Title: Complex formation by the Drosophila MSL proteins: role of the MSL2 RING finger in protein complex assembly
Abstract: Article15 September 1998free access Complex formation by the Drosophila MSL proteins: role of the MSL2 RING finger in protein complex assembly Kyle Copps Kyle Copps Program in Cell and Molecular Biology, Houston, TX, 77030 USA Search for more papers by this author Ron Richman Ron Richman Department of Cell Biology, Houston, TX, 77030 USA Search for more papers by this author Laura M. Lyman Laura M. Lyman Howard Hughes Medical Institute, Baylor College of Medicine, Houston, TX, 77030 USA Search for more papers by this author Kimberly A. Chang Kimberly A. Chang Department of Cell Biology, Houston, TX, 77030 USA Howard Hughes Medical Institute, Baylor College of Medicine, Houston, TX, 77030 USA Search for more papers by this author Joanne Rampersad-Ammons Joanne Rampersad-Ammons Department of Cell Biology, Houston, TX, 77030 USA Search for more papers by this author Mitzi I. Kuroda Corresponding Author Mitzi I. Kuroda Program in Cell and Molecular Biology, Houston, TX, 77030 USA Department of Cell Biology, Houston, TX, 77030 USA Howard Hughes Medical Institute, Baylor College of Medicine, Houston, TX, 77030 USA Search for more papers by this author Kyle Copps Kyle Copps Program in Cell and Molecular Biology, Houston, TX, 77030 USA Search for more papers by this author Ron Richman Ron Richman Department of Cell Biology, Houston, TX, 77030 USA Search for more papers by this author Laura M. Lyman Laura M. Lyman Howard Hughes Medical Institute, Baylor College of Medicine, Houston, TX, 77030 USA Search for more papers by this author Kimberly A. Chang Kimberly A. Chang Department of Cell Biology, Houston, TX, 77030 USA Howard Hughes Medical Institute, Baylor College of Medicine, Houston, TX, 77030 USA Search for more papers by this author Joanne Rampersad-Ammons Joanne Rampersad-Ammons Department of Cell Biology, Houston, TX, 77030 USA Search for more papers by this author Mitzi I. Kuroda Corresponding Author Mitzi I. Kuroda Program in Cell and Molecular Biology, Houston, TX, 77030 USA Department of Cell Biology, Houston, TX, 77030 USA Howard Hughes Medical Institute, Baylor College of Medicine, Houston, TX, 77030 USA Search for more papers by this author Author Information Kyle Copps1, Ron Richman2, Laura M. Lyman3, Kimberly A. Chang2,3, Joanne Rampersad-Ammons2 and Mitzi I. Kuroda 1,2,3 1Program in Cell and Molecular Biology, Houston, TX, 77030 USA 2Department of Cell Biology, Houston, TX, 77030 USA 3Howard Hughes Medical Institute, Baylor College of Medicine, Houston, TX, 77030 USA *Corresponding author. E-mail: [email protected] The EMBO Journal (1998)17:5409-5417https://doi.org/10.1093/emboj/17.18.5409 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Drosophila MSL proteins are thought to act within a complex to elevate transcription from the male X chromosome. We found that the MSL1, MSL2 and MSL3 proteins are associated in immunoprecipitations, chromatographic steps and in the yeast two-hybrid system, but that the MLE protein is not tightly complexed in these assays. We focused our analysis on the MSL2–MSL1 interaction, which is postulated to play a critical role in MSL complex association with the X chromosome. Using a modified two-hybrid assay, we isolated missense mutations in MSL2 that disrupt its interaction with MSL1. Eleven out of 12 mutated residues clustered around the first zinc-binding site of the RING finger domain were conserved in a Drosophila virilis MSL2 homolog. Two pre-existing msl2 alleles, which fail to support male viability in vivo, have lesions in the same region of the RING finger. We tested these in the two-hybrid system and found that they are also defective in interaction with MSL1. Mutation of the second zinc-binding site had little effect on MSL1 binding, suggesting that this portion of the RING finger may have a distinct function. Our data support a model in which MSL2–MSL1 interaction nucleates assembly of an MSL complex, with which MLE is weakly or transiently associated. Introduction In many species, females develop with two X chromosomes, while males have only one. This imbalance has a potentially lethal effect that is countered by the evolution of various forms of X-linked gene dosage compensation. In the fruit fly Drosophila, dosage compensation is accomplished by ∼2-fold hypertranscription of most genes on the male X chromosome (reviewed in Lucchesi and Manning, 1987; Baker et al., 1994). Several genes required for this phenomenon were identified in screens for mutations with recessive, male-specific lethal (msl) phenotypes (Fukunaga et al., 1975; Belote and Lucchesi, 1980; Uchida et al., 1981; Lucchesi et al., 1982; Hilfiker et al., 1997). The 'msls' include msl1, msl2, msl3, maleless (mle) and males absent on the first (mof). Each gene has been cloned and the encoded proteins have been characterized to varying degrees. MSL1 is a novel acidic protein (Palmer et al., 1993); MSL2, a putative zinc-binding protein (Bashaw and Baker, 1995; Kelley et al., 1995; Zhou et al., 1995); MSL3, a chromodomain protein (Gorman et al., 1995; Koonin et al., 1995); MLE, an RNA or DNA helicase of the DExH subfamily of ATPase/helicases (Kuroda et al., 1991; Lee et al., 1997); and MOF, a probable acetyltransferase (Hilfiker et al., 1997). Abundant evidence suggests that the MSL proteins act within a multi-subunit complex (reviewed in Baker et al., 1994; Kelley and Kuroda, 1995; Cline and Meyer, 1996; Lucchesi, 1996). Immunofluorescence studies have shown that the MSLs co-localize at hundreds of discrete sites on the X chromosome in male somatic cells (Bone et al., 1994; Gorman et al., 1995; Kelley et al., 1995; Gu et al., 1998). Localization of each protein in the wild-type pattern requires all five msl+ functions, suggesting that complex formation is a prerequisite for their association with most X chromosome sites. In support of this idea, both MSL1 and MSL3 appear to depend on interaction with the other proteins for stability (Palmer et al., 1994; Gorman et al., 1995; Kelley et al., 1995), and MSL1 and MSL2 have been co-immunoprecipitated (Kelley et al., 1995). There is indirect evidence for the existence of one or more RNA components of the MSL complex. Two non-coding RNAs termed roX1 and roX2 were shown to require the MSLs for their male-specific accumulation, and roX1 was shown specifically to coat the male X chromosome (Amrein and Axel, 1997; Meller et al., 1997). Although a requirement for the roX RNAs in male viability has not been demonstrated, it is possible that their role in dosage compensation is masked by functional redundancy. In addition, it is noteworthy that MLE, the Drosophila homolog of human RNA helicase A, is removed specifically from the X chromosome by treatment of polytene chromosomes with RNase A (Richter et al., 1996). The MSLs appear to function by a mechanism involving the modification of chromatin structure. In the presence of the wild-type MSLs, histone H4 that is mono-acetylated on Lys16 (H4Ac16) is associated preferentially with the X chromosome (Turner et al., 1992; Bone et al., 1994; Hilfiker et al., 1997). A link between histone acetylation and gene transcription was demonstrated recently by the finding that several proteins defined as transcriptional coactivators possess histone acetyltransferase activity (Brownell et al., 1996; Kuo et al., 1996; Ogryzko et al., 1996). Thus, a major function of the MSL complex may be to target the putative acetylase MOF to X chromatin (Hilfiker et al., 1997; Gu et al., 1998). Interaction between MSL2 and MSL1 may serve an initiating role in the association of the MSLs with the X chromosome. First, ectopic expression of the male-specific MSL2 protein in females appears to stabilize MSL1, and causes the assembly of functional MSL complexes on their X chromosomes (Kelley et al., 1995). Secondly, the MSL2 and MSL1 proteins remain co-dependently co-localized at ∼30 sites on the X chromosome in the absence of either MSL3, MLE or MOF (Lyman et al., 1997; Gu et al., 1998). Thus, MSL2 and MSL1 may comprise the chromatin-binding activity of the MSL complex, or may be the first of the known subunits to assemble at pre-defined sites. The MSL2 protein contains a RING finger (C3HC4) domain. The RING finger is present in >60 proteins from plants, bacteria, animals and viruses, many of which are present in macromolecular complexes (reviewed in Freemont, 1993; Saurin et al., 1996). In some proteins, such as the breast cancer protein BRCA1, mutations in the RING finger are correlated with human cancers (Shattuck-Eidens et al., 1995). Solution 1H-NMR structures have been obtained for the equine herpesvirus gene 63 protein (EHV-63) and the human acute promyelocytic leukemia proto-oncoprotein (PML) (Barlow et al., 1994; Borden et al., 1995). In each case, two zinc atoms are coordinated by interleaved pairs of cysteines, or cysteine plus histidine, in a 1–3, 2–4 pattern. A comparison of the EHV-63 and PML domains revealed significant structural differences, suggesting that the RING finger may serve as a scaffold for the evolution of different functions (Borden et al., 1995). A requirement for the MSL2 RING finger in male viability was demonstrated by site-directed mutation of conserved RING finger cysteines (Lyman et al., 1997); however, its function, whether in binding to DNA or to other proteins, has not been established. In this report, we provide direct evidence of complex formation by the MSLs and explore the role of the MSL2 RING finger in this process. We find that the RING finger domain of MSL2 binds the MSL1 protein, and that residues around the first zinc-binding site of the RING finger are critical for this interaction. In addition, we find that MSL3 is tightly complexed with MSL2–MSL1 through interaction with MSL1, but that MLE may be only weakly, or transiently, associated with these proteins. These findings are discussed in relation to current models of dosage compensation and RING finger function. Results MSL1, MSL2 and MSL3 are tightly complexed in vivo The MSL1 and MSL2 proteins previously were reported to co-immunoprecipitate (co-IP) (Kelley et al., 1995). We have extended this analysis to include MSL3 and MLE using extracts from Drosophila Schneider line 2 (SL2) cells. The male character of SL2 cells was documented previously by the presence of male-specific transcripts of the alternatively spliced sex-determination genes Sxl and tra (Ryner and Baker, 1991). We found that these cells also express the MSLs, including the male-specific protein MSL2 (Figure 1A–C). As in male larvae, the MSL-1 and MSL-2 proteins are restricted to a portion of the chromatin in nuclei (presumably the X), while MLE exhibits some general, possibly autosomal, staining in addition to its X-specific association (Kuroda et al., 1991). Assembly of the MSLs into apparently functional complexes in SL2 cells is demonstrated by the subnuclear co-localization of MSL1 and H4Ac16, a marker of dosage-compensated chromatin (Figure 1D–F). Figure 1.Immunostaining and co-immunoprecipitation of the MSLs from Drosophila SL2 cells. (A–C) Separate fields showing cells stained for (A) MSL1, (B) MSL2 or (C) MLE (red) and counterstained with Hoechst 33258 to reveal DNA (blue). In each case, the MSL proteins are localized predominantly to single sites within nuclei, compatible with X chromosome binding. MLE also exhibits some general, possibly autosomal, staining. White arrows indicate association of MSL-1 and MSL-2 with mitotic chromosomes. (D–F) A single field stained for (D) MSL-1 or (E) H4Ac16, a marker of the dosage-compensated X chromosome in male flies. A double exposure micrograph of the same field (F) shows the co-localization of the two signals. (G) Co-IP Western blots. SL2 cell nuclear extract was incubated with immune (Im) and pre-immune (Pre) sera against each of the MSLs. Immunoprecipitates from 200 μg of extract (lanes 2–9) were probed for the presence of individual proteins on separate blots; lane 1 contains 40 μg of the extract. Only MLE was not co-precipitated significantly by antibodies against MSL1, MSL2 and MSL3; similar results were seen with extracts prepared from third instar male larvae by the same method (not shown). Download figure Download PowerPoint For immunoprecipitation reactions, we incubated SL2 cell nuclear extract with immune and pre-immune sera against MSL1, MSL2, MSL3 and MLE (Figure 1G); extracts from third instar male larvae gave qualitatively equivalent results (data not shown). A significant portion of the total MSL1, MSL2 and MSL3 in SL2 extract was precipitated by antisera against any one of these proteins, demonstrating their tight association in vivo (Figure 1G, lanes 2, 4 and 6). In contrast, the fraction of MSL1, MSL2 and MSL3 precipitated by anti-MLE serum was small relative to the amount of MLE precipitated (Figure 1G, lane 8), suggesting that only a fraction of the total MLE was complexed with the other proteins. The simplest explanation for this finding is that there is a substantial pool of free MLE in nuclei in vivo; however, we cannot exclude the possibility that MLE is removed preferentially from the MSL complex during its extraction from chromatin. We further characterized the soluble MSL complex from Schneider cells using gel filtration (Superose 12) and anion exchange (Mono Q) chromatography (Figure 2A and B). Consistent with our IP data, the MSL1, MSL2 and MSL3 proteins co-eluted from both columns, while the bulk of MLE appeared to run as a monomer. The elution of MSL1, MSL2 and MSL3 from Superose 12 chromatography was compatible with their presence in a complex with an Mr of >1 MDa, while the bulk of MLE (143 kDa) co-eluted with a 150 kDa marker protein. In subsequent analyses using Superose 6 chromatography, we found that the MSL complex co-elutes with the Drosophila brahma (BRM) complex (Mr ∼2 MDa; data not shown). We also found that MSL1, MSL2 and MSL3 could be co-precipitated from their peak Mono Q fractions, suggesting that the chromatography was not generally disruptive to protein interaction (data not shown). Thus, the relative abundance of uncomplexed MLE in nuclear extracts may indicate that the association of MLE with the MSL complex is weaker than that of the other MSLs. Figure 2.MSL1, MSL2 and MSL3 co-fractionate on gel filtration and anion exchange chromatography. (A) Superose 12 chromatography. Increasing time (in fraction number) is indicated by the scale; the elution times of calibrating marker proteins (apoferritin, 443 kDa; alcohol dehydrogenase, 150 kDa; cytrochrome c, 14 kDa) are denoted by arrows. MSL1, MSL2 and MSL3 elute as a complex with an Mr >1 MDa. In contrast, elution of MLE (mol. wt 143 kDa) peaked in fraction 14 with the 150 kDa alcohol dehydrogenase marker. (B) Mono Q chromatography. MSL1 and MSL2 are predicted to have an acidic pI (∼5.5) and are strongly retained by Mono Q resin at pH 8.0. Elution of the more neutral MSL3 protein (predicted pI ∼7) peaks in the same fractions as MSL1 and MSL2, while the bulk of MLE (predicted pI ∼7) is eluted in lower salt (Ext., Schneider extract; FT, unbound, or flowthrough, fraction). Download figure Download PowerPoint MSL2 and MSL3 interact with MSL1 in the two-hybrid system To begin molecular dissection of the MSL complex, we assayed for interaction of MSL1, MSL2, MSL3 and MLE in the yeast two-hybrid system. Initially, we made fusions of the full-length proteins to the Gal4 DNA-binding domain (DB) and activation domain (AD) and tested for their ability to activate GAL–lacZ transcription in yeast. The DB–MSL1 and DB–MSL2 fusions strongly activated transcription of the GAL–lacZ reporter in the absence of AD-tagged partners. Therefore, these fusions were not informative for interactions with other proteins. The DB–MSL3 fusion also weakly activated transcription of the reporter; however, interactions with this protein could still be scored by the enhancement of lacZ activity in the presence of strongly interacting partners. Finally, DB–MLE did not activate the reporter in the absence of a partner. To permit two-hybrid assay of MSL1–MSL2 interaction in at least one of the two possible configurations, we examined the effect of C-terminal deletions in these proteins on their ability to potentiate GAL–lacZ transcription. As a series of truncated DB–MSL1 fusions retained the ability to activate the reporter (not shown), we constructed and used a non-activating MSL2 fusion [DB–MSL2 (1–190)], which contains only the N-terminal 190 MSL2 residues, including the RING finger domain (amino acids 37–87). In the complete set of pairwise assays (Figure 3A), both DB–MSL2 (1–190) and DB–MSL3 interacted with AD–MSL1. In contrast, the DB–MLE fusion did not interact with MSL1, MSL2 or MSL3 fusions, but did interact with AD–MLE. It is unclear if the interaction of MLE with itself is relevant to its function in vivo; however, the interaction of MSL2 and MSL3 with MSL1 is consistent with our co-IP data and with the co-purification of these three proteins. In addition, this result is compatible with a network of presumed stability effects in which MSL2 stabilizes MSL1, and both MSL2 and MSL1 are required for the stability of MSL3 (Baker et al., 1994; Gorman et al., 1995; Kelley et al., 1995). Therefore, the two-hybrid interactions between these proteins probably represent their actual interactions at sites on the X chromosome in vivo. Figure 3.Two-hybrid interactions between the MSL proteins. Minus and plus signs indicate the absence or presence of activity of a chromosomal GAL–lacZ reporter; multiple pluses indicate greater relative activity in side-by-side comparisons. (A) Pairwise interaction tests. The full-length proteins or a deletion (MSL2) were fused to the Gal4 DB and AD in the expression plasmids pAS1 and pACT2, and tested for interaction in a mating assay. Informative pairs are in shaded boxes. As a control, the DB–MSL plasmids were mated to the empty pACT2 vector, which expresses the Gal4 AD only. (B) Amino acids 85–186 of MSL1 are sufficient for interaction with MSL2 in the two-hybrid system. The MSL1 protein is depicted as a box, with clusters of negatively charged aspartate and glutamate residues indicated by shading. A segment encoding residues 48–321 was selected for interaction with MSL2 from a library of msl1 fragments. The indicated fusions to the Gal4 AD (small boxes) were used to map the interaction with MSL2 as above. None of the fusions activated GAL–lacZ transcription in the absence of DB–MSL2 (1–190). Download figure Download PowerPoint We screened a library composed of restriction fragments from the msl1 cDNA to identify a region of MSL1 sufficient for interaction with DB–MSL2 (1–190). Several strongly interacting clones isolated from the screen were found to contain an identical Sau3A1 fragment encoding amino acids 48–321 of the MSL1 open reading frame (Figure 3B). Subsequent assay of overlapping subclones within this ∼1 kb fragment showed that amino acids 85–186 of MSL1 were sufficient for interaction with MSL2 in the two-hybrid system; the same region was not sufficient for interaction with MSL3 or with other DB–tagged proteins. A search of protein databases using amino acids 85–186 of MSL1 yielded no significant homologies. The MSL2 RING finger domain is required for interaction with MSL1 To identify specific MSL2 residues necessary for interaction with MSL1, we developed a modified two-hybrid assay that allows selection against protein interaction. In this 'reverse two-hybrid' scheme (Figure 4), yeast cells in which a mutant DB–MSL2 (1–190) protein does not interact with AD–MSL1 fail to activate transcription of a GAL–URA3 reporter, and are thereby made resistant to 5–fluoroorotic acid (5-FOA). A secondary screen of GAL–lacZ reporter activity and other assays (steps 1–4, Figure 4) then confirm that the loss of interaction is due to a missense mutation in MSL2. Variations on this method have been used previously to map protein interactions within the yeast Ste5 protein, and to isolate dominant-negative mutations in the human p53 protein (Brachmann et al., 1996; Inouye et al., 1997a). Figure 4.Scheme for selection of interaction-disruptive mutations in MSL2. Yeast expressing the wild-type AD–MSL1 fusion and a mutagenized DB–MSL2 (1–190) protein (MSL2*) were plated on minimal media containing 5-FOA. Two reporter genes were integrated into the yeast chromosomes. Cells in which interaction of the proteins was preserved ('wild type') activated the GAL–URA3 reporter and were FOA-sensitive (FoaS). Cells in which the interaction was disrupted (mutant) failed to activate GAL–URA3 transcription, were FOA-resistant (FoaR), and lived to produce white colonies in a secondary screen of GAL–lacZ reporter activity. Subsequent assays confirmed that: (i) colonies contained the DB–MSL2 (1–190) plasmid; (ii) an MSL2 fusion protein was made; and (iii) loss of interaction was not due to defects in the MSL1 fusion or in the yeast host cell. Download figure Download PowerPoint Libraries of mutations in the DB–MSL2 (1–190) fusion were generated by subcloning the products of mutagenic PCR reactions into the unmutated DB–MSL2 (1–190) plasmid, ensuring that no mutations were made in the Gal4 DB. From the screening of two small libraries, we isolated 19 plasmids that encoded a stable MSL2 fusion protein incapable of interaction with MSL1. While the majority of plasmids carried more than one amino acid substitution, we were able to identify a single interaction-disruptive mutation in all but one case by subcloning and retesting the mutations from seven of the double mutant plasmids. In all, this analysis yielded 12 mutations in 11 different codons that disrupted binding to MSL1, as well as seven mutations which, by themselves, had no effect (Figure 5A). Figure 5.Results of reverse two-hybrid screening. All mutations were isolated by 5-FOA selection except the fly alleles (ΔV43 and P51L) (Zhou et al., 1995) and the site-directed mutations (C59A, C78A, E82P G83L, E82A G83A and S85K D86G) (Lyman et al., 1997). (A) Western blots and lacZ activity assays for interaction-disruptive (left) and non-disruptive mutations (right). Extracts from yeast expressing wild-type MSL1 and the indicated MSL2 mutants were probed for the ∼37 kDa MSL2 fusion protein; as a control, identical blots were probed with antibodies to γ-tubulin (∼55 kDa). With the exception of the L32P mutant, the various MSL2 mutant proteins were produced at similar levels, regardless of their ability to bind MSL1. After a 24 h incubation in the presence of X-gal, cells in which MSL2–MSL1 interaction is intact are stained blue, while cells in which the interaction is disturbed appear pink due to starvation for adenine. (B) Mutations that disrupt interaction of MSL2 and MSL1 cluster about the first zinc-binding site of the MSL2 RING finger. Interaction-disruptive mutations are indicated by red circles, non-disruptive mutations by green rings; the wild-type and mutant residues are separated by a slash; double mutations are indicated by parentheses. A consensus sequence for MSL2-like RING fingers (Zhou et al., 1995) is shown beneath the protein, with mutations in conserved positions in bold type (ø = hydrophobic residues). Download figure Download PowerPoint We noted a striking tendency of the disruptive mutations to cluster about the first zinc-binding site (Z1) of the RING finger (Figure 5). Mutations were found at several conserved positions within the region, including three of the four cysteines thought to coordinate zinc at Z1. Interestingly, two of the three msl2 mutant alleles sequenced by Zhou et al. (1995) also cluster in this region. The msl2γ136 and msl21 alleles, which fail to support male viability in vivo, carry mutations of residues near Z1 of the RING finger (deletion of V43 and P51L, respectively). To determine whether the encoded proteins would also fail to interact with MSL1, we tested both alleles in our reverse two-hybrid assay (Figure 5A). Both mutations specifically disrupted interaction with MSL1, while random point mutations from our PCR mutagenesis did not. In each case, a stable protein was produced in yeast. Therefore, we conclude that one essential function of the RING finger in MSL2 is to mediate interaction with MSL1. However, further analysis showed that the RING finger is functionally complex. Despite the 'cross-braced' arrangement of the RING finger zinc-binding sites, we found no interaction-disruptive mutations within the second site (Z2). We previously reported that mutations in two of the Z2 cysteines (C59 and C78; Figure 5A) abolish msl2+ function in vivo (Lyman et al., 1997). When tested for interaction in yeast (Figure 5A), fusion proteins carrying these mutations bound to MSL1 as well as the wild-type MSL2 protein. Thus, we propose that the RING finger domain performs at least two functions in MSL2: the first zinc-binding site is involved in MSL1 binding, while the second site may be required for a separate activity. Unfortunately, analysis of this model in vivo is complicated by the fact that non-functional MSL proteins often do not persist in the fly, severely hampering efforts to correlate mutation with the loss or retention of specific protein–protein interactions. Indeed, the protein encoded by the msl21 allele is not detectable in male larval extracts (Lyman et al., 1997), but is stable when expressed in yeast (Figure 5A). The RING finger domain is conserved in MSL2 of Drosophila virilis Previous studies have shown that the MSL components of dosage compensation are conserved between Drosophila melanogaster and Drosophila virilis (Bone and Kuroda, 1996; Marín et al., 1996). These species diverged ∼60 million years ago (Patterson and Stone, 1952). We used oligonucleotides to the RING finger region of the D.melanogaster msl2 gene to amplify a portion of the D.virilis homolog; this PCR product was then used as a probe to obtain a full-length msl2 clone from a D.virilis genomic library. An alignment of the conceptual translation products of the D.virilis and D.melanogaster genes (Figure 6A and B) reveals the excellent conservation (74% identity/83% similarity) of the RING finger domain and surrounding region (amino acids 1–115). All but one of the residues found to be important for MSL1 binding by reverse two-hybrid screening are conserved in the D.virilis protein; the single exception, S53, is adjacent to a conserved proline (P54) and was itself mutated to proline in the screen. Four cysteine or histidine residues outside the RING domain are also conserved. Although these are not positioned to form a canonical zinc finger structure, mutation of one of the residues (C107) to arginine disrupted the interaction of MSL2 with MSL1. Figure 6.Alignment of D.melanogaster and D.virilis MSL2. (A) Identical and similar residues are highlighted by black and gray shading, respectively. Bars above the alignment indicate the position of the RING finger and second cysteine-rich domains; caret marks (ˆ)indicate the position of interaction-disruptive mutations in the RING finger. Dots (●) mark the positions of conserved cysteines/histidines; open circles (○) are cysteines/histidines outside the published RING finger and metallothionein alignments. Polymorphisms (*) in the D.melanogaster protein, including a four amino acid insertion (/ \), lie mainly within the poorly conserved middle third of the protein. The position of an intron (i) (52 bp in D.melanogaster, 69 bp in D.virilis) is conserved. (B) Conserved domains in the two proteins are shown above a box representing the MSL2 protein. Highly variable gap regions described in the text are indicated by hatching. Amino acid identity/similarity within different regions is indicated by percentages below the protein. These sequence data have been submited to the DDBJ/EMBL/GenBank databases under accession No. AFO79368. Download figure Download PowerPoint A second cysteine-rich region (amino acids 521–562), which is loosely related to the PHD motif and to metallothioneins (Bashaw and Baker, 1995; Zhou et al., 1995), is conserved at all cysteines and histidines, including three positions not present in the published metallothionein alignment. The conservation of these residues is compatible with our previous report that mutation of C540 and C542 to alanine diminishes, but does not abolish msl2+ function in vivo (Lyman et al., 1997). The acidic nature of an adjacent region (amino acids 563–592) is also conserved, while a more distal proline-rich region (amino acids 681–701) is incompletely conserved (Zhou et al., 1995). A remarkable feature of the alignment is the presence of three gap regions having little or no homology. A short gap (amino acids 116–128) separates the RING finger and N-terminus from the rest of the protein. The second gap (amino acids 281–520) corresponds to the middle third of the MSL2 protein. This region contains most of the polymorphisms and length variations present within published D.melanogaster msl2 sequences; in addition, neither the previously described repeats (Bashaw and Baker, 1995) nor the acidic character of this region of the D.melanogaster protein are conserved in the D.virilis homolog. Finally, a gap following the second cysteine-rich domain (amino acids 593–614) is conserved in length, but is not conserved in sequence. These data suggest that the functions of the RING finger and other features of MSL2 have been conserved and that these domains may be positioned appropriately in the protein by more or less randomly evolving spa