impair the upregulation of genes in the 12C–F interval, there are at least two potential complications with this simple model. First, it seems unlikely that a reduction in the level of expression of genes in the 12C–F interval would in itself be sufficient to cause male lethality. Unless this chromosomal interval contains genes specifically required for male viability (e.g., encoding components of the dosage compensation machinery), this model would predict that this same interval is haplo-insufficient in females. However, there is no indication that deletions in this chromosomal interval have significant effects on female viability. Second, while the 12DE entry site is absent (in Trl; msl3 mutants), we do not see any obvious perturbation in the distribution

of MSL complexes in this region of the X chromosome in Trl mutants that are wild type for the MSL genes. One explanation for this discrepancy is that defects in MSL-complex distribution in the vicinity of the 12DE site are obscured because the recruitment and spreading of complexes is much more robust in polytene chromosomes (which consist of hundreds of chromosomes whose sequences are aligned in precise register) than in chromosomes from polyploid or diploid nuclei. In fact, Kelley et al. (1999) have found that in polytene chromosomes MSL complexes can spread from ectopic entry sites on the autosomes not only in cis but also in trans and can even skip over large chromosomal segments. This simple model would also not explain why MSL complexes in polytenes of Trl mutants localize to many ectopic sites on the autosomes. The presence of these autosomal complexes indicates that there must be some defect in the loading of complexes onto the X chromosome. Since we do not see any obvious reduction in the amount of complex in the 12C–F interval, it seems unlikely that the loss of the 12DE entry site alone could account for the presence of the autosomal complexes. Instead, this would suggest that the GAGA factor may be important in the loading or spreading of complexes from a number of entry sites located elsewhere on the X in addition to the 12DE entry site. In this respect, it is notable that the GAGA factor binds in close proximity to five MSL-complex entry sites, including roX1. If GAGA is important in the loading or spreading of complexes from a number of “Trl-dependent” entry sites in addition

288


to 12DE, the male lethal effects of the Trl would be explained by the cumulative effects of a reduction in the expression of genes located in several different chromosomal regions rather than in just the 12C–F interval. If loss of GAGA binding reduces the activity of only a subset of the chromatin entry sites, we would expect that dosage compensation would be compromised over some parts of the X chromosome, but not others. This would help to explain why the weak female lethal effects of the H83M2 transgene at 22⬚ are greatly enhanced in Trl mutants. Under conditions in which Msl2 protein expression is limiting, defects in the spreading of MSL complexes from Trl-dependent entry sites could lead to a more efficient loading and subsequent spreading from sites that are independent of Trl function. Female lethality would be induced because of the increased concentration of complexes in regions served by “Trl-independent” entry sites. A hint that this may happen comes from the study of Oh et al. (2003), who found that elimination of one or the other of the roX chromatin entry sites results in a redistribution of the overexpressed Msl1 and Msl2 proteins to the other parts of the X chromosome. Finally, it is important to note that the effects on MSLcomplex distribution seen in salivary gland polytene chromosomes are unlikely to reproduce the defects in nonpolytene tissues that most directly contribute to male lethality. First, the available evidence suggests that GAGA interacts with different sites in different tissues (Bhat et al. 1996; Greenberg and Schedl 2001). Hence, it is possible that the GAGA factor may be important for the formation or maintenance of entry sites in addition to 12DE in other cell types. Second, because of a substantial maternal contribution, homozygous null Trl alleles are not lethal until larval stages, while the lethal phase of the hypomorphic mutant combinations used in the studies reported here is even later, during the pupal stage. It is likely that lethality occurs when the level of the GAGA factor becomes sufficiently depleted by successive rounds of cell division that it drops below a critical threshold in cells that are essential for viability. Since polytenized cells stop dividing, the level of GAGA factor may remain high enough in these cells to keep the functioning of the dosage compensation system from being seriously impaired. Third, GAGA is known to bind to one of the major satellite sequences in centromeric heterochromatin (Raff et al. 1994). Since centromeric heterochromatin is underreplicated in polytenized chromosomes, the effective concentration of the noneuchromatic GAGA-binding sites in salivary gland cells will be much less than that in cells that do not have polytenized chromosomes. This would mean that, under conditions of limiting GAGA factor, more should be available to associate with the euchromatic regions of X chromosome salivary gland cells than in cells with diploid nuclei. An important question is how the GAGA factor functions in the formation and/or maintenance of the 12DE

entry site. Given its role in generating nucleosome-free regions of chromatin, a plausible idea is that GAGA is required to ensure that the 12DE site is readily accessible to appropriate components of the MSL complex. In Trl mutants, the 12DE entry site and/or the immediately surrounding DNA would be packaged into a nucleosomal structure that cannot be used to initiate MSL-complex assembly, is refractory to the assembly of stable complexes, or is not compatible with the spreading of the complex in cis. The idea that a nucleosome-free region of chromatin is critical for entry site function is supported by recent studies of Kageyama et al. (2001) on the roX1 entry site. They found that a small ⵑ200-bp sequence within the roX1 genes can direct the assembly of MSL complexes at ectopic sites and that this sequence is hypersensitive to nuclease digestion in male but not female chromatin. Further correlating the formation of a nuclease hypersensitive site with the assembly of MSL complexes, Park et al. (2003) have recently shown that roX2 also has an internal sequence that functions to recruit MSL complexes and is hypersensitive in male but not in female flies. In light of the effects of Trl mutations on male viability, it is interesting to note that the internal nuclease hypersensitive site in the roX1 and roX2 genes harbors potential GAGA-factor-binding sites. Moreover, these potential GAGA-binding sites appear to be important for the entry site function of these elements (Park et al. 2003). However, we did not detect GAGA at roX2, and although GAGA does appear to be localized close to roX1, it did not significantly overlap with the roX1 Msl complex in our confocal experiments. In this case, why did we not detect GAGA protein at either the roX1 or the roX2 entry sites in polytene chromosomes? It is possible that the amount of GAGA factor at these two sites is too low for us to detect or that the protein is inaccessible to the GAGA antibody because of the Msl complex. Alternatively, the GAGA factor might be required for the initial formation of the roX1/roX2 hypersensitive sites, but would then be largely displaced when the MSL complexes are assembled. In this case it would be required for establishment, not maintenance. Finally, it is also possible that it is not the GAGA factor but one of the other fly proteins that binds to the (GA)n/(CT)n motifs in roX1 and roX2. Clearly, further studies will be required to resolve this question. We thank R. Kelley and M. Kuroda for providing fly stocks and antibodies generously and promptly. The GAGA-581 antibody was a gift of C. Benyajati. M. A. Mortin, S. M. Parkhurst, D. Ish-Horowitz, M. Mlodzik, and D. Wassarman provided crucial unpublished information. We are grateful to G. Deshpande and J. Huie for critical reading of the manuscript. This work was supported by a National Institutes of Health (NIH) grant to P.S. and NIH training grants to A.J.G. and J.L.Y.

LITERATURE CITED Baker, B. S., M. Gorman and I. Marin, 1994 Dosage compensation in Drosophila. Annu. Rev. Genet. 28: 491–521.

GAGA Factor and Dosage Compensation Bashaw, G. J., and B. S. Baker, 1997 The regulation of the Drosophila msl-2 gene reveals a function for Sex-lethal in translational control. Cell 89: 789–798. Benyajati, C., L. Mueller, N. Xu, M. Pappano, J. Gao et al., 1997 Multiple isoforms of GAGA factor, a critical component of chromatin structure. Nucleic Acids Res. 25: 3345–3353. Bhat, K. M., G. Farkas, F. Karch, H. Gyurkovics, J. Gausz et al., 1996 The GAGA factor is required in the early Drosophila embryo not only for transcriptional regulation but also for nuclear division. Development 122: 1113–1124. Biggin, M. D., and R. Tjian, 1988 Transcription factors that activate the Ultrabithorax promoter in developmentally staged extracts. Cell 53: 699–711. Bone, J. R., J. Lavender, R. Richman, M. J. Palmer, B. M. Turner et al., 1994 Acetylated histone H4 on the male X chromosome is associated with dosage compensation in Drosophila. Genes Dev. 8: 96–104. Chang, K. A., and M. I. Kuroda, 1998 Modulation of MSL1 abundance in female Drosophila contributes to the sex specificity of dosage compensation. Genetics 150: 699–709. Cline, T. W., and B. J. Meyer, 1996 Vive la difference: males vs females in flies vs worms. Annu. Rev. Genet. 30: 637–702. Croston, G. E., L. A. Kerrigan, L. M. Lira, D. R. Marshak and J. T. Kadonaga, 1991 Sequence-specific antirepression of histone H1-mediated inhibition of basal RNA polymerase II transcription. Science 251: 643–649. Deshpande, G., J. Stukey and P. Schedl, 1995 scute (sis-b) function in Drosophila sex determination. Mol. Cell. Biol. 15: 4430–4440. Deshpande, G., M. E. Samuels and P. D. Schedl, 1996 Sex-lethal interacts with splicing factors in vitro and in vivo. Mol. Cell. Biol. 16: 5036–5047. Farkas, G., J. Gausz, M. Galloni, G. Reuter, H. Gyurkovics et al., 1994 The Trithorax-like gene encodes the Drosophila GAGA factor. Nature 371: 806–808. Franke, A., and B. S. Baker, 1999 The roX1 and roX2 RNAs are essential components of the compensasome, which mediates dosage compensation in Drosophila. Mol. Cell 4: 117–122. Gorman, M., A. Franke and B. S. Baker, 1995 Molecular characterization of the male-specific lethal-3 gene and investigations of the regulation of dosage compensation in Drosophila. Development 121: 463–475. Granok, H., B. A. Leibovitch, C. D. Shaffer and S. C. Elgin, 1995 Chromatin. Ga-ga over GAGA factor. Curr. Biol. 5: 238–241. Greenberg, A. J., and P. Schedl, 2001 GAGA factor isoforms have distinct but overlapping functions in vivo. Mol. Cell. Biol. 21: 8565–8574. Gu, W., P. Szauter and J. C. Lucchesi, 1998 Targeting of MOF, a putative histone acetyl transferase, to the X chromosome of Drosophila melanogaster. Dev. Genet. 22: 56–64. Henry, R. A., B. Tews, X. Li and M. J. Scott, 2001 Recruitment of the male-specific lethal (MSL) dosage compensation complex to an autosomally integrated roX chromatin entry site correlates with an increased expression of an adjacent reporter gene in male Drosophila. J. Biol. Chem. 276: 31953–31958. Hilfiker, A., D. Hilfiker-Kleiner, A. Pannuti and J. C. Lucchesi, 1997 mof, a putative acetyl transferase gene related to the Tip60 and MOZ human genes and to the SAS genes of yeast, is required for dosage compensation in Drosophila. EMBO J. 16: 2054–2060. Jin, Y., Y. Wang, D. L. Walker, H. Dong, C. Conley et al., 1999 JIL-1: a novel chromosomal tandem kinase implicated in transcriptional regulation in Drosophila. Mol. Cell 4: 129–135. Jin, Y., Y. Wang, J. Johansen and K. M. Johansen, 2000 JIL-1, a chromosomal kinase implicated in regulation of chromatin structure, associates with the male specific lethal (MSL) dosage compensation complex. J. Cell Biol. 149: 1005–1010. Kageyama, Y., G. Mengus, G. Gilfillan, H. G. Kennedy, C. Stuckenholz et al., 2001 Association and spreading of the Drosophila dosage compensation complex from a discrete roX1 chromatin entry site. EMBO J. 20: 2236–2245.

289

Kelley, R. L., and M. I. Kuroda, 2000 The role of chromosomal RNAs in marking the X for dosage compensation. Curr. Opin. Genet. Dev. 10: 555–561. Kelley, R. L., I. Solovyeva, L. M. Lyman, R. Richman, V. Solovyev et al., 1995 Expression of msl-2 causes assembly of dosage compensation regulators on the X chromosomes and female lethality in Drosophila. Cell 81: 867–877. Kelley, R. L., J. Wang, L. Bell and M. I. Kuroda, 1997 Sex lethal controls dosage compensation in Drosophila by a non-splicing mechanism. Nature 387: 195–199. Kelley, R. L., V. H. Meller, P. R. Gordadze, G. Roman, R. L. Davis et al., 1999 Epigenetic spreading of the Drosophila dosage compensation complex from roX RNA genes into flanking chromatin. Cell 98: 513–522. Kerrigan, L. A., G. E. Croston, L. M. Lira and J. T. Kadonaga, 1991 Sequence-specific transcriptional antirepression of the Drosophila Kruppel gene by the GAGA factor. J. Biol. Chem. 266: 574–582. Lu, Q., L. L. Wallrath, H. Granok and S. C. Elgin, 1993 (CT)n (GA)n repeats and heat shock elements have distinct roles in chromatin structure and transcriptional activation of the Drosophila hsp26 gene. Mol. Cell. Biol. 13: 2802–2814. Lucchesi, J. C., and J. E. Manning, 1987 Gene dosage compensation in Drosophila melanogaster. Adv. Genet. 24: 371–429. Lyman, L. M., K. Copps, L. Rastelli, R. L. Kelley and M. I. Kuroda, 1997 Drosophila male-specific lethal-2 protein: structure/function analysis and dependence on MSL-1 for chromosome association. Genetics 147: 1743–1753. Meller, V. H., and B. P. Rattner, 2002 The roX genes encode redundant male-specific lethal transcripts required for targeting of the MSL complex. EMBO J. 21: 1084–1091. Meller, V. H., P. R. Gordadze, Y. Park, X. Chu, C. Stuckenholz et al., 2000 Ordered assembly of roX RNAs into MSL complexes on the dosage-compensated X chromosome in Drosophila. Curr. Biol. 10: 136–143. Oh, H., Y. Park and M. I. Kuroda, 2003 Local spreading of MSL complexes from roX genes on the Drosophila X chromosome. Genes Dev. 17: 1334–1339. Palmer, M. J., V. A. Mergner, R. Richman, J. E. Manning, M. I. Kuroda et al., 1993 The male-specific lethal-one (msl-1) gene of Drosophila melanogaster encodes a novel protein that associates with the X chromosome in males. Genetics 134: 545–557. Palmer, M. J., R. Richman, L. Richter and M. I. Kuroda, 1994 Sexspecific regulation of the male-specific lethal-1 dosage compensation gene in Drosophila. Genes Dev. 8: 698–706. Pannuti, A., and J. C. Lucchesi, 2000 Recycling to remodel: evolution of dosage-compensation complexes. Curr. Opin. Genet. Dev. 10: 644–650. Park, Y., G. Mengus, X. Bai, Y. Kageyama, V. H. Meller et al., 2003 Sequence-specific targeting of Drosophila roX genes by the MSL dosage compensation complex. Mol. Cell 11: 977–986. Parkhurst, S. M., and D. Ish-Horowicz, 1991 wimp, a dominant maternal-effect mutation, reduces transcription of a specific subset of segmentation genes in Drosophila. Genes Dev. 5: 341–357. Raff, J. W., R. Kellum and B. Alberts, 1994 The Drosophila GAGA transcription factor is associated with specific regions of heterochromatin throughout the cell cycle. EMBO J. 13: 5977–5983. Tsukiyama, T., P. B. Becker and C. Wu, 1994 ATP-dependent nucleosome disruption at a heat-shock promoter mediated by binding of GAGA transcription factor. Nature 367: 525–532. Wang, Y., W. Zhang, Y. Jin, J. Johansen and K. M. Johansen, 2001 The JIL-1 tandem kinase mediates histone H3 phosphorylation and is required for maintenance of chromatin structure in Drosophila. Cell 105: 433–443. Wilkins, R. C., and J. T. Lis, 1997 Dynamics of potentiation and activation: GAGA factor and its role in heat shock gene regulation. Nucleic Acids Res. 25: 3963–3968. Zeidler, M. P., K. Yokomori, R. Tjian and M. Mlodzik, 1996 Drosophila TFIIA-S is up-regulated and required during Ras-mediated photoreceptor determination. Genes Dev. 10: 50–59. Communicating editor: K. Anderson