Extent of Chromatin Spreading Determined by roX RNA Recruitment of MSL Proteins

See allHide authors and affiliations

Science  22 Nov 2002:
Vol. 298, Issue 5598, pp. 1620-1623
DOI: 10.1126/science.1076686


The untranslated roX1 and roX2 RNAs are components of the Drosophila male-specific lethal (MSL) complex, which modifies histones to up-regulate transcription of the male X chromosome. roX genes are normally located on the X chromosome, and roX transgenes can misdirect the dosage compensation machinery to spread locally on other chromosomes. Here we define MSL protein abundance as a determinant of whether the MSL complex will spread in cis from an autosomalroX transgene. The number of expressed roX genes in a nucleus was inversely correlated with spreading fromroX transgenes. We suggest a model in which MSL proteins assemble into active complexes by binding nascent roXtranscripts. When MSL protein/roX RNA ratios are high, assembly will be efficient, and complexes may be completed while still tethered to the DNA template. We propose that this local production of MSL complexes determines the extent of spreading into flanking chromatin.

A key mechanism for regulating eukaryotic gene expression is alteration of DNA packaging into chromatin (1). Modified chromatin architecture can sometimes be propagated long distances in cis from an initiation point (2–6), but the mechanism of such spreading is not understood. The MSL dosage compensation complex is thought to spread along the single male X chromosome inDrosophila (7). The MSL complex is composed of at least six proteins and two noncoding roX RNAs that paint the male X chromosome, leading to covalent modification of the NH2-terminal tails of histones H3 and H4 and twofold hypertranscription of hundreds of linked genes (8–10).

The two roX RNAs perform redundant functions (11,12). The lethality of roX1 roX2 double-mutant males can be rescued by expression of either roX1 orroX2 RNA from autosomal locations, showing thatroX RNAs can be supplied in trans to coat the X chromosome (12). However, both genes synthesizing roX RNAs are normally located on the X chromosome, and we have suggested that this contributes to targeting dosage compensation to the correct chromosome (7).

In certain msl mutant backgrounds, the MSL complex is absent from most locations on the X chromosome, but a small subset of sites, termed chromatin entry sites, retain partial complexes (7,13). Two of these sites are the roX genes. When aroX gene is moved to an autosome, it recruits MSL complex, which occasionally spreads up to 1 megabase (Mb) into the flanking autosome in a pattern that varies considerably (Fig. 1A). This suggested that the MSL complex recognizes the X chromosome by first binding at roX genes (and perhaps additional sites) and then spreading in cis (7). The MSL proteins could recognize theroX genes by binding DNA, nascent RNA, or both. MSL proteins bind roX RNAs to form active complexes, and eachroX gene also contains an MSL binding site (9,14).

Figure 1

Extensive spreading of MSL complex from autosomal roX transgenes. (A) When aroX gene is moved from the X chromosome to an autosome, it recruits the MSL complex to spread in cis from the transgene (7, 13). (B to E) Typical MSL staining (red is antibody to MSL1) in GMroX2-86F males with a wild-type X chromosome (B), roX1 X chromosome (C),roX2 X chromosome (D), or roX1 roX2 double-mutant X chromosome (E). Arrow indicates transgene location. Most nuclei in (B) to (D) have a single MSL band at the autosomal transgene (arrow), but modest MSL spreading can be seen in a low percentage of such nuclei: (B) 4%, (C) 2%, (D) 15%. (F to I) The extent of cis spreading is shown in transgenic roX1 roX2males carrying GMroX1-84C (F), GMroX1-65B (G), GMroX1-69C (H), and GMroX1-80C (I). The MSL bands in (I) are seen on either side of the chromocenter (c). The full genotype of the roX1 roX2chromosome is y w roX1ex6 Df(1)roX252P{w +4Δ4.3}. The P{w +4Δ4.3} element, needed to supply essential genes lost in Df(1)roX252 (12), is on the X chromosome at 18F (15).

The ectopic MSL spreading observed from autosomal roXtransgenes was seen in only a small fraction of nuclei compared with the invariant MSL pattern in the wild-type male X chromosome (7, 13). During complementation analyses ofroX1 roX2 mutants, we unexpectedly found that the genotype of the X chromosome strongly influenced ectopic MSL spreading from autosomal transgenes. We observed essentially no spreading in the presence of a wild-type X chromosome, but mutations in eitherroX1 or roX2 separately allowed modest MSL spreading from autosomal roX transgenes in some nuclei (Table 1; Fig. 1, B to D). In contrast, roX1 roX2 mutants displayed extensive autosomal MSL spreading [>1 megabase pair (Mbp)] in nearly all nuclei regardless of their insertion site (Fig. 1, E to I; Fig. 2, A and B), including centric heterochromatin (Fig. 1I). In each case, MSL complexes still painted the X chromosome. Autosomal roX transgenes were poor sites of MSL spreading if one or both endogenous roX genes were functioning on the X chromosome, but the same transgenes supported efficient MSL spreading over autosomes in a roX1 roX2 double mutant. Thus, roX genes appear to compete for limiting components for chromatin spreading.

Figure 2

Pairs of roX transgenes compete for limiting MSL proteins. (A to C) Extent of MSL spreading (red) in roX1 roX2 males with only GMroX1-67B (arrow) (A), only GMroX2-97F (arrowhead) (B), and both transgenes present in the same animal (C). (D) Northern blot showing that RNA made from GMroX2 is abundant when alone but decreases when roX1 RNA is also made. The same membrane was hybridized to roX1, roX2, and rp49 probe as a loading control. The number of endogenous roX genes (X) and autosomal transgenes (A) present is indicated above each lane. Lanes: 1, wild-type male; 2, wild-type female; 3 and 4,roX1 + roX2 males (lane 3, no transgene; lane 4 carries GMroX2-85B); 5 to 9, roX1 roX2 males (lane 5 carries GMroX2-85B, lane 6 carries both GMroX2-85B and GMroX1-67B, lane 7 has GMroX2-97F, lane 8 has both GMroX2-97F and H83roX1-61B, lane 9 has both GMroX2-97F and the nontranscribed Δpro roX1-66C). (E)y w roX1 ex6/Y; GMroX1-67B/+ males show MSL spreading in only 2% of nuclei. (F) Similar males that also overexpress MSL1 and MSL2 show modest MSL spreading in 84% of nuclei. (G) w Df(1)roX252 /Y; P{w + 4Δ4.3}/+; GMroX2-97F/+ males have MSL spreading in only 3% of nuclei. (H) Similar males that overexpress MSL1 and MSL2 show MSL spreading in 45% of nuclei.

Table 1

roX genes compete for MSL spreading. (Upper) Each transgene was placed in four different genetic backgrounds: wild-type X chromosome, roX1 single mutant,roX2 single mutant, and roX1 roX2 double mutant. The percentage of male nuclei showing MSL spreading from theroX transgene is given, with the number of nuclei counted in parentheses. The ability of each transgene to restore male viability toroX1 roX2 males is in the last column (male:female ratio). The MSL spreading observed in roX1 roX2 usually extended over 1 Mb, while only small clusters of MSL bands around the transgene (<300 kb) spread in other backgrounds. (Lower) All assays were performed in roX1 roX2 males. Transgenes were tested separately (transgene A only) or in pairs (both A and B) for their ability to support MSL spreading. N, number of nuclei examined. Single band, percentage of nuclei in which no MSL spreading occurred at the transgene(s). When autosomal MSL spreading was seen in nuclei carrying two roX transgenes, it was much less extensive than when only one transgene was present.

View this table:

We next asked if only X-linked roX genes could compete with autosomal MSL spreading. We found that a second autosomalroX transgene strongly reduced spreading from a referenceroX transgene. For example, the MSL complex spread several megabase pairs from P{w +GMroX2}97F (henceforth transgenics will be referred to as GMroX1-location or GMroX2-location, i.e., GMroX2-97F) in nearly all nuclei when it was the only source of roX RNA (Table 1; Fig. 2B). However, spreading was greatly reduced when GMroX1-67B was also present (Fig. 2C; Table 1). We tested seven pairs of roX transgenes and found that spreading from one site was reduced in both frequency and extent by the presence of a second roX gene (Table 1) (15). This confirms that the factors on the wild-type X chromosome responsible for competing for MSL spreading from an autosomal transgene are the endogenous roX genes and shows that roX genes are potent inhibitors of ectopic MSL spreading regardless of location.

The ability to compete with ectopic MSL spreading might reside in theroX RNAs or in the MSL binding sites within theroX genes. We constructed stocks in which MSL cis spreading from a reference GMroX2-97F transgene was challenged with two differentroX1 cDNA transgenes, both of which contain an MSL binding site. In one case, the roX1 cDNA was transcribed from the constitutive Hsp83 promoter (13). This transgene strongly competed with MSL spreading from GMroX2-97F (Table 2). The other roX1 cDNA transgene lacked a promoter and produced no detectable RNA (Fig. 2D, lane 9) (14). The nontranscribed cDNA failed to compete for MSL spreading from GMroX2-97F (Table 2). Thus, the active factor competing for ectopic autosomal MSL spreading is theroX RNA product. The MSL binding site within roX1clearly is not sufficient and may play no role.

Table 2

roX RNA, not the MSL binding site, competes for MSL spreading. The number of nuclei showing extensive spreading from GMroX2-97F in transgenic roX1 roX2 males was counted. Small clusters are 2 to 5 closely grouped MSL bands around 97F. Extensive spreading is >5 MSL bands extending more than 1 Mbp from 97F.N, number of nuclei counted. Line 1 shows GMroX2-97F alone. Lines 2 and 3 show nontranscribed roX1 cDNAs at 66C or 85D. Lines 4 and 5 show the roX1 cDNA transcribed from theHsp83 promoter at 61B and 87B. Although MSL complexes bound to roX1 cDNAs, they showed no detectable spreading in this assay. RNA production in the absence of spreading appeared to be sufficient for competition.

View this table:

One possible explanation for our results is that roXtranscripts compete to assemble MSL proteins into functional complexes. Because roX RNAs are unstable unless bound by MSL proteins (13, 15), steady-state levels of oneroX RNA species might fall if another roX species captured most MSL proteins. We measured the steady-state level ofroX2 RNA made by GMroX2 when it was the only source ofroX RNA or when roX1 RNA was also made from the endogenous gene on the X chromosome or an autosomal transgene. When the GMroX2-97F or the GMroX2-85B transgene was the only source ofroX RNA and complexes spread efficiently in cis, we observed high levels of roX2 RNA (Fig. 2D, lanes 5 and 7). WhenroX1 RNA was also present, MSL spreading was greatly inhibited around the GMroX2 transgenes and the amount ofroX2 RNA fell. It made no difference if the roX1RNA came from the endogenous locus on the X chromosome (Fig. 2D, lane 4), the GMroX1 autosomal transgene (Fig. 2D, lane 6), or the H83roX1 transgene (Fig. 2D, lane 8).

Transcription at GMroX2 might be indirectly repressed by the presence of roX1 RNA through feedback regulation. Alternatively, similar amounts of roX2 RNA may be made in all cases, but, when roX1 RNA competes for MSL proteins,roX2 transcripts may be inefficiently packaged into complexes and the naked RNAs then degraded. We reexamined ectopic MSL spreading from eight roX transgenes under conditions in which both MSL1 and MSL2 were overexpressed from constitutive promoters (16). GMroX1-67B and GMroX2-97F are typical examples that show dramatic spreading in the absence of any other roXgenes (Fig. 2, A and B) but rare spreading when the X chromosome carried a functional roX1 or roX2 gene (Fig. 2, E and G). This competition between roX genes was partially overcome when MSL1 and MSL2 proteins were overexpressed, as spreading from either transgene was clearly increased in both frequency and extent, despite the presence of a functioning roX gene on the X chromosome (Fig. 2, F and H) (15). These results strongly suggest that MSL proteins are normally recruited byroX RNAs to begin the spreading process and that the local concentration of MSL complexes at a roX gene determines the extent of epigenetic spreading into flanking chromatin.

Although the MSL complex has been reported to up-regulate transcription of flanking genes by histone modification (17,18), this model for dosage compensation inDrosophila has been disputed (19). The ability to cause consistent MSL spreading on autosomes provides an opportunity to examine the direct effect of MSL complexes on the transcription status of individual flanking genes. We used a GMroX1 transgene inserted at position 69C, where the linked reporter gene mini-white is expressed only in the dorsal part of the eye (Fig. 3, A and B) (20). GMroX1-69C flies showed a striking sex difference in eye pigmentation superimposed on this dorsoventral pattern. Males displayed sporadic red sectors of mini-white expression ventrally, suggesting that theroX gene recruited the MSL complex to partially overcome the local repressive chromatin environment. When the X chromosome carried aroX + gene, derepression of mini-whitewas a rare event, resulting in a few pigmented sectors, but in aroX1 roX2 mutant background, mini-whiteexpression in the ventral half of the eye was almost totally derepressed in males (Fig. 3, D versus B and C). As a control, we tested another nearby mini-white transgene, P{lacW}mirr B1–12, lacking anyroX sequences (21), and it was not affected byroX mutations (Fig. 3, E and F). These results suggest that spreading is likely to be a general phenomenon, occurring in many or most tissues of the animal. When the MSL complex spreads over genes surrounding roX1, the chromatin is remodeled for increased transcription. If ectopic MSL spreading occurs in a repressive chromatin environment, even strong silencing may be overcome. The consistency of ectopic spreading correlates with increased transcription of flanking chromatin.

Figure 3

MSL spreading visualized by eye color phenotype. w +GMroX1-69C is inserted in the second intron of the ara gene within the iroquoiscluster (15). The mini-white eye color marker shows a typical darker pigment in males than in females in the dorsal eye but is silenced by the local chromatin environment in the ventral half of the eye (20). (A) y w;w +GMroX1-69C/+ females silence mini-white in the ventral ∼70% of the eye. (B)y w/Y; w +GMroX1-69C/+ males have red sectors in the ventral part of the eye. (C) y w roX1ex6 Df(1)roX2 52P{w +4Δ4.3};w +GMroX1-69C/+ females have mini-white expression unchanged fromw +GMroX1-69C. Light orange color comes from the P{w +4Δ4.3} transgene that provides essential genes lost in Df(1)roX2 52(12, 15). (D) y w roX1ex6 Df(1)roX2 52P{w +4Δ4.3}/Y;w +GMroX1-69C/+ males show almost complete derepression of mini-white in the ventral half of the eye. (E) w/Y; P{lacW}mirr B1–12/TM6, Tb male carries a mini-white marked P element about 100 kb fromw +GMroX1-69C and shows similar ventral silencing (21). (F) y w roX1ex6Df(1)roX2 52P{w +4Δ4.3}/Y; P{lacW}mirr B1–12/+ male shows no derepression of ventral silencing because the B1–12 transgene does not carry a roX gene to attract the MSL complex. Light orange ventral eye is due to the P{w +4Δ4.3} transgene. This is a rare escaper male lacking any source of roX RNA. About 99% of such males die because of a failure of dosage compensation.

We have found conditions in which roX RNAs either assemble into MSL complexes that preferentially spread in cis from the site of transcription or diffuse to the X chromosome. We propose that these outcomes are determined by MSL proteins assembling onto growingroX transcripts tethered to the chromosome. The local pool of MSL proteins would control the efficiency of this process. If active complexes are completed by the time 3′ RNA processing releases the RNA from the chromosome, the most likely outcome is immediate entry and spreading into flanking chromatin. If the supply of MSL subunits is reduced by a competing source of roX RNA, release ofroX transcripts might precede complex maturation. These MSL complexes are unlikely to return to the transgene after assembly in solution and instead diffuse to the X chromosome, which, in addition toroX genes, has unknown features that make it the best target for MSL complexes.

  • * To whom correspondence should be addressed. E-mail: mkuroda{at} (M. I. K.); vmeller{at} (V.H.M.)


Stay Connected to Science

Navigate This Article