Functional Redundancy Within roX1, a Noncoding RNA Involved in Dosage Compensation in Drosophila melanogaster

Corresponding author: Mitzi I. Kuroda, Howard Hughes Medical Institute, Baylor College of Medicine, One Baylor Plaza, Houston, TX 77030., mkuroda{at}bcm.tmc.edu (E-mail)

Communicating editor: B. J. MEYER

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Drosophila melanogaster males dosage compensate by twofold upregulation of the expression of genes on their single X chromosome. This process requires at least five proteins and two noncoding RNAs, roX1 and roX2, which paint the male X chromosome. We used a deletion analysis to search for functional RNA domains within roX1, assaying RNA stability, targeting of the MSL proteins to the X, and rescue of male viability in a roX1^- roX2^- mutant background. We found that deletion of 10% segments of the RNA did not dramatically reduce function in most cases, suggesting extensive internal redundancy. The 3' 600 nt of roX1 were most sensitive to mutations, affecting proper localization and 3' processing of the RNA. Disruption of an inverted repeat predicted to form a stem-loop structure was found partially responsible for the defects observed.

NONCODING RNAs are postulated to regulate gene expression in many different ways (reviewed in EDDY 2001 ). Some, such as AIR, are implicated in the regulation of imprinted expression of nearby genes (SLEUTELS and BARLOW 2002 ; SLEUTELS et al. 2002 ). Noncoding RNAs also participate in dosage compensation, where they are important for the regulation of gene expression along the length of the X chromosome in mammals and Drosophila (reviewed in FRANKE and BAKER 2000 ; PANNUTI and LUCCHESI 2000 ; AVNER and HEARD 2001 ; LEE 2002 ; MELLER and KURODA 2002 ). In mammals, females inactivate one of their two X chromosomes, thus equalizing gene expression to males with a single X. This process requires expression of a 15- to 17-kb noncoding RNA, Xist, from the future inactive X and results in silencing of thousands of genes. Xist is the central player in correct targeting of X inactivation, because when the Xist gene is moved to an autosome, its RNA spreads into flanking autosomal genes causing them to be silenced (LEE and JAENISCH 1997 ; WUTZ and JAENISCH 2000 ).

Drosophila melanogaster achieve dosage compensation by the opposite mechanism, twofold upregulation of gene expression from the single male X chromosome (reviewed in FRANKE and BAKER 2000 ; PANNUTI and LUCCHESI 2000 ; MELLER and KURODA 2002 ). Yet, two noncoding RNAs, roX1 and roX2 (RNA on X), are required for this process as well (AMREIN and AXEL 1997 ; MELLER et al. 1997 ; FRANKE and BAKER 1999 ; MELLER and RATTNER 2002 ). These two RNAs are present in a complex with at least five proteins, collectively referred to as the male-specific lethals (MSLs; MELLER et al. 2000 ; SMITH et al. 2000 ). Targeting the X chromosome by the MSL complex is thought to occur, at least in part, through spreading in cis from roX genes (KELLEY et al. 1999 ; PARK et al. 2002 ). One of the MSLs, males-absent on the first (MOF), is a MYST histone acetyltransferase that specifically acetylates histone H4 at lysine 16 (H4Ac16; TURNER et al. 1992 ; HILFIKER et al. 1997 ; SMITH et al. 2000 ; AKHTAR and BECKER 2001 ). The histone kinase JIL-1 is also associated with the MSL complex and phosphorylates histone H3 at Ser10 (JIN et al. 1999 , JIN et al. 2000 ; WANG et al. 2001 ). By restricting the MSL complex to the male X, JIL-1- and MOF-mediated histone modifications are enriched on X chromatin, presumably to mediate the twofold upregulation of X-linked gene expression.

All of the MSL proteins are required for male viability. Loss-of-function mutations in the msl genes cause lethality in males because of a lack of dosage compensation, but have no effect on female viability (MELLER and KURODA 2002 ). In contrast, available roX1 mutations have no adverse effect on either sex (MELLER et al. 1997 ; KELLEY et al. 1999 ). A complete deletion of roX2 is also male viable (MELLER and RATTNER 2002 ). However, removing both roX1 and roX2 dramatically reduces male viability with no apparent effect in females (MELLER and RATTNER 2002 ). The functional redundancy of the roX RNAs is surprising, given that the roX1 RNA is 3.7 kb, compared to only 600 nucleotides (nt) for the major roX2 transcript (SMITH et al. 2000 ; Y. PARK, personal communication). The two share little primary sequence, with homology searches detecting only a 30-nt motif in common (AMREIN and AXEL 1997 ; FRANKE and BAKER 1999 ).

In a roX1^- roX2^- double mutant, MSL complex localization is undetectable in embryos (FRANKE and BAKER 1999 ). In mutant larvae, the complex binds weakly to a variable number of sites on the polytene male X chromosome, but can also be found at autosomal sites and the chromocenter (MELLER and RATTNER 2002 ). Therefore, one likely function of the roX RNAs is to facilitate X chromosome targeting or binding by the MSL complex.

We have used a genetic approach to search for functional domains within roX1. We created a series of roX1 deletions and tested them for their ability to support dosage compensation in vivo. The RNAs were also assayed for their ability to target the MSL proteins to the X and modify chromatin. We found that the 3' end of roX1 is important for full activity and X localization. Surprisingly, most roX1 transcripts carrying 10% deletions retained near normal activity. Larger deletions were almost nonfunctional, suggesting that multiple, redundant functional domains may be interspersed along the length of roX1 and that the structure of the RNA may be quite flexible. As roX1 lacks obvious repeated sequence motifs, functional domains may operate at the level of tertiary structure, not primary sequence. This is consistent with overlapping genetic function shared by roX1 and roX2 in the absence of primary sequence similarity.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Flies were maintained on standard cornmeal-molasses food at room temperature. Transgenic flies were obtained using standard protocols, mapped, and crossed to homozygosity in a y w roX1^ex6 background (KELLEY et al. 1999 ).

We adapted the roX1^- roX2^- double-mutant X chromosome by meiotic recombination to include a w^- allele to track the presence of w⁺ marked transgenes and a y^- allele to track differentially marked X chromosomes. We also recombined an X chromosome insertion of the cosmid transgene [w⁺ 44.3] (MELLER and RATTNER 2002 ) onto the roX1^- roX2^- X chromosome. The cosmid [w⁺ 44.3] is required to rescue the embryonic lethality of flanking genes deleted in the roX2 deficiency, Df(1)52. Df(1)52 flies carrying cosmid [w⁺ 44.3] still lack CG11695, a predicted Zn finger gene with no mutant phenotype, as well as nod, a kinesin motor required for fidelity of chromosome segregation during female meiosis (ZHANG et al. 1990 ). Supplemental Fig 1 (at http://www.genetics.org/supplemental/) shows the crosses used to create the y w roX1^ex6 Df(1) 52 [w⁺ 44.3] X chromosome. For all complementation tests, 30 y w roX1^ex6 Df(1) 52 [w⁺ 44.3] virgins were mated to 30 w; [w⁺ roX1]/Balancer males. Bottles were flipped two times and progeny were counted for 2 weeks following eclosion of the first offspring.

View larger version (18K): In this window In a new window Download PPT slide   Figure 1. Complementation test. Depending on the insertion site of the [w+ roX1] transgene, fathers carrying either the CyO or the TM3 Sb e balancer were crossed to females homozygous for the roX1- roX2- double-mutant X chromosome. Four different classes of progeny are expected with males and females for each. The ratio of males of class 1 to females of class 1 is the rescue frequency. The ratio of males of class 2 to females of class 2 is the escaper frequency. Classes 3 and 4 represent progeny arising from X chromosome nondisjunction during meiosis of the nod- females.

roX1 deletions were created by long-range PCR amplification of the roX1 cDNA c20, including its pBlueScript backbone, with primers pointing away from each other and carrying a unique NheI restriction site. PCR products were digested and recircularized. Deletions were subcloned into a pCaSpeR derivative carrying the hsp83 promoter and a 450-nt PstI fragment containing polyadenylation signals from the tra2 locus (MELLER et al. 2000 ).

Site-directed mutants in the stem-loop were created by a two-step PCR scheme using overlapping primers in opposite orientation, each carrying the desired nucleotide changes. Independent PCR reactions were run with each mutated primer and an appropriate nonmutated primer that annealed outside the 1.2-kb HindIII fragment containing the roX1 3' end. The two PCR fragments were purified, allowed to anneal where the mutated primers overlap, and used as template in a second PCR step, using only the outside primers. The resulting PCR product was cut with HindIII and used to replace the unmutated HindIII sequence in the c20 cDNA. Candidates were confirmed by sequencing. These mutant cDNAs were subcloned into the H83pCaSpeR vector and injected as above.

Transgenic RNA for Northern analysis and RNase protection was isolated from sexed third instar larvae homozygous for the transgene in a y w roX1^ex6 background using Trizol (Invitrogen, San Diego). In Fig 6, larvae homozygous for the c20 wild-type cDNA were in a y w roX1^MB710 background (MELLER et al. 1997 ), but identical results were obtained for the roX1^ex6 background (data not shown). Northern analysis was performed as previously described using 20 µg of total RNA (KELLEY 1993 ; MELLER et al. 2000 ).

View larger version (26K): In this window In a new window Download PPT slide   Figure 2. Rescue by roX1 deletion constructs. Overview of 5' and 3' roX1 constructs and deletions roX11–roX130. The arrow indicates the Hsp83 promoter and the orientation of the transgene. The top line shows the beginning and end of roX1 RNA. The 5' and 3' ends of each deletion are shown; numbering is for c20 roX1 cDNA (accession no. AB051842). Constructs 5' roX1 and 3' roX1 are based on the c3 cDNA and have been described previously (MELLER et al. 1997 , MELLER et al. 2000 ). Also shown are the number of independent insertions analyzed, the number of rescued males and females counted for the insertion with the highest rescue (class 1 in Fig 1), and a bar graph for the rescue frequency. (*) For the complementation test without a rescuing transgene, classes 1 and 2 (Fig 1) both represent escaper males and their sisters. These are combined and shown both in actual numbers and as ratios in the bar graph.

View larger version (23K): In this window In a new window Download PPT slide   Figure 3. Rescue by roX1 stem-loop mutants. Overview of the predicted stem-loop region in the 3' end of roX1. Shown are nucleotides 2700–3390 of the roX1 cDNA c20. The sequence of the stem-loop is indicated in uppercase letters. Mutations in the stem-loop are indicated to the side in lowercase letters (indicated as open boxes in the diagrams of constructs). Also shown are the endpoints of each deletion (see Fig 2), number of independent insertions analyzed, and the actual number of males and females counted to calculate the rescue frequency for the insertion with the highest rescue.

View larger version (24K): In this window In a new window Download PPT slide   Figure 4. RNA in situ hybridization to transgenic polytene chromosomes. Chromosome spreads from male transgenic larvae with a roX1- roX2- X chromosome. A and B are from a roX110 transgenic; C and D, roX111; and E and F, roX19. A, C, and E show the roX1 signal in Texas red. B, D, and F show the roX1 signal merged with the DNA counterstain 4',6-diamidino-2-phenylindole (DAPI; blue). In this and subsequent figures, X marks the X chromosome.

View larger version (18K): In this window In a new window Download PPT slide   Figure 5. MSL1 immunostain of transgenic polytene chromosomes. Chromosome spreads from male larvae with a roX1- roX2- X chromosome. A and B are from larvae without a roX1 transgene; C and D, with a wild-type roX1 transgene; E and F, roX110; and G and H, Overlap. A, C, E, and G show the MSL1 signal in Texas red. B, D, F, and H show the MSL1 signal merged with the DNA counterstain DAPI (blue).

View larger version (69K): In this window In a new window Download PPT slide   Figure 6. Northern analysis of transgenic RNAs. Total RNA from y w roX1ex6; [w+ roX1] transgenic larvae was probed successively with a roX1 and an rp49 probe.

Probes for RNase protection were generated by PCR (AATGAACACAGCCAAGCAAG and TTGATTAACCCTTAGCATGTCC for transgenic probe, AAATGAACACAGCCAAAGCAAG and CCGAAAGCACATATTCCCAAC for genomic probe) and subcloned into pCR Topo II (Invitrogen) using the TOPO TA cloning kit (Invitrogen). Plasmid was digested with XbaI or BamHI and transcribed using Ambion's (Austin, TX) MaxiScript kit with T7 or Sp6, depending on orientation of the insert. Because of the length of the transgenic probe, this transcription reaction was supplemented with 1 µl of 60 µM unlabeled UTP. Either 50,000 cpm of genomic or 100,000 cpm of transgenic probe and 10 µg total RNA were used for each reaction. RNase protection assays were carried out using Ambion's HybSpeed RPA kit per manufacturer's instructions, using only RNase T1. Reactions were run on a denaturing sequencing gel (8 M urea, 4% polyacrylamide) and exposed to film.

Larvae for squashes were obtained by crossing roX1^ex6 Df(1)52 [w⁺ 44.3] virgins (proximal recombination event only; Table 1 and Supplemental Fig 2 at http://www.genetics.org/supplemental/) to homozygous transgenic males with a y w roX1^ex6 X chromosome, reared at 18° in uncrowded vials, and stained as described previously for MSL1 (KELLEY et al. 1999 ) and for roX1 (MELLER et al. 2000 ). To exclude squashing nondisjunction males, only larvae with y⁺ mouthhooks were analyzed. The template for the tra2 probe used to detect hybrid RNAs on the X chromosome (Fig 7) was made by PCR amplification using primers GCGACGATTAACTTGGTGAATG and CTGCAGCTCATCGACCAATTCAGCAC.

View larger version (31K): In this window In a new window Download PPT slide   Figure 7. Abnormal processing at the roX1 3' end. (A) Overview of RNase protection probes used. Both probes have the same roX1 sequence at their 5' end (indicated by nt position from cDNA c20), but are fused to sequence from the injection vector (transgenic probe) or genomic sequence (genomic probe). Lollipops indicate predicted polyadenylation sites. The dashed line below the transgenic probe diagram indicates the tra2 probe used for C. (B) RNase protection. Diagrams of the protected products are shown to the right of the autoradiograph. These correspond to the probe maps shown in A. (C) RNA in situ hybridization of y w roX1ex6; [w+ wt roX1] male larvae using the tra2 probe indicated in A.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Complementation of roX1^- roX2^- double mutants with roX1 transgenes:
Most males carrying a roX1^- roX2^- double-mutant X chromosome die, but can be rescued if one of the roX RNAs is expressed from a cDNA transgene inserted on an autosome (MELLER and RATTNER 2002 ). However, when either the 5' 900 nt (construct 5' roX1) or the 3' 2.4 kb (construct 3' roX1) is expressed, neither is sufficient for rescue (MELLER and RATTNER 2002 ). Therefore, to map functionally important domains within the RNA, we constructed a series of roX1 transgenes deleting 10% segments and tested them for rescue of male lethality.

We set up the complementation test outlined in Fig 1 using a modified roX1^- roX2^- X chromosome (see MATERIALS AND METHODS and Supplemental Fig 1 at http://www.genetics.org/supplemental/). roX1^- roX2^- females were crossed to males carrying a y⁺ X chromosome and a roX1 transgene balanced by either CyO or TM3, depending on whether the insertion of the transgene was on the second or third chromosome. The balancer allowed us to distinguish between two classes of male offspring: rescued males, i.e., males surviving because of expression of a roX1 transgene, and escaper males, i.e., males surviving despite a lack of roX RNAs. The frequency of either rescued males or escaper males is calculated as the ratio of these males to their respective sisters (class 1 males/class 1 females for the rescue frequency; class 2 males/class 2 females for the escaper frequency; Fig 1). A complication arises because the mothers in the complementation test are also mutant for nod, a kinesin motor required for the fidelity of nonrecombinant chromosome segregation during female meiosis (ZHANG et al. 1990 ). Females mutant for nod produce a high frequency of gametes carrying either zero or two X chromosomes (ZHANG and HAWLEY 1990 ). This enables abnormal father-to-son transmission of the paternal X chromosome, giving rise to male offspring with an X chromosome wild type for both roX1 and roX2. By marking the paternal X chromosome with y⁺, it is possible to identify such nondisjunction XO offspring, which occurred at a frequency of 3% (classes 3 and 4 in Fig 1).

Since transgenes insert randomly into chromatin and are subject to position effects, we analyzed several insertions per construct. For each, we selected the transgene with the highest rescue frequency as being representative of its rescue potential and used it for further analysis. Our rationale for following this approach, rather than deriving an average value for each set of insertions, was that poor rescue should reflect a functional defect in the RNA, not low abundance of a fully active RNA. However, comparing the averaged rescue values for each construct yielded results very similar to those obtained by comparing only the highest rescue values (data not shown).

Using the complementation cross outlined above, we found that an unmutated roX1 cDNA c20, expressed from the constitutively active Hsp83 promoter (MELLER et al. 2000 ), was capable of substantial rescue in the sense orientation, but not in the antisense orientation (66 vs. 4%; Fig 2). Likewise, both 5' roX1 and 3' roX1 were significantly impaired in their ability to rescue male lethality (3 and 6%, respectively; Fig 2). The frequency of escaper males was <1%. Repeating the complementation test with y⁺ males without a transgene, we also found double-mutant males at <1% (Fig 2). These rare escapers died within a few days of eclosion (data not shown).

Since roX1 and roX2 share an essential function in dosage compensation, we tested whether the short stretch of primary sequence homology (25/30-nt match located at the 3' end of each RNA; FRANKE and BAKER 1999 ) was required for function. We found that roX1 RNA lacking the 30-nt motif (roX130) was able to rescue at a level similar to that of the wild-type cDNA construct (65 vs. 66%; Fig 2). The 30-nt motif was previously found neither necessary nor sufficient for MSL binding to the roX1 gene (KAGEYAMA et al. 2001 ). Therefore, the function of the only identified sequence common to roX1 and roX2 remains elusive.

To scan the entire roX1 gene, we created a series of 11 overlapping deletions in the roX1 cDNA c20. Each deletion was between 260 and 400 nt in length (Fig 2). Since the roX1 c20 cDNA misses 200–300 nt of sequence from the roX1 5' end (R. L. KELLEY, personal communication), it in itself represents a deletion. Fig 2 is a summary of our findings with the deletion constructs using the highest rescuing constructs. We placed these (and subsequent) mutants into three broad classes based on rescue of male viability. Most deletions, like roX11–roX19, rescued about as well as the wild-type c20 cDNA. Almost 80% of roX1 sequence can be deleted in small intervals without obvious consequence. A second group, here represented by roX111, gave weaker rescue of 20%, suggesting impaired function. Finally, the transgene roX110 rescued only slightly better than the roX1 antisense negative control and the 3' roX1 construct.

We noted that the 3' roX1 construct, which removes the 5' 900 nt of roX1, was impaired in its ability to rescue, but that none of the small deletions in this region (roX11–roX13) showed a similar reduction in rescue ability. One explanation for these findings is that several redundant elements are contained within the 5' 900 nt of roX1. roX11–roX13 may remove only one of these elements and therefore encode an RNA that is functional. Whether these elements are important for RNA function or stability is unclear.

The deletion analysis pointed to the 3' end of the RNA as playing an important role in roX1 function. Both roX110 and roX111 had poor male rescue activity, with roX110 showing the least rescue (7%). An analysis of the region deleted in these transgenes revealed a large inverted repeat in roX1, which was predicted to form a stem-loop structure by computer analysis with mfold (GCG; see Fig 3). Since stem-loops are important structural features of RNAs and are frequently the site of RNA-protein interactions, we analyzed the importance of this stem-loop by removing the sequence deleted in both roX110 and roX111 (Overlap) or precisely removing the inverted repeat (Inverted Repeat). In addition, we created site-directed mutants within the inverted repeat predicted to either disrupt the potential stem-loop structure (5' and 3' stem mutants) or restore the structure with a different primary sequence (double mutant). We also scrambled the sequence inside the seven-nucleotide loop (loop mutant).

All transgenes were assayed in the complementation test described in Fig 1. We found that the inverted repeat and its associated potential stem-loop structure, but not the actual nucleotide sequence of the inverted repeat, were important for full roX1 function (Fig 3). All transgenes deleting a portion of the inverted repeat (Overlap, Inverted Repeat) resulted in impaired rescue frequency (27 and 22%, respectively) similar to the rescue frequency obtained with roX111 (20%), but not as low as for roX110 (7%). Transgenes destroying the structure by site-directed mutagenesis also resulted in impaired rescue (27 and 25%), similar to roX111. In contrast, transgenes changing the sequence, but not the structure of the stem-loop (loop mutant and double mutant) rescued at levels similar to that of the wild-type cDNA (52 and 57% vs. 66%). We conclude that the inverted repeat and its associated stem-loop structure play an important role in roX1 function. We propose that roX110 removes a second upstream element in addition to the inverted repeat, as none of the stem-loop mutants are as severe as roX110.

Localization of mutant RNAs to the X chromosome:
Since roX110 and roX111 were able to rescue only partially, we asked if the mutant transgenic RNAs could paint the X chromosome. Polytene chromosomes were prepared from male third instar larvae whose only source of roX RNA was the transgene. roX1 RNA localization to the X chromosome was analyzed by in situ hybridization with digoxigenin-labeled antisense roX1 riboprobes (MELLER et al. 2000 ). We found that neither roX110 nor roX111 was capable of painting the male X chromosome normally (Fig 4, A–D). We observed weak staining of the X chromosome, but also additional staining of the chromocenter and autosomes. This RNA in situ pattern is reminiscent of MSL staining in roX1^- roX2^- larvae without a rescuing transgene (MELLER and RATTNER 2002 ; Fig 5A and Fig B), but more X staining was observed. The significance of the ectopic association of mutant MSL complexes with autosomes and the chromocenter remains unknown. Although roX111 rescued better than roX110 (Fig 2, 20 vs. 7%), the roX111 staining pattern did not appear significantly better than that of roX110. In contrast, roX19, a deletion that rescues well, showed strong staining of the mutant RNA on the X chromosome (Fig 4E and Fig F). All site-directed stem-loop mutant RNAs appeared to paint the X chromosome normally (data not shown). Therefore, the RNA in situ assay did not distinguish between transgenic lines that rescued well and ones that exhibited only intermediate rescue. However, RNA localization in the roX110 and roX111 transgenic lines was clearly weaker than that in all other lines tested.

Transgenic roX1 RNAs restore MSL localization to the X chromosome:
Since the MSL proteins are unable to bind the male X normally without roX RNA (MELLER and RATTNER 2002 ), we tested whether our set of mutant RNAs could direct the MSL complex to the X chromosome. In the roX1^- roX2^- double mutants, MSL proteins are found at only a few sites on the X chromosome and, aberrantly, at the chromocenter and at a number of autosomal sites (Fig 5A and Fig B). The wild-type roX1 cDNA c20 expressed from a transgene is capable of restoring the normal X staining pattern of the MSL complex (Fig 5C and Fig D). Of the deletions, roX11–roX19 showed normal MSL staining (data not shown), consistent with their ability to rescue roX1^- roX2^- male lethality at levels similar to that of the wild-type c20 cDNA. roX110 and roX111, however, showed autosomal bands in addition to weak X staining, which appeared less dense than normal (for roX110 see Fig 5E and Fig F; similar results were observed for roX111). The autosomal MSL1 staining pattern was reminiscent of that of roX1^- roX2^- larvae (Fig 5A and Fig B). We observed weak levels of H4Ac16 on the X chromosome in roX110 and roX111 (data not shown), suggesting that these mutant RNAs are capable of at least partially restoring functionality to the complex.

Overlap showed increased X staining compared to roX110 and roX111, but still showed more ectopic staining than wild type showed (Fig 5G and Fig H). roX130, Inverted Repeat, and site-directed stem-loop mutants showed normal X staining (data not shown). Since Inverted Repeat and 5' and 3' stem mutants rescue at a lower frequency than that of the wild-type cDNA, their apparently normal X staining pattern is puzzling. These RNAs may therefore paint the X chromosome and cause the MSL complex to localize in a wild-type pattern, without being fully functional. This has been observed for the Xist RNA, where the ability to silence the X chromosome and to paint the X chromosome was genetically separated (WUTZ et al. 2002 ). On the other hand, since MSL immunostaining is not quantitative, subtle differences in the level of MSL complex on the X chromosome would not be detected. Consistent with this possibility, Inverted Repeat and 5' and 3' stem mutant RNAs are found at lower steady-state levels by Northern analysis (see below).

Alteration of 3' processing in roX1 3' deletion mutants:
Previous work has shown that roX1 RNA is unstable unless the MSL proteins are also coexpressed (AMREIN and AXEL 1997 ; MELLER et al. 1997 , MELLER et al. 2000 ). To determine whether mutant RNAs were capable of assembling with MSL proteins, we tested all roX1 transgenics by Northern analysis for stable RNA accumulation and integrity. We found that all transgenes make stable transgenic RNA that can be readily detected (Fig 6), suggesting that these RNAs do assemble with MSL proteins. This was particularly surprising for roX110 and to a lesser extent for roX111, since these transgenes rescued poorly and the RNA was impaired in painting the X chromosome. However, it was generally true that transgenic lines rescuing at a lower frequency exhibited less roX1 RNA. All transgenes that removed or destroyed the stem-loop structure and had impaired rescue also showed lower levels of roX1 RNA. On the contrary, it was not strictly true that lower RNA levels correlated with poor rescue. Both roX12 and roX13 had lower levels of roX1 RNA, yet rescued at least 46%.

The unmutated roX1 cDNA encodes an RNA of 3.4 kb. Most of the deletions (roX11–roX19) showed the expected reduction in size of 300 nt. However, roX110 and roX111 transgenic RNAs were abnormally long: the deletions of 349 or 341 nt, respectively (Fig 2), should reduce the size of these RNAs to a size similar to roX11–roX19. Instead, roX110 and roX111 encode transcripts of approximately the same length as the wild-type cDNA transgene. The inverted repeat mutants also showed an upper band (3.7 kb) in addition to the expected band (3.4 kb).

Little is known about roX1 RNA transcription or processing. All transgenes in this study use the Hsp83 promoter (a PolII promoter), but it is not known which RNA polymerase transcribes the endogenous roX genes. roX RNAs are spliced, but it is not clear if they are also capped and polyadenylated, since none of the isolated roX1 cDNAs contain long poly(A) tails. Because the mutants at the 3' end of the RNA showed the most prominent size discrepancies, we focused our analysis on the 3' processing of the roX1 transgenic RNAs. Shown in Fig 7A is a schematic of this region in the pCaSpeR vector. Downstream of the roX1 3' end is the 3' genomic sequence from tra2 containing a polyadenylation site (MELLER et al. 2000 ). The pCaSpeR vector sequence also contains a potential polyadenylation site. One explanation for the longer mutant transcripts was failure to terminate at the tra2 polyadenylation site and read through to the pCaSpeR polyadenylation site. Using RNase protection assays, we found that normal, unmutated roX1 transgenes terminate at the 3' end of the cDNA, even though they have been removed from their genomic context and the cDNA lacks a consensus polyadenylation site in this region (compare Fig 7B, transgenic wild-type roX1 c20 RNA and y w RNA with both transgenic and genomic probes). roX110, however, completely failed to terminate at this site; its RNA was not processed until reaching the tra2 polyadenylation sites. We noted that even wild-type cDNA transgenes encode some transcripts that read through the normal roX1 3' processing site. Using an RNA probe complementary to the tra2 sequence that would detect only readthrough transcripts (Fig 7A), we found that hybrid transcripts with this extra sequence were capable of painting the X chromosome with normal specificity (Fig 7C). We conclude that the presence of the additional tra2 sequences at the 3' end of some roX1 transgenic RNAs does not interfere with their localization, although an effect on function cannot be assayed. The increased presence of heterologous 3' sequences on mutant RNAs could be a contributing factor to the phenotype of the mutant RNAs. However, the localization of unmutated hybrid RNAs to the X chromosome suggests that the reduced localization and function of mutant RNAs is most likely a consequence of the altered or deleted roX1 sequences.

roX1^- roX2^- escaper frequency variability:
MELLER and RATTNER 2002 reported 3.5% escaper males that appear healthy in the absence of roX RNAs. When these males are crossed to roX mutant females, the escaper frequency appears constant, suggesting that the recovery of mutants is stochastic rather than due to a genetic alteration. In our complementation tests, we consistently recovered smaller numbers of escapers than reported previously (<1%). When constructing the roX1^- roX2^- X chromosome for this study, we recombined y^- and w^- alleles onto the distal end of the original X isolated by MELLER and RATTNER 2002 . In addition, we recombined a cosmid transgene [w⁺ 44.3], inserted at position 18F-19A in the proximal X, to rescue the embryonic lethality of flanking genes deleted in the roX2 deficiency, Df(1)52. By examining various intermediate recombinant chromosomes, we noted that the lower escaper frequency could be mapped to the replacement of the proximal X chromosome (Table 1). The insertion site of [w⁺ 44.3] in the proximal X might be detrimental to subviable roX1^- roX2^- mutant males, although it is tolerated in homozygous females. Alternatively, other loci on the X chromosome proximal to the roX2 locus might influence the ability of males to live without roX RNAs.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The 3.7-kb roX1 RNA is predicted to make up over half the mass of the MSL complex. Yet, surprisingly, our principal finding is that most of the roX1 is dispensable for function, as constructs lacking almost any 10% of the sequence still rescued roX1^- roX2^- double-mutant male lethality similar to a wild-type roX1 cDNA. Comparison of a 900-bp deletion construct (3' roX1), with the overlapping 300- to 400-nt deletions roX11–roX13 showed that this region contains at least two redundant elements. It is possible that several additional redundant elements are located within the region of roX1 defined by deletions roX14–roX19. Our finding that most small domains are dispensable is consistent with the absence of primary sequence homology between the functionally interchangeable roX1 and roX2 RNAs. Whatever roX RNAs do, their functions likely depend on multiple, complex tertiary folds, not primary nucleotide sequence. The one short 30-nt sequence element shared between roX1 and roX2 could be deleted without obvious consequence. This finding is similar to results obtained with Xist, where conserved sequence elements were not necessarily important for RNA function (WUTZ et al. 2002 ). The deletion in roX14 is identical to a deletion called roX1DHS that removes a male-specific DNaseI hypersensitive (DHS) site in the roX1 gene (KAGEYAMA et al. 2001 ). The DHS site is the principal DNA-binding site for the MSL complex within the roX1 gene, but here we find that the sequence is dispensable in the roX1 RNA.

In addition to functional redundancy in the 5' end of roX1, we found that the 3' end is most important for full function of the RNA. Deleting sequences within the last 600 nt of the RNA resulted in markedly lowered rescue efficiency and abnormal RNA and MSL complex localization. A predicted stem-loop structure in this region is partially responsible for the loss of rescue activity, but we suspect a second, partially redundant element is removed by the roX110 deletion.

roX1 RNA may not undergo typical polyadenylation, but our studies do not address this in detail. Whether roX1 RNA is polyadenylated or is enriched on oligo(dT) columns through internal A-rich stretches is not known. The absence of consensus polyadenylation sites raises the possibility that an unusual mechanism might be employed to terminate the RNA. This mechanism might be important for keeping the RNA in the nucleus and thus for localization of both the MSL proteins and the RNA to the X chromosome (ZHAO et al. 1999 ).

No physical information about the structure of the MSL RNA-protein complex is available. We do not know whether both roX1 and roX2 are present in one MSL complex or if two different complexes exist, one with roX1 and one with roX2. It is not known whether roX1 RNA is flexible or adopts a rigid structure in the MSL complex, possibly stabilized by many weak RNA-protein interactions. However, our results are compatible with a model in which the RNA is more flexible and perhaps surrounds an MSL protein core. In one view of dosage compensation, the histone-modifying enzymes account for the altered chromatin found on the male X and the resulting twofold increase in transcription. The main function of the roX RNAs might therefore be to restrict MSL action to the X chromosome in vivo. This is consistent with our observations that roX110 and roX111 fail to rescue well and mislocalize the mutant MSL complexes to many autosomal sites. It might also account for the presence of escaper roX1^- roX2^- males: since MSL complexes still bind some sites on the X chromosome in the absence of any roX RNAs, enzymatically active MSL proteins might weakly act on the entire genome. Weak H4Ac16 histone acetylation has been observed in roX1^- roX2^- males (MELLER and RATTNER 2002 ); therefore, rare males might have sufficient levels of histone modifications on the X chromosome without fatal levels of increased autosomal activity.

Our finding of extensive redundancy within the roX1 RNA is reminiscent of a recent study examining Xist using a similar deletion strategy. This study also found multiple redundancies within the Xist RNA (WUTZ et al. 2002 ). Additional examples from other areas of RNA biology include redundant RNA localization signals in the 3' untranslated region of bicoid in Drosophila and of Vg1 in Xenopus and interactions between HIV Rev and its cognate RNA element, the RRE (KARN et al. 1994 ; GAUTREAU et al. 1997 ; MACDONALD and KERR 1997 ). The interaction between Rev and the RRE may prove relevant: the Rev protein binds at a bulge in the RRE, a large RNA structure. Once the first Rev molecule is bound to the RRE, however, additional Rev molecules can load onto the same RNA (DALY et al. 1993 ; KARN et al. 1994 ). It is possible that the region within roX110 is required for the efficient binding of a first MSL protein to the RNA. Once this first protein is bound, more MSL proteins can bind to the RNA and make it fully functional. In this model, most of the RNA would consist of secondary MSL-binding sites of lower affinity, which could result in inefficient loading of the MSL complex to the roX110 mutant RNA.

roX1 and roX2 are redundant with each other (FRANKE and BAKER 1999 ; MELLER and RATTNER 2002 ) and we have demonstrated functional redundancy within roX1. If both RNAs exert their function through a common tertiary structure, this structure must be able to form despite completely different primary sequences. There are examples of proteins that fold into highly similar structures despite their lack of sequence identity (PERUTZ et al. 1965 ; KRISHNA et al. 1994 ; MITTON-FRY et al. 2002 ). The versatility in intramolecular interactions within RNAs may be instrumental in allowing them to assume many varied tertiary structures (DOUDNA 2000 ), but most RNA structures are still completely unknown. Presumably RNAs as different as roX1 and roX2 fold into a common tertiary structure necessary for dosage compensation in Drosophila.

Note added in proof:
A second conserved sequence element present in both roX1 and roX2 was recently reported (Y. PARK et al., 2003, Sequence-specific targeting of Drosophila roX genes by the MSL dosage compensation complex. Mol. Cell 11: 977–986).

	ACKNOWLEDGMENTS

We thank R. L. Kelley for numerous helpful suggestions in experimental design and X. Bai, R. L. Kelley, and Y. Park for critical reading of the manuscript. We are grateful to P. R. Gordadze, H. G. Kennedy, X. Chu, and C. Olson for excellent technical assistance. This work was supported by grants from the National Institutes of Health and the Welch Foundation. M.I.K. is an Investigator of the Howard Hughes Medical Institute.

Manuscript received October 28, 2002; Accepted for publication March 13, 2003.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AKHTAR, A. and P. B. BECKER, 2001  The histone H4 acetyltransferase MOF uses a C₂HC zinc finger for substrate recognition. EMBO Rep. 2:113-118.[Medline]

AMREIN, H. and R. AXEL, 1997  Genes expressed in neurons of adult male Drosophila.. Cell 88:459-469.[Medline]

AVNER, P. and E. HEARD, 2001  X-chromosome inactivation: counting, choice and initiation. Nat. Rev. Genet. 2:59-67.[Medline]

DALY, T. J., R. C. DOTEN, P. RENNERT, M. AUER, and H. JAKSCHE et al., 1993  Biochemical characterization of binding of multiple HIV-1 Rev monomeric proteins to the Rev responsive element. Biochemistry 32:10497-10505.[Medline]

DOUDNA, J. A., 2000  Structural genomics of RNA. Nat. Struct. Biol. 7(Suppl.):954-956.

EDDY, S. R., 2001  Non-coding RNA genes and the modern RNA world. Nat. Rev. Genet. 2:919-929.[Medline]

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.[Medline]

FRANKE, A. and B. S. BAKER, 2000  Dosage compensation rox!. Curr. Opin. Cell Biol. 12:351-354.[Medline]

GAUTREAU, D., C. A. COTE, and K. L. MOWRY, 1997  Two copies of a subelement from the Vg1 RNA localization sequence are sufficient to direct vegetal localization in Xenopus oocytes. Development 124:5013-5020.[Abstract]

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.[Medline]

JIN, Y., Y. WANG, D. L. WALKER, H. DONG, and C. CONLEY et al., 1999  JIL-1: a novel chromosomal tandem kinase implicated in transcriptional regulation in Drosophila.. Mol. Cell 4:129-135.[Medline]

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.[Abstract/Free Full Text]

KAGEYAMA, Y., G. MENGUS, G. GILFILLAN, H. G. KENNEDY, and 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.[Medline]

KARN, J., M. J. GAIT, M. J. CHURCHER, D. A. MANN, I. MIKAÉLIAN et al., 1994 Control of human immunodeficiency virus gene expression by the RNA-binding proteins tat and rev, pp. 192–220 in RNA-Protein Interactions, edited by K. NAGAI and I. W. MATTAJ. Oxford University Press, New York.

KELLEY, R. L., 1993  Initial organization of the Drosophila dorsoventral axis depends on an RNA-binding protein encoded by the squid gene. Genes Dev. 7:948-960.[Abstract/Free Full Text]

KELLEY, R. L., V. H. MELLER, P. R. GORDADZE, G. ROMAN, and 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.[Medline]

KRISHNA, T. S. R., X.-P. KONG, S. GARY, P. M. BURGERS, and J. KURIYAN, 1994  Crystal structure of the eukaryotic DNA polymerase processivity factor PCNA. Cell 79:1233-1243.[Medline]

LEE, J. T., 2002  Is X-chromosome inactivation a homology effect? Adv. Genet. 46:25-48.[Medline]

LEE, J. T. and R. JAENISCH, 1997  Long-range cis-effects of ectopic X inactivation centres on a mouse autosome. Nature 386:275-279.[Medline]

MACDONALD, P. M. and K. KERR, 1997  Redundant RNA recognition events in bicoid mRNA localization. RNA 3:1413-1420.[Abstract]

MELLER, V. H. and M. I. KURODA, 2002  Sex and the single X chromosome. Adv. Genet. 46:1-24.[Medline]

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.[Medline]

MELLER, V. H., K. H. WU, G. ROMAN, M. I. KURODA, and R. L. DAVIS, 1997  roX1 RNA paints the X chromosome of male Drosophila and is regulated by the dosage compensation system. Cell 88:445-457.[Medline]

MELLER, V. H., P. R. GORDADZE, Y. PARK, X. CHU, and 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.[Medline]

MITTON-FRY, R. M., E. M. ANDERSON, T. R. HUGHES, V. LUNDBLAD, and D. S. WUTTKE, 2002  Conserved structure for single-stranded telomeric DNA recognition. Science 296:145-147.[Abstract/Free Full Text]

PANNUTI, A. and J. C. LUCCHESI, 2000  Recycling to remodel: evolution of dosage-compensation complexes. Curr. Opin. Genet. Dev. 10:644-650.[Medline]

PARK, Y., R. L. KELLEY, H. OH, M. I. KURODA, and V. H. MELLER, 2002  Extent of chromatin spreading determined by roX RNA recruitment of MSL proteins. Science 298:1620-1623.[Abstract/Free Full Text]

PERUTZ, M. F., J. C. KENDREW, and H. C. WATSON, 1965  Structure and function of haemoglobin II. Some relations between polypeptide chain configuration and amino acid sequence. J. Mol. Biol. 13:669-678.

SLEUTELS, F. and D. P. BARLOW, 2002  The origins of genomic imprinting in mammals. Adv. Genet. 46:119-163.[Medline]

SLEUTELS, F., R. ZWART, and D. P. BARLOW, 2002  The non-coding Air RNA is required for silencing autosomal imprinted genes. Nature 415:810-812.[Medline]

SMITH, E. R., A. PANNUTI, W. GU, A. STEURNAGEL, and R. G. COOK et al., 2000  The Drosophila MSL complex acetylates histone H4 at lysine 16, a chromatin modification linked to dosage compensation. Mol. Cell. Biol. 20:312-318.[Abstract/Free Full Text]

TURNER, B. M., A. J. BIRLEY, and J. LAVENDER, 1992  Histone H4 isoforms acetylated at specific lysine residues define individual chromosomes and chromatin domains in Drosophila polytene nuclei. Cell 69:375-384.[Medline]

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.[Medline]

WUTZ, A. and R. JAENISCH, 2000  A shift from reversible to irreversible X inactivation is triggered during ES cell differentiation. Mol. Cell 5:695-705.[Medline]

WUTZ, A., T. P. RASMUSSEN, and R. JAENISCH, 2002  Chromosomal silencing and localization are mediated by different domains of Xist RNA. Nat. Genet. 10:167-174.

ZHANG, P. and R. S. HAWLEY, 1990  The genetic analysis of distributive segregation in Drosophila melanogaster. II. Further genetic analysis of the nod locus. Genetics 125:115-127.[Abstract]

ZHANG, P., B. A. KNOWLES, L. S. B. GOLDSTEIN, and R. S. HAWLEY, 1990  A kinesin-like protein required for distributive chromosome segregation in Drosophila.. Cell 62:1053-1062.[Medline]

ZHAO, J., L. HYMAN, and C. MOORE, 1999  Formation of mRNA 3' ends in eukaryotes: mechanism, regulation, and interrelationships with other steps in mRNA synthesis. Microbiol. Mol. Biol. Rev. 63:405-445.[Abstract/Free Full Text]

This article has been cited by other articles:

				S.-W. Park, Y. I. Kang, J. G. Sypula, J. Choi, H. Oh, and Y. Park An Evolutionarily Conserved Domain of roX2 RNA Is Sufficient for Induction of H4-Lys16 Acetylation on the Drosophila X Chromosome Genetics, November 1, 2007; 177(3): 1429 - 1437. [Abstract] [Full Text] [PDF]