BIOLOGICAL OVERVIEW

The male-specific-lethal (MSL) proteins in Drosophila melanogaster serve to adjust gene expression levels in male flies containing a single X chromosome to equal those in females with a double dose of X-linked genes. Together with noncoding roX RNA, MSL proteins form the 'dosage compensation complex' (DCC), which interacts selectively with the X chromosome to restrict the transcription-activating histone H4 acetyltransferase MOF (Males-absent-on-the-first) to that chromosome. MSL3 is essential for the activation of MOF's nucleosomal histone acetyltransferase activity within an MSL1-MOF complex. By characterizing the MSL3 domain structure and its associated functions, it has been found that the nucleic acid binding determinants reside in the N terminus of MSL3, well separable from the C-terminal MRG signatures that form an integrated domain required for MSL1 interaction. Interaction with MSL1 mediates the activation of MOF in vitro and the targeting of MSL3 to the X-chromosomal territory in vivo. An N-terminal truncation that lacks the chromo-related domain and all nucleic acid binding activity is able to trigger de novo assembly of the DCC and establish an acetylated X-chromosome territory (Morales, 2005).

Male-specific lethality in fruit flies results from loss of function of a few genes involved in dosage compensation. Dosage compensation is a process that serves to adjust the expression levels of genes residing on the single male X chromosome to meet the expression originating from the two female X chromosomes. The five male-specific lethal proteins known to be critically involved in dosage compensation, MSL1, MSL2, and MSL3 (male-specific-lethal 1, 2, and 3, respectively), MOF (males-absent-on-the-first), and MLE (maleless), associate with the noncoding roX RNA to form the dosage compensation complex (DCC; also referred to as the MSL complex or compensasome). Due to the male-specific expression of MSL2, the complex only assembles in males, where it selectively associates with many sites on the X chromosome. This association effectively concentrates MOF, a histone acetyltransferase (HAT) with specificity for lysine 16 of histone H4 (H4K16), on the X chromosome. According to the prevailing model, the DCC is involved in a twofold increase of transcription of X-linked genes in males. Acetylation of H4K16 (H4K16ac) can reverse chromatin-mediated repression of transcription, and targeting MOF to a heterologous promoter in yeast leads to activation of transcriptio. The H4 N terminus is involved in folding of the nucleosomal fiber, and acetylation may thus modulate chromatin organization. In mammals, H4K16 acetylation also correlates with less-repressive chromatin. Sequestration of MOF to the X chromosome may also affect the level of autosomal gene expression globally. Whatever the mechanisms by which gene expression levels are adjusted in a twofold range, it is likely that essential feedback regulations are involved in fine-tuning (Morales, 2005 and references therein).

Knowledge of the molecular functions of the MSL proteins is rather limited. Each of them is required for faithful association of the DCC with the X chromosome. MSL2 is exclusively expressed in males and required for accumulation of roX RNA, an essential component of DCC. Interaction of MSL2 with MSL1 is a mutual requirement for interaction of the two proteins with the X chromosome and for formation of the complete DCC. MSL1 appears to be a scaffolding protein since it is able to associate not only with MSL2 but also with MOF and MSL3. MLE, the homolog of human RNA helicase A, appears responsible for stabilizing roX RNA and for its incorporation into DCC. A core complex will form in the absence of roX RNA or of any of the three RNA binding subunits MLE, MOF, or MSL3, but its association with only a subset of X-chromosomal sites does not lead to faithful dosage compensation. Histone H4 acetylation by MOF is likely to play a crucial role in the modification of chromatin structure that regulates transcription. In addition, its acetylase activity is important for the distribution of the DCC over the X chromosome. The fact that MOF can acetylate MSL1 and MSL3 suggests possible involvement of nonhistone targets as well (Morales, 2005).

MSL3 (Gorman, 1995) contains an MRG domain, a domain found throughout eukaryotes from yeasts to humans, with as yet unknown function (Bertram, 2001; Marin, 2000). Interestingly, several members of this 'MRG family' are associated with HATs in complexes known or suspected to be involved in transcriptional regulation. Prominent examples are the MSL3-relative Eaf3p, which interacts with the acetyltransferase Esa1 in the yeast NuA4 complex (Eisen, 2001), and human MRG15, which can be found associated with the HAT Tip60 (Doyon, 2004) in a human complex highly related to the yeast NuA4 (Morales, 2005 and references therein).

MSL3 interacts with roX RNA, and its association with the X chromosome is sensitive to RNase treatment (Buscaino, 2003). It can also bind DNA and chromatin and forms a salt-resistant complex with MSL1 in vitro (Morales, 2004). The reconstitution of a four-subunit MSL complex recently led to insight into functional interactions of MSL3 (Morales, 2004). Remarkably, the association of MSL3 with the C terminus of MSL1 in complex with MOF resulted in a dramatic activation of acetyltransferase activity on nucleosome substrates. At the same time, binding of MSL3 leads to a refinement of substrate specificity: in the absence of MSL3, the rudimentary acetylation activity of MOF was entirely directed towards MSL1 and does not modify nucleosomes. However, in contact with MSL3, MSL1 is no longer subject to modification, but robust acetylation of nucleosomal H4 is observed. This finding has led to a hypothesis for the existence of a molecular 'checkpoint' that serves to restrict the H4K16ac mark to the X chromosome. According to this model, association of MOF with MSL1 (Morales, 2004) leads to its sequestration on the X chromosome, but only completion of the DCC by incorporation of MSL3 unleashes its H4 acetylase activity. Whether the MOF that is expressed in female cells in the absence of a functional DCC has a lower activity toward histone H4 or profits from an equally forthcoming molecular environment remains to be explored (Morales, 2005).

The direct interaction of MSL3 with MOF in vitro is comparably weak and does not result in an activation of its HAT activity. Rather, the stimulatory effect of MSL3 on MOF requires simultaneous contact of MSL3 and MOF with the C terminus of MSL1 (Morales, 2004). Since MSL3 can bind DNA and chromatin in vitro (Morales, 2004), its presence in a complex with MOF may stabilize the interaction of the enzyme with the chromatin substrate. It is also possible that contact of MSL3 with MSL1 indirectly affects acetyltransferase activity through conformational changes. In order to address these possibilities, the functional interactions of MSL3's domains have been examined. Using a set of mutants that are no longer able to bind nucleic acids or MSL1, it was found that the stimulation of MOF’s HAT activity, as well as targeting to the X-chromosomal territory, depends on interactions with MSL1 via the MRG signature of MSL3. In contrast, the nucleic acid binding N terminus including the chromo-related domain (CRD) is not a primary determinant of targeting MSL3 to the X-chromosome territory, nor does it contribute to stimulation of MOF's HAT activity in vitro or to establish the accumulation of the H4K16 acetylation mark on the X chromosome in vivo (Morales, 2005).

The MSL1 interaction surface maps to the C-terminal half of MSL3. This part of MSL3 is characterized by similarities to the MRG domain that subsumes MRG15, MSL3, and related proteins in multiple species into the so-called MRG family (Bertram, 2001, Marin, 2000). The msl3 gene is related to the Drosophila mrg15 gene, suggesting an early gene duplication event. Accordingly, MRG sequences in MSL3 are highly conserved between D. melanogaster and Drosophila virilis (Marin, 2003; Marin, 2000). The MRG domain defined by Marin and Baker (Marin, 2000) consists of three blocks of strong sequence similarity separated by short amino acid stretches of lesser conservation. Interestingly, these 'linker' regions harbor rather long insertions in MSL3 of flies and humans. The C terminus of MSL3 may thus be organized by folding of MRG signature sequences, which are disconnected in the primary sequence, into a compact unit from which the MSL3-specific structures 'loop out.' Consistent with this idea, it was found that every deletion in the C terminus of MSL3 compromises interaction with MSL1. Most of these deletions affect at least one of the blocks of MRG sequence similarity, most likely leading to global misfolding. However, one deletion that abolishes MSL1 binding (Delta328-433) selectively removed MSL3-specific sequences between two MRG blocks. There is considerable conservation of these sequences in the Drosophila species for which sequence information has recently become available, suggesting a conserved function, but whether this sequence contains a dedicated MSL1 interface remains to be explored. In any case, this analysis suggests that the MRG sequence similarity reflects a functional domain. The MRG-MSL1 contact is essential for targeting MSL3 to the X-chromosomal territory, confirming the functional importance of the interactions defined in vitro. It is suggested that MRG modules in other MRG family members may also constitute protein-protein interaction units (Morales, 2005).

In vitro analysis showed that MSL3 interacts better with single-stranded nucleic acids than with dsDNA. The significance of ssDNA interaction, if any, is unclear at the moment. In contrast, there is evidence that MSL3 interacts with roX RNA in vivo and in vitro (Akhtar, 2000; Buscaino, 2003), but the domain involved in RNA binding had not been defined. Biochemical analysis demonstrates that the nucleic acid binding structures reside in the N-terminal half of MSL3, which also contains the CRD. Previously, Buscaino (2003) suggested that RNA interaction of MSL3 is affected by its acetylation at lysine 116, close to the CRD. In the current studies, a fragment comprising the first 140 amino acids (and hence the CRD as well as K116) was not sufficient for nucleic acid binding, but sequences up to amino acid 259 contributed significantly. To what extent the CRD of MSL3 contributes to RNA binding needs to be established. The CRDs of MSL3 and MOF appear more related to each other than to canonical chromodomains. They lack the alpha-helix supporting the ß-sheet bundle and aromatic residues that may be involved in recognition of methylated histone N termini. The CRD of MOF also appears not to be sufficient for RNA binding. A further interesting similarity between MOF and MSL3 is that nucleic acid interactions are not the primary targeting determinant for either MOF (Morales, 2004) or MSL3. Although impairment of the CRDs leads to somewhat increased binding of the corresponding GFP fusion protein to autosomes, their concentration on the X-chromosomal territory is still obvious. However, the CRDs and noncoding RNA may have functions that are not assayed for in simple recruitment experiments. It is also possible that the CRDs of MOF and MSL3 provide partially redundant functions for DCC assembly. In contrast, mutations in MOF or MSL3 that abrogate their interaction with the C terminus of MSL1 prevent faithful recruitment to the X chromosome. Obviously, the recruitment assay employed may just reveal the strongest binary interaction that MSL3 or MOF are involved in. However, the fact that overexpression of an MSL3 lacking all nucleic acid binding capacity was able to complement an MSL3 deficiency and to trigger the accumulation of MOF and H4K16 acetylation on the X-chromosomal territory emphasizes the importance of the MSL protein interactions for the assembly of a functional DCC (Morales, 2005).

MSL complexes can be formed in vitro in the absence of RNA. A deficiency of roX RNA in vivo can be partially overcome by overexpression of the 'platform' proteins MSL1 and MSL2. It is possible that transient overexpression of MSL3 overcomes the RNA requirement and that under normal conditions of limiting MSL protein concentrations RNA is required for faithful DCC assembly (Morales, 2005).

The remarkable stimulation of MOF's HAT activity upon association of MSL3 with an MSL1-MOF complex was not due to enhanced binding of MSL3 to nucleic acids but rather required contact of MSL3 with the MSL1 scaffold. MOF and MSL3 are brought into proximity by interaction with adjacent structures in the C terminus of MSL1 (Morales, 2004). It is possible that the MSL1 scaffold stabilizes an otherwise transient and therefore nonproductive direct contact between MSL3 and MOF (Morales, 2004). The existence of such a contact has been inferred from the fact that MSL3 can be acetylated by MOF (Buscaino, 2003). However, when it comes to acetylation, MSL1 is a much better substrate for MOF than MSL3 (Morales, 2004). The new data reinforce a previous model of an acetylation 'checkpoint' built into DCC assembly. Accordingly, the regulatory potential of H4K16 acetylation would only be fully realized upon binding of MOF with MSL1 and the completion of the complex by association of MSL3 (Morales, 2004). Such a checkpoint would render full activation of MOF dependent on proper DCC assembly and hence 'maleness' and serve to restrict the critical epigenetic mark to the X chromosome (Morales, 2005).

GENE STRUCTURE

cDNA clone length - 1963

Bases in 5' UTR - 101

Exons - 6

Bases in 3' UTR - 323

PROTEIN STRUCTURE

Amino Acids - 512 (Isoform A)

Structural Domains

Dosage compensation in Drosophila is mediated by genes known as 'male-specific lethals' (msls). Several msls, including male-specific lethal-3 (msl-3), encode proteins of unknown function. The Drosophila virilis msl-3 gene has been cloned. Using the information provided by the sequences of the Drosophila melanogaster and D. virilis genes, it was found that sequences of other species can be aligned along their entire lengths with msl-3. Among them, there are genes in yeasts (the Schizosaccharomyces pombe Alp13 gene, as well as a putative Alp13 homolog, found in Saccharomyces cerevisae) and in mammals (MRG15 and MSL3L1 and their relatives) plus uncharacterized sequences of the nematode Caenorhabditis elegans and the plants Arabidopsis thaliana, Lycopersicon esculentum, and Zea mays. A second Drosophila gene of this family has also been found. It is thus likely that msl-3-like genes are present in all eukaryotes. Phylogenetic analyses suggest that msl-3 is orthologous to the mammalian MSL3L1 genes, while the second Drosophila melanogaster gene (which has been called Dm MRG15) is orthologous to mammalian MRG15. These analyses also suggest that the msl-3/MRG15 duplication occurred after the fungus/animal split, while an independent duplication occurred in plants. The proteins encoded by these genes have similar structures, including a putative chromodomain close to their N-terminal end and a putative leucine zipper at their C-terminus. The possible functional roles of these proteins are discussed (Marin, 2000).

The msl-3 gene was part of a small gene family that was widespread in eukaryotes. Both in mammals and in Drosophila there are two related genes, msl-3 and MRG15. Considering their high similarities, the origin of the sequences MORF 4 in one case and MSL3L1-b in the other must have been relatively recent retrotranscriptions and insertions of processed mRNAs derived from MRG 15 and MSL3L1, respectively. In spite of the fact that ESTs corresponding to MSL3L1-b, were found it is unclear at present whether it corresponds to a functional gene (such as MORF4) or to a processed pseudogene with some residual transcriptional activity. MRGX contains introns, and thus it can be interpreted as a conventional duplication of MRG15, which arose quite recently and with a truncated N-terminal end (Marin, 2000).

There are two hypotheses to explain why both plants and animals have duplicated genes, while yeasts do not. First, it is possible that there was a single duplication preceding the plants/animals/fungi split and that yeast have subsequently lost one of the two duplicates. Alternatively, the duplications in plants and animals may be independent, with the plant duplication occurring after the plants-versus-fungi/metazoans split and the duplication found in animals occurring after the fungi/metazoans split. This second hypothesis is favored, considering the high similarity of the plant paralogs. If they were produced by a duplication before the split of plants with fungi and animals, their degree of differentiation should be very high, branching in the inner part of the tree. Thus, it is concluded that the simplest explanation for these data is that the plant paralogs have been duplicated relatively recently, while msl-3 and MRG15 are older duplicates, probably produced soon after the fungi/animals split. It is also suggested that Sc Alp13 is the Alp13 ortholog in Saccharomyces. This is likely, considering that the S. cerevisiae genome has been completely sequenced and no other gene has shown any similarity to the S. pombe Alp13 gene (Marin, 2000).

Data are available to suggest that msl-3 evolved at a higher rate than MRG15 after the msl-3/MRG15 duplication. This hypothesis is supported by two independent lines of evidence. (1) The tree, which is based only on the five conserved motifs, shows that the distance between the Drosophila and mammalian MRG 15 genes is shorter than the distance between the mammalian and Drosophila msl-3 genes. (2) There are qualitative features, in regions not included in the analysis that provided the phylogenetic tree, that are characteristic of the msl-3 gene in Drosophila and mammals: the very long stretch of amino acids between the conserved regions 4 and 5, as well as the characteristic five additional amino acids in the middle of conserved motif 1. These changes might be related to the acquisition of new functional roles in the msl-3 genes. Recent data suggest that, contrary to the theoretical expectations that most duplicated genes should accumulate mutations and be lost, acquisition of new functions may be frequent after gene duplication (Marin, 2000).

The tree, as well as the fact that the C. elegans sequence Ce-1 lacks the long region between motifs 4 and 5 characteristic of msl-3 in Drosophila and mammals, suggests that Ce-1 corresponds to an ortholog of MRG15. Considering the data supporting the hypothesis that nematodes are close relatives of insects, it is hypothesized that the msl-3 ortholog has been lost in the nematode lineage. Results for plant genes are more ambiguous. Although the simplest hypothesis is that a duplication occurred relatively recently in plants, the topology of the tree in fact does not correspond to that expected for two paralogs (the At-1, Zm-1, and Le-1 sequences appear to be derived from the At-2 sequence). However, the fact that two of the sequences that were analyzed in plants are incomplete (Le-1 is missing the last domain, while Le-2 lacks part of it) may significantly alter the relationships among the plant sequences. The relationships among these plant genes should be reconsidered when more sequences are available (Marin, 2000).

It is interesting to speculate about the possible function of the msl-3-related genes in other species, as well as that of MRG 15 in Drosophila. The function of MRG 15 in mammals is unknown, although it has been suggested that the related chromodomain-lacking gene MORF 4 may be involved in cellular senescence. One possibility to consider is whether MRG 15 may also be acting in dosage compensation in fly species. However, finding msl-3-related genes in many species, including yeasts, that do not have a dosage compensation system, raises the possibility that these genes may perform some ancestral gene function from which the msl-3 dosage compensation function evolved. Although the precise function of yeast gene Alp13 is also currently unknown, it is known that ALP13 is a nuclear protein and that Alp13 mutants are viable but sterile and lead to alterations of cell shape. At another level, the description of msl-3-related genes in mammals opens up the possibility of studying their possible interactions with the previously described mammalian maleless relative (Helicase A) to determine whether a similar MSL complex is present in mammalian species (Marin, 2000).

The comparison of all the msl-3-related sequences has allowed reevaluation of the existence of two chromodomains in MSL-3. It is concluded that only the more N-terminal chromodomain-containing regions have a pattern of conserved residues in the members of this gene family that is compatible with being a chromodomain-related structure. However, the reason why this pattern was not originally detected in D. melanogaster MSL-3 is because that region has a very low similarity with the canonical chromodomain-containing proteins, such as Polycomb or HP1. Most data suggest that the chromodomain is a protein-protein interaction domain, but it is an open question whether the chromodomain and other related sequences, like the chromo-shadow domain or the chromodomain-like sequences in MSL-3, actually perform the same role. In contrast, the most C-terminal putative chromodomain-containing region actually does not show the expected pattern of conserved residues in the different MSL-3-related proteins, suggesting that this region does not correspond to a chromodomain. The regular disposition of some conserved hydrophobic amino acids suggests that it may instead correspond to a leucine zipper–like structure. Thus, MSL-3 and its relatives might be able to form dimers or interact using such a surface with MSL-1 or MSL-2. All of these structural features could be related to MSL-3 being a structural component of the compensasome, important for binding together proteins that provide the different enzymatic activities (e.g., helicase, acetyltransferase) required for the action of the complex. MSL-3 relatives could be performing similar mediator functions in other, related or unrelated, protein complexes (Marin, 2000).

After all these analyses, it is still unclear whether MSL-3 and its relatives are able to interact in some way with nucleic acids or other structural components of chromatin, such as histones. Two interesting features observed in these proteins which may be significant for their function in regulating chromatin structure are the highly polar region between motifs 2 and 3 and the putative KH domain in D. melanogaster msl-3. Many other chromatin-associated proteins, such as topoisomerase I, nucleoplasmin, and the chromodomain-containing protein SWI6, also have characteristic charged regions. It is possible that the differences in disposition of charges among MSL-3-related proteins detected in these analyses, for example, the longer polar domain in D. virilis MSL-3 than in D. melanogaster MSL-3, may be related to the fine tuning of the functions of those proteins in the different species. Similarly, the fact that D. melanogaster msl-3, but apparently not its D. virilis homolog or other MSL3-related proteins, has a region with some similarity to the KH domain, a type of RNA-binding domain, may be a demonstration of slightly different roles of the two homologous proteins in their respective complexes. Finally, the finding of alternative splicing for mammalian MSL3L1, with one of the isoforms produced lacking part of the chromodomain, is interesting. Chromodomain-lacking duplicates of MRG15 have been described. It may be significant, then, that in Drosophila melanogaster, there are two types of msl-3 RNAs, and one of them would lack part of the chromodomain-like region. These results suggest that there may be, in both mammals and Drosophila, significant roles for the truncated versions of the MSL-3-type proteins (Marin, 2000).

The sequence analysis of MSL3 reveals two prominent features. A CRD resides within the first 90 amino acids. This domain deviates significantly from the canonical chromodomains known to interact with methylated histone H3 N termini and rather resembles the CRD of MOF, which is involved in RNA binding. In the C-terminal half of MSL3, sequences are present that relate MSL3 to the MRG protein family (Bertram, 2001). However, whereas the MRG domain appears as a contiguous stretch of similarity in MRG15 and its closer relatives, it is fragmented by insertions in MSL3 (Bertram, 2001; Morales, 2005).

In Drosophila, dosage compensation of X-linked genes is achieved by transcriptional upregulation of the male X chromosome. Genetic and biochemical studies have demonstrated that male-specific lethal (MSL) proteins together with roX RNAs regulate this process. MSL-3 is essential for cell viability. Three domains in the protein have distinct roles in dosage compensation. The chromo-barrel domain (CBD) is not necessary for MSL targeting to the male X chromosome but is important for male viability and equalization of X-linked gene transcription. The polar region cooperates with the CBD in MSL-3 function, whereas the MRG domain is responsible for targeting the protein to the X chromosome. These results demonstrate that MSL-3 localization to the male X chromosome and transcriptional upregulation of X-linked genes are two separable functions of the MSL-3 protein (Buscaino, 2006).

date revised: 12 April 2022

Home page: The Interactive Fly © 2006 Thomas Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.