Recruitment of C. elegans dosage compensation proteins for gene-specific versus chromosome-wide repression

Sex is determined in many organisms by an X-chromosome counting mechanism that distinguishes one X chromosome from two (e.g. XO male/XX female in nematodes) or by the presence of a specific sex chromosome, such as the Y chromosome (e.g. XY male/XX female in mammals)(Meller and Kuroda, 2002; Meyer, 2000). In these organisms, a chromosome-wide regulatory process called dosage compensation copes with the difference in X-linked gene dose between males and females by equalizing levels of X-chromosome transcripts(Meller and Kuroda, 2002; Meyer, 2000). To achieve dosage compensation, specialized complexes are targeted to the X chromosome(s)of one sex to modulate X transcript levels. This sex-specific, chromosome-wide regulation is superimposed upon the regulation of individual X-linked genes that occurs in both sexes. Failure to accomplish dosage compensation causes sex-specific lethality. In the nematode C. elegans, dosage compensation is achieved by a protein complex that binds to both X chromosomes of hermaphrodites to reduce their transcript levels by half(Meyer, 2000). Remarkably, the dosage compensation complex is similar to the evolutionarily conserved 13S condensin complex required for mitotic and meiotic chromosome resolution and compaction, implying the recruitment of ancient chromosome segregation proteins to the new task of regulating gene expression(Chuang et al., 1994; Hagstrom et al., 2002; Lieb et al., 1998; Lieb et al., 1996).

Typically, protein complexes that regulate gene expression across entire chromosomes or subchromosomal domains do not function as specific regulators of individual genes. However, the C. elegans dosage compensation complex is unusual in this regard. Not only does the complex repress expression of X chromosomes by twofold, it represses transcription of the autosomal sex determination gene her-1 by 20-fold(Chu et al., 2002; Dawes et al., 1999). This contrast led us to investigate how a protein complex achieves uniformly weak repression of numerous genes in one context and strong repression of a specific gene in another. The work presented here on the dosage compensation gene dpy-21 reveals distinctions in the composition and recruitment of the complexes that achieve these two levels of repression.

Prior research showed that sex determination and dosage compensation in C. elegans are coordinately controlled in response to the X-chromosome counting mechanism (DeLong et al., 1993; Klein and Meyer,1993; Miller et al.,1988; Nusbaum and Meyer,1989; Rhind et al.,1995; Villeneuve and Meyer,1987). In XX embryos, SDC-2 is the pivotal factor that initiates dosage compensation (Dawes et al.,1999). It is the only dosage compensation protein known to be expressed exclusively in hermaphrodites, and it can localize to X chromosomes independently of other dosage compensation proteins. SDC-2 acts together with SDC-1 and SDC-3 to trigger assembly of the condensin-like dosage compensation proteins (DPY-26, DPY-27 and MIX-1) onto hermaphrodite X chromosomes to reduce gene expression by half. In fact, ectopic expression of SDC-2 in males is sufficient to trigger assembly of this complex onto the single X chromosome,causing death from inappropriately low X-linked gene expression. The SDC complex also induces hermaphrodite sexual development in XX embryos by associating with the promoter of the male sex-determining gene her-1to repress its transcription (Chu et al.,2002; Dawes et al.,1999). Again, SDC proteins recruit DPY-26, DPY-27 and MIX-1, this time to her-1 regulatory regions(Chu et al., 2002). This localization, together with the observation that dpy mutations affect sexual fate in specific genetic backgrounds, suggests that DPY proteins may act directly with SDC proteins to repress her-1.

dpy-21 is also required for dosage compensation; however, dpy-21 differs significantly from other dosage compensation genes. Like mutations in other dosage compensation genes, dpy-21 mutations cause elevated X-linked gene expression and morphological phenotypes dependent upon X-chromosome dose: XO animals appear wild type, while XX animals are dumpy and egg-laying defective (L. DeLong, PhD thesis, Massachusetts Institute of Technology, 1990) (DeLong et al.,1987; Hodgkin,1983; Meneely and Wood,1984; Meneely and Wood,1987; Meyer and Casson,1986; Plenefisch et al.,1989). But unlike other dosage compensation mutations, which disrupt the stability or X-localization of the dosage compensation complex,causing extensive XX-specific lethality, dpy-21 mutations, shown here to be null, cause no embryonic lethality and only infrequent larval lethality(Chuang et al., 1996; Lieb et al., 1998; Lieb et al., 1996; Plenefisch et al., 1989). Consistent with the weak dosage compensation phenotype, dpy-21 null mutations cause no obvious defect in the assembly or localization of the dosage compensation complex (Chuang et al.,1996; Davis and Meyer,1997; Lieb et al.,1998; Lieb et al.,1996). We place DPY-21 in the molecular pathway of dosage compensation.

We show that DPY-21 is a member of the dosage compensation complex and localizes to X chromosomes in a sex-specific manner. DPY-21 is the first member of the dosage compensation complex that does not also localize to her-1. Moreover, we show that recognition of her-1 target DNA is initiated by a different SDC protein from that responsible for initial recognition of X-chromosome targets. These findings represent the first molecular differences in the repression complexes that achieve gene-specific versus chromosome-wide regulation.

dpy-21 was mapped with sequence tagged site (STS) polymorphisms to a 500 kb interval of chromosome V located between cosmids K03D8 and C48B12 and covered by YACs Y59A8 and Y94A7. To identify cDNA clones from this interval,the Data Base of Expressed Sequence Tags (dbEST from Y. Kohara, National Institute of Genetics, Mishima, Japan) was searched with the partially assembled YAC sequences using the Basic Local Alignment Search Tool (BLAST). The Smallest Sum of Probability score [P(N)] was calculated between the sequence of each EST and the YAC sequences to determine the probability that the two sequences had the same overall similarity by chance. ESTs with a P(N)score less than e^-10 were chosen for further study, resulting in 108 ESTs. EST candidates were grouped into 30 families based on predicted amino acid sequence similarity to related proteins. This approach prevented the analysis of multiple ESTs from the same ORF. Six out of 30 cDNA families were judged unlikely to participate in dosage compensation and were therefore eliminated from preliminary consideration.

Plasmids were excised from cDNA-containing phagemids (provided Y. Kohara),linearized with either KpnI or NotI, and used as template for either Ribomax (Promega) T3 or T7 transcription reactions. Single-stranded RNAs (ssRNAs) were injected at a concentration of 1 mg/ml.

ssRNA was tested for its ability to mimic a dpy-21 mutation in two genetic assays. In the first assay, single-stranded RNA (ssRNA) corresponding to each EST family was tested for its ability to rescue the XO-specific lethality caused by a xol-1 mutation. xol-1 mutations inappropriately activate the hermaphrodite program of dosage compensation in XO males, causing male lethality from reduced X-linked gene expression(Miller et al., 1988). Mutations in dpy-21 suppress the XO lethality by disrupting dosage compensation. ssRNAs were injected into a him-5(e1485); xol-1(y9)strain and the progeny were examined for males. The him-5 mutation causes X-chromosome non-disjunction and was included in the xol-1strain to generate XO animals. Partial rescue of male lethality was observed with ssRNA corresponding to cDNA clones yk132a2 and yk278g2.

In the second assay, the strongest xol-1 suppressor (ssRNA to yk132a2) was tested for its ability to suppress the masculinization caused by sdc-3(Tra). Only 1% of sdc-3(y52Tra) XX homozygous animals from sdc-3(y52Tra)/+ mothers are hermaphrodites at 20°C,but 65% of sdc-3(y52Tra) dpy-21(e428) XX animals from sdc-3(y52Tra) dpy-21(e428)/++ mothers are hermaphrodites(DeLong et al., 1993). To test yk132a2, ssRNA was injected into sdc-3(y52Tra)unc-76(e911)/sdc-3(y128Dpy) animals and the percentage of Unc Tra male and Unc hermaphrodite progeny was determined. Thirty-five percent of the Unc progeny (n=220) were hermaphrodites, indicating suppression similar to a dpy-21 mutation. Together, these assays indicate that yk132a2 is likely to represent dpy-21.

DNA sequence analysis of yk132a2 and yk278g2 showed that both cDNAs represented the same gene (ORF Y59A8b.1), a conclusion previously obscured by the fact that the cDNAs were incomplete and differed in their 5′ and 3′ ends. The true 5′ end of Y59A8b.1 was obtained using PCR with a primer to the C. elegans trans-splice leader SL1 and an internal primer.

Total C. elegans RNA was isolated as described(DeLong et al., 1993). mRNA was purified using Oligotex (Qiagen). dpy-21 cDNA was made by reverse transcription using primers DPY-21.51 (5′ GCAAATAGGGGTACTCCATTG 3′) and SuperScriptII (Invitrogen). To amplify dpy-21,first-round nested PCR used primers DPY-21.52 (5′ GATCTCATCGGGTAAAGGATTC 3′) and NOTSL1, which includes the SL1 splice leader and a NotI restriction site. Second-round nested PCR used primers DPY-21.53 (5′GTGTATGAAGCGAAGAACTTCG 3′) and NOTSL1.

Molecular lesions in the dpy-21 mutants were identified by sequence analysis of genomic DNA (e428 and y59) or RT-PCR products (e459, y43, y47, y188ts and y150ts). Sequence analysis was performed on three different DNA preparations for each mutation. The DNA changes were as follows: y428, CAG to TAG at codon 394; y59, CAG to TAG at codon 417; e459, GGA to GAA at codon 1291; y150ts, CTC to TTC at codon 1383; y88ts, AAG to GAG at codon 1396; y47, CAG to TAG at codon 1423. dpy-21(y43) has both a silent change at bp 4290 and a 134 bp deletion that removes the intron/exon 8 boundary. A splice site 87 base pairs upstream of the normal splice site is used, resulting in an in-frame deletion of 29 amino acids, as shown by sequence analysis of RT-PCR products.

RNA was made from wild-type embryos, L1, L2, L3 and L4 larvae, and non-gravid adults according to a protocol by G. Csankovszki. Trizol (10 ml,Invitrogen) was added to 2-3 ml of packed animals and the mixture homogenized using a polytron. The aqueous sample was subjected to three sequential extractions using 2 ml chloroform, then an equal volume of phenol and 0.2 volumes of chloroform, and, finally, an equal volume of chloroform. For each extraction, the phases were separated by centrifugation at 9000 g for 15 minutes at 4°C. The aqueous layer was mixed with 0.5 volumes of 2-propanol and incubated at room temperature for 10 minutes. The RNA was pelleted at 9000 g for 15 minutes at 4°C and washed with 70% ethanol. mRNA was purified from total RNA using Oligotex(Qiagen). Poly-A RNA (5 μg) was separated by gel electrophoresis, and northern analysis was performed (Wutz and Jaenisch, 2000) using random-primed probes transcribed from nucleotides 1-3347 of dpy-21.

Rabbit anti-DPY-21 antibodies were raised against a fusion protein composed of a GST tag and DPY-21 amino acids 1-173 that had been expressed from vector pGEX-5X-2 (Amersham) and purified using Glutathione Sepharose 4B (Amersham). Affinity purification of the anti-DPY-21 antibodies was performed as described(Davis and Meyer, 1997), using a 6×-His-tagged fusion protein that was made to the same DPY-21 region,expressed from vector pRSETA (Invitrogen), purified with Ni-NTA spin (Qiagen)and coupled to 6×Reacti-Gel (Pierce).

Rabbit anti-DPY-21 antibodies were also raised against a fusion protein composed of DPY-21 amino acids 467-1102, a GST tag at the N terminus and a 6×-His tag at the C terminus. The plasmid used to express the antigen was constructed by subcloning a RT-PCR product corresponding to nucleotides 1533-3460 of the dpy-21 transcript into pGEX-5X-2 vector that had been modified by adding 6 histidines prior to the stop codon. Affinity purification of these anti-DPY-21 antibodies was performed(Davis and Meyer, 1997) using the 6×-His-tagged fusion protein coupled to 6×Reacti-Gel (Pierce)with one modification. The serum was first passed through a GST-only column(gift of R. Chan) to remove antibodies against GST, and the flow-through was collected for antigen-specific purification.

SDC-3 was detected in embryos and in her-1 array experiments using affinity-purified rat polyclonal antibodies raised against amino acids 1067-1340 (from C. Tsai).

Extracts were prepared by boiling wild-type and mutant embryos in four volumes of SDS-PAGE loading buffer (100 mM Tris, pH 6.8, 2% SDS, 0.1%bromophenol blue, 7.5 M Urea, and 0.1 M DTT) for 10 minutes. Insoluble debris was removed by centrifugation at 9300 g for 1 minute. SDS-PAGE(Novex) and immunoblotting (Bio-Rad) using chemiluminescent detection (ECL,Amersham) were performed with the extract.

Lysates were prepared by sonicating embryos in chromatin buffer (200 mM sucrose, 5 mM Tris, pH 7.5, 80 mM NaCl, 1 mM CaCl₂) supplemented with a protease inhibitor cocktail (Calbiochem). Lysates were titrated to 5 mM EDTA and allowed to rock at 4°C for 20 minutes. Cellular debris was cleared by centrifugation at 4500 g for 5 minutes at 4°C. The supernatant was incubated with 5 μg primary antibody for 4 hours at 4°C and then incubated with Protein A Sepharose beads for 1 hour. Immunocomplexes bound to the Protein A Sepharose beads were washed with CHIP buffer (50 mM HEPES, pH 7.6, 140 mM KCl, 1 mM EDTA, and 0.5% NP-40). Proteins were eluted by boiling the beads in an SDS dye (50 mM Tris, pH 6.8, 2% SDS,0.1% bromophenol blue, 10% glycerol and 0.1 M DTT). Immunoprecipitates were analyzed by SDS-PAGE and immunoblotting.

Embryos were collected, fixed, and stained as described(Chuang et al., 1994; Davis and Meyer, 1997), except that embryos were fixed at room temperature for 15 minutes, and all washes were in PBST (0.14 M NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄,1.8 mM KH₂PO₄, 0.5% Triton X-100 and 1 mM EDTA, pH 8.0). Gut cells were stained as described (Howe et al., 2001). For confocal microscopy, animals were mounted in Vectashield (Vector Laboratories) containing 1 μg/ml of DAPI. Laser confocal microscopy was performed on a Leica TCS-NT confocal microscope. Over 100 embryos were analyzed in most immunostaining experiments.

The X-chromosome FISH protocol was developed in collaboration with G. Csankovszki and T. Wu based on the FISH protocol previously described(Dernburg and Sedat, 1998). Gravid hermaphrodites were removed from plates and washed with M9 (8.5 mM NaCl, 41 mM Na₂HPO₄, 8.5 mM KH₂PO₄and 18 mM NH₄Cl) and treated with hypochlorite to obtain embryos. After several washes with M9, an equal volume of embryos was added to sperm salts (50 mM PIPES, pH 7.0, 25 mM KCl, 1 mM MgSO₄, 45 mM NaCl and 2 mM CaCl₂) containing 3% paraformaldehyde (Electron Microscopy Sciences) on a poly-lysine coated slide and incubated in a humid chamber for 5 minutes. The slide was freeze cracked on dry ice, placed in 95% ethanol for 1 minute and washed with PBST. Subsequently, the slide was dehydrated for 2 minutes each in 70% ethanol, 80% ethanol, 95% ethanol and 100% ethanol, then allowed to air dry. X-chromosome fluorescent probe (10 μl) made by random primed labeling (Promega) of X-specific YAC DNA (gift of G. Csankovszki) was added. The slide was denatured at 95°C for 3 minutes, then incubated overnight in a humid chamber at 37°C. The slide was washed three times for 5 minutes each at 39°C with 2×SSC (0.3 M NaCl and 30 mM Na₃C₆H₅O₇) in 50% formamide, three times for 5 minutes each at 39°C with 2×SSC, and once for 10 minutes at 39°C with 1×SSC. The slide was rinsed in PBST and antibody staining was performed as described for immunostaining of adult germline and gut cells.

Localization of DPY-21 was analyzed in dosage compensation mutants. Dosage compensation mutations that cause XX-specific lethality were maintained in XO strains for which the hermaphrodite mode of sex determination and dosage compensation had been switched on by a xol-1 mutation. XO-specific lethality caused by the xol-1 mutation was suppressed by the dosage compensation mutation being assayed. For sdc-2 and sdc-3mutant strains, a her-1 mutation was included to allow the XO animals to develop as hermaphrodites. The genotypes of strains were:

sdc-1(n485);

dpy-26(n199) unc-30(e191) IV; lon-2(e678) xol-1(y9) V;

unc-32(e189) dpy-27(y167) III; flu-2(e1003) xol-1(y9) V;

dpy-28(s939) III; him-5(e1490); xol-1(y9) V;

her-1(hv1y101) V; xol-1(y9) sdc-2(y74) unc-9(e101) X; and

her-1(e1520) sdc-3(y126) V; xol-1(y9) X.

The dosage compensation mutations used in the staining experiments were molecular or genetic nulls.

Except for her-1 array strains carrying either an sdc-2or sdc-3 mutation, previously established her-1 array and control lines were used to examine DPY-21 localization(Chu et al., 2002; Dawes et al., 1999). All her-1 arrays contained the her-1 region indicated and plasmids encoding LacI::GFP and Lac O repeats. Animals were heat-shocked at 34°C for 30 minutes and allowed to recover at room temperature for 30 minutes to induce production of the LacI::GFP fusion protein, which is controlled by a heat-shock promoter.

We cloned dpy-21 to determine its molecular identity and its role in the regulatory hierarchy that controls dosage compensation. dpy-21was first mapped with sequence tagged site (STS) polymorphisms to a 500 kb interval of chromosome V. An RNAi strategy involving two genetic assays was then used to identify a candidate dpy-21 EST from the pool of ESTs correlated with this interval (Materials and methods). The EST yk132a2 emerged as the most likely candidate to represent dpy-21. The gene structure and complete DNA sequence of the ORF (Y59A8b.1) that corresponded to yk132a2 were determined.

The detection of molecular changes in ORF Y59A8b.1 from several dpy-21 mutant alleles proved that dpy-21 had been correctly identified (Fig. 1A). dpy-21(e428) is a nonsense mutation predicted to truncate DPY-21 at amino acid 394. This mutation causes the most severe dpy-21 mutant phenotype: 17% larval lethality and all adult survivors are dumpy (Dpy) and egg-laying defective (Egl). dpy-21(y59am) and dpy-21(y47am) are also nonsense mutations, predicted to terminate translation at codons 417 and 1396, respectively, and both can be suppressed by the amber mutant tRNA suppressor sup-7. y47 and e428 cause very similar phenotypes. dpy-21(e459) and the temperature-sensitive mutations dpy-21(y88ts) and dpy-21(y150ts) all have missense mutations(Fig. 1A). y88 causes the weakest phenotype: no lethality and all adults are Dpy and Egl. Finally, dpy-21(y43) has a 134 bp deletion that results in an in-frame deletion of 29 amino acids.

Previous genetic analysis of the 24 dpy-21 mutations was inconclusive regarding the null phenotype, as the chromosomal deficiencies that eliminate dpy-21 are haploinsufficient, causing lethality by themselves (L. DeLong, PhD thesis, Massachusetts Institute of Technology,1990). The identification of nonsense mutations and their effect on DPY-21 protein levels (see below), together with our finding that dpy-21RNAi treatment of dpy-21(e428) mutants does not enhance the mutant phenotype, strongly suggests that the weakness of the dpy-21 dosage compensation phenotype is not caused by incomplete disruption of the gene. Rather, dpy-21 is genuinely less essential than other dosage compensation genes.

The dpy-21 gene includes 11 introns and is predicted to encode a 5629 bp transcript. DPY-21 appears to be a novel, conserved protein of 1641 amino acids with a proline-rich N terminus. No other motifs could be identified based on sequence. However, homologs of DPY-21 share a unique,evolutionarily conserved C-terminal domain (amino acids 1230-1620)(Fig. 2). The biochemical functions of the proteins in this family are unknown.

Transcript levels of most dosage compensation genes peak during embryogenesis and decline during larval stages(Chuang et al., 1994; Klein and Meyer, 1993). To determine the developmental profile of dpy-21 transcripts, mRNA prepared from animals in each development stage was blotted and hybridized with a probe specific to the first 3347 nucleotides of dpy-21(Fig. 1B). The probe detected a single dpy-21 transcript that is expressed throughout development and migrates at ∼5700 bp, consistent with the predicted transcript size of 5629 bp. The dpy-21 transcript levels appeared highest during embryogenesis after normalization to the myo-1 (pharyngeal myosin)control transcript. This expression pattern resembles that of other dosage compensation genes.

All previously identified dosage compensation proteins form a complex that localizes to hermaphrodite X chromosomes and represses their gene expression(Chu et al., 2002; Chuang et al., 1996; Davis and Meyer, 1997; Dawes et al., 1999; Lieb et al., 1998; Lieb et al., 1996). To determine whether DPY-21 is a member of the dosage compensation complex,antibodies were raised to two different regions of DPY-21, the first 173 amino acids and an internal peptide (amino acids 467-1102) (Materials and methods). Two findings indicate that these antibodies are specific to DPY-21. First,both antibodies recognized a 210 kDa protein in extracts from wild-type gravid adults that was not detectable in extracts from either dpy-21(e428)or dpy-21(y59) mutant gravid adults(Fig. 3A and data not shown). The apparent molecular weight of this protein is slightly larger than the predicted size of 185 kDa. A control assessing levels of SMC-1, a chromosome cohesion protein, demonstrated that the dpy-21(e428) mutant and wild-type extracts had comparable levels of proteins(Fig. 3A). Second, the N-terminal antibody detected a protein of 60 kDa in the dpy-21(e428)mutant extract, consistent with the size of a truncated DPY-21 product predicted from the location of the nonsense mutation at codon 394(Fig. 3A). The truncated protein was detected variably.

To test whether DPY-21 associates physically with other members of the dosage compensation complex, immunoprecipitation reactions were performed using antibodies to several dosage compensation proteins. DPY-21 antibodies co-immunoprecipitated the dosage compensation protein SDC-3(Fig. 3B), and antibodies to the dosage compensation protein DPY-27 co-immunoprecipitated DPY-21(Fig. 3C). Three controls showed that the precipitation reactions were specific, allowing us to conclude that DPY-21 interacts biochemically with dosage compensation proteins. Mock reactions without antibodies (Fig. 3B), reactions with preimmune sera(Fig. 3B,C), and reactions with antibodies against CBP-1, a DNA-associated CREB-binding protein(Fig. 3C), failed to precipitate dosage compensation proteins.

Although DPY-21 interacts with other dosage compensation proteins, two differences were noticed in the behavior of DPY-21 compared with the other proteins. First, the immunoprecipitation reactions were not reciprocal in that DPY-27 antibodies immunoprecipitated DPY-21, but DPY-21 antibodies did not precipitate DPY-27. Second, antibodies to the dosage compensation protein DPY-26 precipitated DPY-21 only weakly, even though it precipitates DPY-27 and MIX-1 robustly. Although we do not know the reasons for these differences, we speculate that the antibodies might disrupt interactions between DPY-21 and other dosage compensation proteins or that the association between DPY-21 and the other proteins might not be as strong as that of other components. These observations suggest that the function of DPY-21 within the complex may be different from that of the other proteins.

If DPY-21 functions as a member of the dosage compensation complex in vivo,as predicted by the biochemical experiments, DPY-21 should localize to X chromosomes of XX embryos at or beyond the 40-cell stage, when SDC-2 recruits dosage compensation proteins to X. Immunofluorescence experiments using DPY-21 antibodies showed DPY-21 to be diffusely localized in nuclei of XX embryos(<40 cells) that had not yet activated dosage compensation(Fig. 4A). In embryos with greater than 40 cells, DPY-21 formed punctate, subnuclear foci that coincided with the X-localized SDC-3 foci (Fig. 4B). We showed directly that the DPY-21 foci co-localized with X chromosomes by probing simultaneously with DPY-21 antibodies and X-chromosome-specific DNA probes that were visualized by fluorescence in situ hybridization (FISH) (Fig. 5A). DPY-21 maintains its localization to X chromosomes throughout C. elegans development, like other dosage compensation proteins, as indicated by its presence on X chromosomes of adult gut nuclei(Fig. 4D). The in vivo specificity of both DPY-21 antibodies was confirmed by the absence of staining in the two dpy-21 mutant strains homozygous for early nonsense mutations, dpy-21(e428) (Fig. 4C and data not shown) or dpy-21(y59) (data not shown). These results establish that DPY-21 is recruited to the X chromosomes of hermaphrodites with other members of the dosage compensation complex at the onset of dosage compensation.

dpy-21 mutations disrupt X-linked gene expression in males as well as hermaphrodites; yet, disruption of a hermaphrodite-specific process is not expected to alter expression of the X chromosome in males. Paradoxically, of five X-linked genes assayed, dpy-21 mutations caused an elevation in transcript levels from four (Meneely and Wood, 1987; Meyer and Casson,1986) and a reduction in transcript levels from one(DeLong et al., 1987). Despite their altered transcript levels, dpy-21 mutant XO males appear phenotypically wild type. Our results show that in males, the effect of dpy-21 mutations on X-linked genes does not occur through the dosage compensation pathway. We stained XO male embryos with DPY-21 antibodies and an X-chromosome FISH probe to confirm that DPY-21 is expressed in males and to determine whether it localizes to X, as it does in hermaphrodites. In XO embryos of all ages, the pattern of localization resembled that in XX embryos that had not yet activated dosage compensation. DPY-21 was dispersed throughout the nucleus, in multiple foci of intense staining that were not coincident with the X chromosome (Fig. 4E). Thus, DPY-21 appears to influence gene expression in males through a route that is independent of dosage compensation. Consistent with this conclusion, mutations in the dosage compensation genes dpy-26,dpy-27 or dpy-28 do not suppress the reduction in X-linked gene expression caused by a dpy-21 mutation(Plenefisch et al., 1989).

Activity of the dosage compensation proteins is switched off in males by the male-specific gene xol-1, which represses the hermaphrodite-specific sdc genes and thereby prevents recruitment of the dosage compensation complex to the male X chromosome(Davis and Meyer, 1997; Dawes et al., 1999; DeLong et al., 1993; Miller et al., 1988; Rhind et al., 1995). If DPY-21 is controlled by the genetic hierarchy that regulates other dosage compensation proteins, we would expect DPY-21 to be localized ectopically to the X chromosome of xol-1 mutant XO animals. This expectation was met in the experiment of Fig. 4F.

In hermaphrodites, SDC-2, SDC-3 and DPY-30 are essential for the recruitment of all dosage compensation proteins to X chromosomes(Chuang et al., 1996; Lieb et al., 1998; Lieb et al., 1996). SDC-2 confers both hermaphrodite specificity and X-chromosome recognition to the dosage compensation process (Dawes et al.,1999). To explore the mechanism by which DPY-21 is recruited to X chromosomes, we determined whether mutation of individual dosage compensation genes blocks the recruitment of DPY-21 to X. In sdc-2 and sdc-3 mutants, DPY-21 was present but distributed throughout the nucleus in multiple foci of intense staining, as it is in males(Fig. 5A-C). These DPY-21 foci did not coincide with X chromosomes, as demonstrated by the combination of DPY-21 antibody staining and X-chromosome FISH(Fig. 5A-C). Therefore, SDC-2 and SDC-3 are required for the recruitment of DPY-21 to X.

In hermaphrodites, DPY-26, DPY-27, DPY-28 and MIX-1 have a complex dependence on each other for their stability and X localization. For example,without DPY-27, both DPY-26 and MIX-1 are stable but cannot localize to X(Lieb et al., 1998; Lieb et al., 1996). Without DPY-28, the DPY-26, DPY-27 and MIX-1 proteins are unstable(Chuang et al., 1996; Lieb et al., 1998; Lieb et al., 1996). Unlike these DPY proteins, DPY-21 does not require DPY-26, DPY-27 or DPY-28 for its accumulation and stability. However, DPY-21 does require DPY-26, DPY-27 and DPY-28 for its localization to X chromosomes(Fig. 5D-F). The dependence of DPY-21 on SDC and DPY proteins for its recruitment to X further indicates that DPY-21 downregulates X-linked gene expression in XX animals through the dosage compensation complex.

DPY-21 is not required for the stability or X localization of the other dosage compensation proteins, even though it is a member of the complex(Chuang et al., 1996; Davis and Meyer, 1997; Dawes et al., 1999; Lieb et al., 1998; Lieb et al., 1996). Our biochemial analysis further demonstrated this point. The level of MIX-1, a dual-functional protein with roles in both dosage compensation and chromosome condensation, is equivalent in wild-type and dpy-21(428) mutant extracts (Fig. 3A). Regarding the stability and localization of the complex, DPY-21 behaves like the dosage compensation complex member SDC-1 and unlike the other DPY proteins(Chuang et al., 1994; Davis and Meyer, 1997; Dawes et al., 1999; Lieb et al., 1998; Lieb et al., 1996). Nonetheless, the localization of DPY-21 and SDC-1 occurs independently. DPY-21 localizes to X chromosomes in sdc-1 mutants(Fig. 5G), and SDC-1 localizes to X in dpy-21 mutants (not shown).

In addition to activating dosage compensation, SDC proteins induce hermaphrodite sexual differentiation by repressing transcription of the male-specific, autosomal gene her-1(Chu et al., 2002; Dawes et al., 1999; DeLong et al., 1993; Trent et al., 1991). SDC proteins repress her-1 transcription 20-fold by binding to three separate regulatory regions within the gene(Chu et al., 2002). The first binding site overlaps the start point of her-1 transcription. The second and third sites lie within the second intron of her-1; each contains a 15 bp sequence that is necessary for SDC localization. The SDC proteins recruit the DPY components of the X-chromosome dosage compensation complex to her-1, implying the direct participation of these DPY proteins in gene-specific repression (Chu et al., 2002). Given that DPY-21 is a member of the repression complex on X, we asked whether DPY-21 is a member of the repression complex on her-1.

Recruitment of proteins to her-1 can be assayed in hermaphrodites carrying an extrachromosomal DNA array containing multiple tandem repeats of the her-1 regulatory regions, lac operator repeats(Straight et al., 1996), and a transgene encoding LacI::GFP (Chu et al.,2002; Dawes et al.,1999). LacI::GFP binds to the lac operator sequence,allowing the array to be visualized by GFP fluorescence. If a protein is recruited to her-1, an antibody raised against the protein will co-localize with GFP fluorescence.

Surprisingly, DPY-21 was not recruited to her-1. In both embryos(Fig. 6A) and adult gut nuclei(Fig. 6B) co-stained with SDC-3 and DPY-21 antibodies, SDC-3 co-localized with her-1 arrays and X chromosomes, but DPY-21 only co-localized with X chromosomes. DPY-21 provides the first indication that the repression complex on X differs from the repression complex on her-1.

Having found a difference in the composition of repression complexes at her-1 versus X, we determined whether the two repression complexes recognize their DNA targets in a similar or different way. SDC-2 confers X-chromosome recognition. It can localize to X chromosomes in the absence of other components of the dosage compensation complex, and it triggers assembly of these components onto X chromosomes(Dawes et al., 1999). We addressed whether SDC-2 was equivalently important for her-1recognition. We found that SDC-3 played the more central role in recognizing her-1.

While examining the localization of SDC-3 in young embryos (<30 cells)we noticed that SDC-3 could associate with her-1 regulatory regions at a time well before it localized to X chromosomes(Fig. 7A,B), suggesting another difference between the X and her-1 repression complexes. To verify this result, we assayed SDC-3 localization in four strains(Table 1A). Two strains contained independent arrays with all three SDC binding sites in her-1, and a third strain contained arrays with only SDC binding sites 2 and 3 (Fig. 7G). The fourth strain contained control arrays with only the vector sequences present in the other arrays. In embryos with fewer than 20 cells, SDC-3 localized infrequently to X chromosomes or to arrays with vector sequence only, but localized robustly to the arrays carrying her-1 regulatory regions(Table 1A). Thus, SDC-3 localizes to her-1 earlier than it localizes to X.

The fact that SDC-3 localized to her-1 before SDC-2 was detectable in embryos suggested that recruitment of SDC-3 to her-1 was independent of SDC-2. To test whether SDC-3 localization to her-1 is indeed SDC-2 independent, we examined SDC-3 localization in sdc-2(null) mutant embryos carrying her-1 arrays. SDC-3 co-localized with her-1 arrays but not with X chromosomes in sdc-2(null) mutants (Table 1B, Fig. 7C,D),demonstrating that localization of SDC-3 to her-1 does not require SDC-2. By contrast, SDC-2 failed to co-localize with her-1 arrays in sdc-3(null) mutants (Fig. 7F). Thus, although SDC-2 can localize to X without other dosage compensation proteins, SDC-2 requires SDC-3 for its recruitment to her-1. Conversely, SDC-3 localizes to her-1 independently of SDC-2, but requires SDC-2 for its localization to X. SDC-3 thus appears to play the lead role in recognizing her-1 sequences, whereas SDC-2 plays the lead role in recognizing X sequences.

The dependence of SDC-3 on other dosage compensation proteins for its recruitment to her-1 was also examined. SDC-3 is expressed in low levels in dpy-27(null) mutants and does not localize to X chromosomes(Davis and Meyer, 1997). By contrast, we found that SDC-3, like SDC-2(Chu et al., 2002), localizes to her-1 arrays in dpy-27(null) mutant embryos(Fig. 7E, Table 1B), providing another example that SDC-3 has different requirements for its recruitment to her-1 versus X. Together these results demonstrate that SDC-2 and SDC-3 assemble onto their regulatory targets in a different order and have different requirements for their localization. Moreover, different SDC proteins are pivotal for the recognition of her-1 versus X-chromosome targets.

The molecular analysis presented here revealed the direct participation of DPY-21 in the dosage compensation process. DPY-21 associates physically with other components of the dosage compensation complex and localizes to X chromosomes of XX embryos to repress gene expression. Recruitment of DPY-21 to X chromosomes is regulated by the genetic hierarchy that coordinately controls sex determination and dosage compensation. The discovery of DPY-21 as a member of the X-chromosome dosage compensation complex was somewhat unexpected given the relative weakness of dpy-21 mutant phenotypes compared with those caused by the loss of other members. Unlike other dosage compensation mutations, dpy-21 mutations do not cause extensive XX-specific lethality, nor do they overtly disrupt the stability or localization of the dosage compensation complex. All 24 dpy-21 alleles cause very similar phenotypes: elevated X-linked gene expression and an XX-specific dumpy,egg-laying-defective phenotype, both characteristic of rare adults that survive the lethality of more severe dpy mutations. The strongest dpy-21 alleles cause XX-specific lethality, but only at a low level(17%). Our molecular and biochemical analysis indicates that the phenotype of the strongest dpy-21 mutations represents the null phenotype. Previous genetic tests on this point had been inconclusive because the chromosomal deletions that remove dpy-21 are haploinsufficient on their own.

The partial disruption of dosage compensation caused by dpy-21null mutations can now be understood in the context of DPY-21 behavior. Although DPY-21 is a member of the dosage compensation complex, its association with the complex is not as stable as that of other members. Antibodies to dosage compensation proteins do not immunoprecipitate DPY-21 as readily as other components. Moreover, the stability of DPY-21 does not depend on the presence of other dosage compensation proteins, further suggesting that DPY-21 is not as integral to the complex as other members. Finally,recruitment of DPY-21 to X requires all other dosage compensation proteins except SDC-1. By contrast, the stability and/or X localization of these other dosage compensation proteins requires only a subset of dosage compensation proteins, implying an earlier and/or stronger association with the complex than DPY-21.

How might DPY-21 regulate X-chromosome gene expression? The DPY-21 protein sequence provided no clue as to its function. DPY-21, unlike the dosage compensation proteins DPY-26, DPY-27 and MIX-1, has no similarity to subunits of condensin, a complex that controls mitotic and meiotic chromosome structure. Rather than functioning in X-chromosome gene regulation through modification of X-chromosome structure, as proposed for the other proteins,DPY-21 might modify the activity of the dosage compensation complex, for example through a covalent modification or an allosteric effect. A more intriguing possibility is that DPY-21 might act directly on X chromatin to stabilize a repressed chromatin state initiated by the core dosage compensation complex, perhaps through histone modification or a direct association with chromatin.

The dosage compensation complex associates with X chromosomes of hermaphrodites to repress transcription twofold, while it associates with the regulatory sites of the male sex-determination gene her-1 to repress transcription 20-fold (Chu et al.,2002; Dawes et al.,1999). Our experiments revealed the first molecular differences in the repression at her-1 versus X: in the composition and recruitment of repression complexes and in the proteins that recognize these DNA targets.

Although DPY-21 localizes to X chromosomes as a component of the dosage compensation complex, it does not localize to her-1, unlike the other dosage compensation proteins. It is not yet known how this difference in DPY-21 localization contributes to the mechanisms of gene-specific versus chromosome-wide repression. DPY-21 does not, for example, counteract the activity of other dosage compensation proteins to limit the degree of X repression to only twofold. If it did, dpy-21 mutations would cause enhanced X repression rather than the diminished repression observed. Perhaps DPY-21 is absent from her-1 because its replacement in the repression complex with a different protein would increase the repressive capacity of the complex. Alternatively, DPY-21 may have a mechanism of action on X that would be detrimental to the development or viability of the organism if DPY-21 were allowed to bind to her-1. For example, if DPY-21 helps transmit the repressed chromatin state along X, then binding of DPY-21 to her-1might adversely influence the activity of genes neighboring her-1.

We have shown that SDC-2 and SDC-3 have different protein requirements for recruitment to their targets at her-1 versus X. Moreover, the SDC protein sufficient for her-1 target recognition is different from the one sufficient for X target recognition. SDC-2 can localize to X independently of SDC-3, but requires SDC-3 for its localization to her-1. SDC-2 triggers assembly of dosage compensation proteins onto X chromosomes and appears pivotal for X-chromosome recognition(Dawes et al., 1999). By contrast, SDC-3 can localize to her-1 independently of SDC-2 but requires SDC-2 for its localization to X. Similarly, SDC-3 does not require DPY-27 for its association with her-1 but it does require DPY-27 for its association with X. SDC-3 is essential for the localization of all dosage compensation proteins to her-1, and SDC-3 appears pivotal for her-1 target recognition.

These results concur with and extend previous genetic analysis indicating that SDC-3 functions differently at her-1 versus X. More specifically, the sex determination and dosage compensation activities of SDC-3, unlike SDC-2, are separately mutable: SDC-3 requires its zinc fingers to localize to X but its ATP binding motif to localize to her-1(Chu et al., 2002; Dawes et al., 1999; DeLong et al., 1993; Klein and Meyer, 1993). The initial mapping of recognition elements within her-1 identified three SDC binding sites, each of which differed in sequence, but two of which shared an identical 15 bp sequence essential for SDC binding(Chu et al., 2002). Because that 15 base pair sequence is not present on X chromosomes, it cannot be central for the recruitment of dosage compensation components to X chromosomes. This observation can now be interpreted in light of our finding that different SDC proteins play lead roles in recognizing her-1versus X regulatory targets. The 15 bp sequence might be part of a recognition sequence used by SDC-3 but not by SDC-2. In conclusion, our analysis has revealed important molecular differences in the composition and targeting of protein complexes that achieve both gene-specific and chromosome-wide regulation, opening the way to a deeper understanding of the mechanisms underlying these two major forms of gene regulation.

We thank L. DeLong for genetic analysis of the dpy-21 mutations;E. Cookson for initial SNP mapping of dpy-21; R. Chan, D. Chu, G. Csankovszki, E. Ralston, D. Reiner and C. Tsai for reagents and advice on experimental design; and R. Chan, T. Cline, E. Ralston, A. Severson and C. Tsai for thoughtful comments on the manuscript. This work was funded in part by NIH Grant R37 GM30702 to B.J.M. and NIH pre-doctoral training grants GM07232 and HD07375 to S.A.Y. B.J.M. is an investigator of the Howard Hughes Medical Institute.