Transcriptional activation in Drosophila spermatogenesis involves the mutually dependent function of aly and a novel meiotic arrest gene cookie monster

Differential control of gene expression during development is crucial for the specification and maintenance of differentiated cell types. One of the most dramatic switches in the pattern of gene expression occurs during spermatogenesis, many genes are switched on only in this cell type. Some are germline-specific homologues of ubiquitously expressed genes (e.g. β2 tubulin) (Kemphues et al.,1982) and the mitochondrial fusion protein fzo(Hales and Fuller, 1997),others are spermatogenesis specific proteins, e.g. the protamines that replace histones to tightly package sperm DNA(Peschon et al., 1987). Even genes transcribed in other tissues often have a separate, testis-specific,enhancer element in their promoter region, e.g. gonadal(Schulz et al., 1990).

In Drosophila spermatogenesis, transcription is essentially shut off upon entry into the meiotic divisions, therefore all transcripts required at later stages must be made in primary spermatocytes(Olivieri and Olivieri, 1965). Four mitotic divisions of a spermatogonial cell produce a cyst of 16 primary spermatocytes, which immediately undergo premeiotic S phase. They then enter an extended G2 period, characterised by high transcriptional activity and cell growth. The meiotic divisions result in a cyst of 64 round spermatids, which use stored mRNAs to produce proteins needed for their dramatic morphological changes during elongation, before finally individualising to form motile sperm(reviewed by Fuller,1993).

Although entry into spermatid differentiation is independent of progression through the meiotic divisions, these processes are subject to coordinate control, mediated by the `meiotic arrest' class of genes, including aly,can, mia and sa (Lin et al.,1996), and several other unpublished mutations (M. Fuller,personal communication). The meiotic arrest genes are essential for the transcription of many mRNAs involved in spermatid differentiation, and thus are required for spermatid differentiation. The meiotic arrest genes also control accumulation of proteins involved in the meiotic divisions, e.g. thecdc25 homologue Twine, and thus link differentiation to the cell cycle (White-Cooper et al.,1998). The meiotic arrest genes of Drosophila have been split into two classes, based on the mechanism by which they control accumulation of Twine. The can class (including can, mia andsa) post-transcriptionally regulate Twine production. By contrastaly regulates transcription of twine. Two other meiotic regulators, cyclin B and boule, are also transcriptional targets of aly, but not can, mia or sa.

Three meiotic arrest genes have been cloned to date, aly, can andnht (Hiller et al.,2001; White-Cooper et al.,2000). can and nht encode testis-specific homologues of TAF_II80 and TAF_II110, respectively. These are subunits of the basal transcription factor TF_IID, whose role is to facilitate RNA polymerase II binding to the proximal promoter region. Thealy (always early) gene discussed in this paper should not be confused with the regulator of RNA export Aly(Zhou et al., 2000).aly encodes a homologue of a C. elegans negative regulator of vulval induction, the SynMuvB gene lin-9. The SynMuv B pathway includes several genes whose products form a complex (NURD) that regulates chromatin structure (Lu and Horvitz,1998; Solari and Ahringer,2000).

The mechanism by which lin-9 regulates the NURD complex is not understood. LIN-9 has not been shown to be a component of the NURD complex itself, so may therefore act upstream. The subcellular localisation of LIN-9 in C. elegans has not been determined. We previously showed that Aly protein localised to chromatin in maturing Drosophila primary spermatocytes. This localisation suggests that the lin-9 homologue,Aly, may act in close concert with a NURD complex on chromatin(White-Cooper et al., 2000). The nuclear localisation of Aly protein is both regulated and essential for the normal function of the protein, as the protein produced by several mutant alleles remains cytoplasmic, despite the presence of a nuclear localisation signal.

We report the identification and cloning of cookie monster(comr), a novel aly-class Drosophila meiotic arrest gene. Like aly, comr transcription is testis specific in males, but low levels of transcript were detected at earlier stages of development. We show that Comr protein is localised to the nucleus of primary spermatocytes,and concentrated on decondensed regions of chromatin. The Comr pattern is similar but not identical to that of Aly. We show that the nuclear localisation of these two proteins is mutually dependent. Finally, we show that active RNA polymerase II is limited to discrete regions of the nuclei of primary spermatocytes. These regions of high transcriptional activity are a subset of the Comr localisation domain, but the level of Comr protein does not predict the level of active RNA polymerase II.

Drosophila were maintained on standard cornmeal/agar/sucrose medium at 25°C. All the markers used here have been described previously(Lindsley and Zimm, 1992) and at FlyBase (FlyBase, 1999). Deficiency chromosome lines were obtained from Bloomington Stock Centre. Wild-type flies were Oregon R for molecular biology or red e for immunostaining.

A large scale mutagenesis screen to isolate new viable mutants induced by EMS on second chromosomes marked with cn and bw, was conducted by Charles Zuker and colleagues (E. Koundakjian, R. Hardy, D. Cowen and C. Zuker, personal communication). All the lines generated were tested for male sterility by Barbara Wakimoto and Dan Lindsley, who then conducted a preliminary morphological examination by phase-contrast microscopy to determine the defect in each male sterile line. These lines were re-screened in Margaret Fuller's laboratory, and those with a meiotic arrest phenotype were tested inter se for allelism. comr corresponds to lineZ2-1340, and was represented by a single mutant allele. Mutant phenotypes were scored by squashing cut testes in testis buffer (183 mM KCl,47 mM NaCl, 10 mM Tris pH 6.8) and observing with phase contrast optics. Hoechst 33342 (4 μg/ml) was included in the dissection buffer to visualise chromosomes in live spermatocytes.

Df(2R)Egfr3/Df(2R)X58-7 and Df(2R)XE2900/Df(2R)X58-7males were selected from crosses based on the lack of the balancer chromosome. Testes of some mutant males were dissected to confirm the mutant phenotypes,the remaining mutant flies were used for genomic DNA extraction. Genomic DNA was simultaneously extracted from wild-type male flies, and from flies homozygous for a different mutation generated in the same mutagenesis screen. These DNAs were used as PCR templates to amplify the following candidate genes predicted by the genome sequencing and annotation projects: CG9284, CG4386,CG13492, CG4363, CG4377, CG4372, CG9294, CG13493 and PpN58A(Adams et al., 2000).

PCR primers were designed to amplify all of the predicted ORF from candidate genes, CG4363, CG4377, CG4372, CG9294 and CG13493 as sets of overlapping products. The amplified fragments were sequenced from both ends using BigDye terminator cycle sequencing reagent (ABI), reactions were run on a ABI 377 automated DNA sequencing system. Sequence alignments were carried out using Sequencher 3.1 (GeneCodes Corp).

Total RNA was prepared using TRIzol reagent (Roche) from testes of wild-type male flies, the whole bodies of wild-type males and females respectively, and from wild-type male carcasses (testes were removed by dissection). The total RNA was treated by DNAse I to remove possible contamination of genomic DNA. RT-PCR was carried out using Superscript Preamplificaion System (Invitrogen). A pair of oligo primers(5′-GATTACCAGGGTATGCAGGA and 5′-GGCTTTGCTTTAAACCTGGT) was used to amplify a 979 bp cDNA fragment from the open reading frame of comr. Lack of DNA contamination was confirmed by RT-PCR of the ubiquitously expressed gene PP1-87B using primers that spanned an intron.

The 979 bp comr RT-PCR product was subcloned into pBluescript(Stratagene). pBluescript clones containing cyclin B, cyclin A, Mst87F,boule and polo were prepared as previously described(White-Cooper et al., 1998). Antisense dig-labelled RNA probes were made using the Roche Dig-RNA labelling mix, according to the manufacturer's instructions. Hydrolysis was for 15 minutes per 500 bp. Hybridisation was carried out as previously described(White-Cooper et al.,1998).

Anti-peptide antibodies were raised by Moravian-Biotechnology. The synthesised oligo-peptide, KSRRYDLRNSKRNPR (amino acids 586-600) at the C terminus of the predicted Comr protein, was coupled to BSA and used to immunise a rabbit. The serum was preadsorbed against fixed Drosophilaovaries to deplete non-specific or low affinity antibodies. Rabbit anti-Comr was used at a dilution of 1:5000 for western blotting.

Testes were dissected from young males in testis buffer and transferred to an Eppendorf tube and snap frozen in liquid nitrogen. They were thawed,5×SDS sample buffer added and the samples were boiled for 5 minutes then spun in a microfuge for 5 minutes. Typically 20 testes were loaded on each lane of a 10% SDS-PAGE gel. Proteins were transferred onto a nitrocellulose membrane, which was then blocked in 5% fat-free powdered milk in PBS+0.1%Tween-20. Primary and HRP-conjugated secondary antibodies were diluted in blocking solution and incubated with the blot overnight at 4°C and room temperature for 1 hour, respectively. Signals were detected using Pierce SuperSignal chemi-luminescence reagent and Kodak MBX blue sensitive X-ray film.

For immunohistochemistry, testes were dissected from 0 to 1-day-old males in testis buffer and fixed in 4% paraformaldehyde in HEPES buffer (0.1 M HEPES pH 6.9, 2 mM MgSO₄, 1 mM EGTA) for 30 minutes. The testes were rinsed once then transferred to cell culture inserts (Falcon) in a 24-well tissue culture plate, washed three times for 20 minutes each in PBS+0.1%Triton-X100 (PBSTx), blocked for 30 minutes in PBSTx+5%FCS, then incubated overnight at 4°C or 4 hours at room temperature in primary antibody diluted in this blocking buffer. After four 15 minute washes in PBSTx the testes were incubated 2 hours at room temperature or overnight at 4°C with Universal biotinylated secondary antibody (Vector), preadsorbed againstDrosophila embryos and diluted 1:1000 in blocking buffer. After four 15 minute washes in PBSTx, the testes were incubated for 1 hour at room temperature with ExtrAvidin-HRP (Sigma), diluted 1:1000 in PBSTx. Testes were washed three times for 15 minutes each in PBSTx, rinsed once in PBS then stained with diamino-benzidine (0.5 mg/ml) + 0.001%H₂O₂in PBS. The colour reaction was monitored periodically and stopped by washing with PBS. Testes were mounted in 85% glycerol and observed with an Olympus BX50 microscope equipped with DIC optics. Rabbit anti-Aly was diluted 1:4000 and rabbit anti-Comr was preadsorbed against ovaries and diluted 1:1000.

For immunofluorescence, testes were dissected from 0-1 day old males in testis buffer, about five males were used per slide. The testes were transferred to a 25 μl drop of testis buffer on a poly-l-lysine-treated slide and cut open with tungsten needles. Paraformaldehyde (25 μl of 4%solution) in HEPES buffer was added and the testes were left to fix for 12 minutes at room temperature. Testes were squashed by adding a coverslip, the slide was dunked in liquid nitrogen and the coverslip was removed with a scalpel. Testes were then stored in PBS+0.1% Tween-20 (PBSTw) until all the samples had been prepared. Samples were blocked for 30 minutes with PBSTw+5%FCS, then incubated with primary antibody diluted in blocking solution overnight at 4°C. Testes were rinsed once then washed four times for 10 minutes each in PBSTw. Testes were incubated with secondary antibodies diluted in blocking solution for 2 hours at room temperature. When DNA was to be stained with propidium iodide, RNAse A (0.5 mg/ml) was included with the secondary antibody incubation. Samples were washed as before and mounted in 85% glycerol + 2.5% n-propyl gallate. For DNA staining propidium iodide(1μg/ml) was included in the mounting medium. Coverslips were sealed with nail varnish and the samples were imaged using a BioRad Radiance Plus confocal microscope.

The monoclonal antibody against RNA polymerase II phospho-C-terminal domain(P-CTD) (Clone H5) was obtained from Covance/BAbCO and used at a dilution of 1:100. Pre-adsorbed anti-Comr antibody was used at 1:1000. Secondary antibodies coupled to Cy3 or FITC were used at 1:1000 (Jackson).

A large collection of EMS-induced male sterile mutations (seeWakimoto, 2000) were examined by phase-contrast microscopy of testes squashes to identify potential new meiotic arrest loci. Wild-type testes contain cysts of cells at many different stages of spermatogenesis: pre-meiotic, meiotic and post-meiotic(Fig. 1A). By contrast, testes from one of the Zuker lines (Z2-1340) contained mature pre-meiotic primary spermatocytes, but no cysts undergoing meiotic divisions or attempting post-meiotic spermatid differentiation(Fig. 1B). Z2-1340testes were smaller than wild type because of the lack of later stages of spermatogenesis. The arrested cells degenerated towards the distal end of the mutant testes. Z2-1340 females were fertile. As Z2-1340 maps to the second chromosome, and all of the previously characterised meiotic arrest genes were on the third chromosome, Z2-1340 must represent a previously uncharacterised meiotic arrest locus, which we call cookie monster (comr) because the cells look like a `whole bunch of cookie monster eyes'.

All of the four previously described meiotic arrest loci show defects in transcription in primary spermatocytes of many genes required for spermatid differentiation, including Mst87F and fzo. The meiotic arrest genes were subdivided into aly class and can class because certain cell cycle genes, namely cyclin B, twine andboule were transcribed in can-class mutants (can,mia and sa) but not in aly(White-Cooper et al., 1998). A set of diagnostic RNA in situ hybridisation experiments revealed thatcomr resembled aly rather than the can-class mutantmia (Fig. 2).comr mutant testes transcribed polo, but failed to transcribe both boule and Mst87F.

The phenotype of comr also resembled aly rather than thecan-class genes in terms of the chromosome morphology in mutant primary spermatocytes, after staining with vital Hoechst (33342).can-class mutants had apparently normal chromosome morphology whereasaly had defects in chromatin structure. The major chromosome bivalents of wild-type primary spermatocytes formed three discrete, clearly delineated domains within each nucleus(Fig. 1C). By contrast the chromosome bivalents of comr^Z1340 mutant primary spermatocytes were fuzzy, with indistinct boundaries(Fig. 1D). comr mutant spermatocytes therefore resemble aly rather than can based on their failure to express mRNA for certain cell cycle control genes and their aberrant chromosome morphology. Thus, comr is the second member of the aly-class meiotic arrest genes of Drosophila.

Meiotic recombination mapping localised comr genetically to 96.0±5.7, between c and px on the right arm of chromosome 2. Complementation tests with deficiency chromosomes placed thecomr locus in the physical region 57F-58A(Fig. 3). The left-hand end of the region is defined by the proximal break point of deficiencyDf(2R)X58-7, which is reported as 58A1-2(Kerrebrock et al., 1995). The right-hand end is defined by the distal breakpoint of Df(2R)Egfr3,reported as 57F5-11 (Price et al.,1989). Both these deletions uncovered comr, suggesting that they overlap. Although Df(2R)Egfr3/Df(2R)X58-7 flies were viable and female fertile they were male sterile, indicating that they do indeed overlap. The testes of transheterozygous males had a phenotype indistinguishable in squash preparations fromcomr^Z1340/comr^Z1340 orcomr^Z1340/Df. The boundaries of the overlapping deleted region were identified by PCR of predicted genes from transheterozygousDf(2R)Egfr3/Df(2R)X58-7 males. PCR products were generated for CG4386, CG9284 and PpN58A from this template DNA, indicating that these genomic regions were not disrupted in at least one of the two deletion lines. By contrast, no PCR products were generated for the predicted genes CG13492,CG4363, CG4377, CG4372, CG9294 and CG13493, indicating that these loci are disrupted by both deletions. Clearly none of these six loci is required for viability or female fertility, but at least one is required for male fertility. A third deletion, Df(2R)XE-2900, which deletes from 57F2-58A1 also uncovered comr^Z1340. AdditionallyDf(2R)X58-7/Df (2R)XE-2900 flies were viable, but the males had a phenotype indistinguishable in squash preparations fromcomr^Z1340/comr^Z1340. PCR from these transheterozygotes revealed that CG13493 was deleted inDf(2R)XE-2900, but none of the other candidate genes was deleted,making CG13493 the strongest candidate for the comr gene. The open reading frame of CG13493 was sequenced from wild-type,comr^Z1340 and another mutant derived from the Zuker screen to represent the background chromosome. comr^Z1340contained a single base change (C-T) in CG13493, which was not present in the background chromosome. This mutated codon 12 (CAG) to a stop codon (TAG),presumably resulting in a null allele. No coding region mutations were found in any of the other candidate genes sequenced. Thus CG13493 iscomr.

The comr ORF, based on genome project predictions and RT-PCR (see below) encodes a novel 600 amino acid protein, with a predicted molecular weight of 68.4 kDa, and a predicted pI of 5.1(Fig. 4A). The Comr protein showed no significant homology to any protein in the protein or translated EST databases. The predicted protein contained an acidic domain in the C terminus of the protein (amino acids 518-570), and a predicted nuclear localisation sequence (NLS) (amino acids 583-589). In addition, a region that may represent a very divergent PB1 domain (amino acids 348-431) was identified(Fig. 4B). PB1 domains have been found in several signal transduction proteins, including kinase C iota(KPCI), and have been shown to mediate protein-protein interactions(Ito et al., 2001;Ponting et al., 2002).

comr was represented only once in the Drosophila EST database (of 246620 EST sequences), this cDNA (AT27686) is derived from a testis library and contains a 65 bp 5′UTR. The developmental expression profile of comr was determined by RT-PCR. The transcript was testis specific in adult males, being undetectable in gonadectomised males(Fig. 5A), consistent with the fully viable but male sterile phenotype of the mutant. Low levels of the transcript were also detected in whole females(Fig. 5A), embryos and larvae(data not shown). RNA in situ hybridisation revealed a uniform transcript distribution in embryos (data not shown).

In testes, the comr transcript was expressed at highest levels in early primary spermatocytes, as revealed by RNA in situ hybridisation. The transcript level decreased as the spermatocytes grew, and became undetectable as the primary spermatocytes entered the meiotic divisions(Fig. 5B). This expression pattern of comr is essentially identical to the pattern we have previously reported for aly(White-Cooper et al., 2000). The earliest defect we have detected in comr or aly mutant testes is failure to initiate transcription of target genes in very early primary spermatocytes. The strong expression of comr transcript in early primary spermatocytes is consistent with this phenotype.

An antibody raised against a peptide from the C terminus of Comr protein recognised a 100 kDa protein in wild-type testes that was absent fromcomr^Z1340 mutant testes(Fig. 6A). Although this protein was significantly larger than the predicted size for Comr (68 kDa) the low pI of the protein could affect its mobility in SDS-PAGE. We expressed Comr with an N-terminal Flag tag in mammalian tissue culture cells. Western blot analysis of these transiently transfected cells using anti-flag antibodies and the anti Comr antibody showed that ectopically expressed tagged Comr protein migrated as a single band with an apparent molecular weight of 100 kDa (data not shown). The anti-Comr antibody also crossreacts with a 120 kDa protein present in wild type, Df(2R)Egfr3/Df(2R)X58-7 andcomr^Z1340 mutant testes (upper band inFig. 6A).

In wild-type testes the anti-Comr antibody recognised the nuclei of young and maturing primary spermatocytes (Fig. 6B). Whole-mount immunohistochemistry revealed staining throughout the primary spermatocytes nuclei, as well as a single more concentrated spot of staining within each nucleus. No staining was detected in cells undergoing the meiotic divisions, or at any later stage. Whencomr^Z1340 mutant testes were stained under the same conditions, the antibody still recognised the single spot within each nucleus,but the general nuclear staining was absent(Fig. 6C). Therefore Comr protein appeared to localise throughout the nucleus in primary spermatocytes. The concentrated spot in the nucleus is likely to be due to crossreactivity of the antibody to the 120 kDa protein.

To explore the relationship between Comr localisation and chromatin in primary spermatocytes, we carried out indirect immunofluorescent staining of Comr double labelled with the DNA dye propidium iodide. Staining with the anti-Comr antibody was restricted to the nuclei of primary spermatocytes. Meiotic and post-meiotic cells did not stain with the antibody. Additionally no staining of any somatic cell type (e.g. cyst cells, sheath or accessory gland) was observed, showing that the protein is germ cell specific. Within primary spermatocyte nuclei the anti-Comr antibody showed two distinct types of staining pattern. A brightly stained `stringy' region (arrow inFig. 6E) probably corresponds to the darkly stained spot seen in the immunohistochemistry. This may be the Y-loops, a very decondensed region of the Y chromosome, which is transcribed in primary spermatocytes. Weaker, somewhat spotty, staining was found throughout the nucleus, concentrated near the condensed regions of chromatin(arrowheads in Fig. 6D,E). Again, to determine which component(s) of this pattern were attributable to the crossreacting antigen, we stained comr^Z1340 mutant testes. The stringy staining in a small region of the nucleus was still detected in comr mutant spermatocytes(Fig. 6H, arrow). However the general nuclear staining was absent from these cells(Fig. 6G,H, arrowheads),confirming our findings from the immunohistochemistry. We conclude that Comr protein is expressed only in primary spermatocytes in the testis, that it is nuclear and that is associated with regions of chromatin.

The identical mutant phenotypes of comr and aly suggest that they act in the same pathway; we were interested in examining whethercomr regulated aly or vice versa. To determine howcomr interacts with known meiotic arrest loci, we first investigated the expression of comr and aly in various mutant backgrounds. comr transcript was detected by in situ hybridisation and RT-PCT at a level similar to wild type in aly mutant testes(Fig. 5C, and data not shown). Transcription of comr was also detected in testes ofcan³ (Fig. 5D). Comr protein was detected by western blotting in testis extracts from aly mutant males, as well as in testis extracts from males mutant for the can-class meiotic arrest gene, mia(Fig. 6A). Therefore,transcription and translation of comr is upstream of the action of any known meiotic arrest mutant. The level of Comr protein in alymutant testes was lower than in mia mutant testes, suggesting a potential role for aly in ensuring the accumulation of Comr protein to normal levels. The crossreacting 120 kDa antigen also appeared to be less abundant in all the mutant genotypes compared with wild type. alytranscript and protein levels were similar to wild type incomr^Z1340 mutant testes (data not shown).

The similar localisation pattern of Comr and Aly proteins, combined with their identical mutant phenotypes, suggested that the proteins may interact directly or indirectly. To dissect this relationship, we examined the subcellular localisation pattern of one protein in a background mutant for the other. As both aly and comr are expected to act upstream of the can-class genes, staining of mia mutant testes was used to control for nonspecific effects of the developmental arrest characteristic of all the meiotic arrest mutants. In wild-type testes Aly protein was localised to the nuclei of maturing primary spermatocytes(Fig. 7A). Nuclear staining of primary spermatocytes in cysts gave a spotty appearance to the apical region of the testis. By contrast, Aly protein showed a honeycomb distribution pattern in comr^Z1340 mutant spermatocytes(Fig. 7B). The holes in this pattern correspond to the nuclei; therefore, Aly protein failed to translocate to the nucleus and instead remained cytoplasmic incomr^Z1340 mutant spermatocytes. This indicates that the nuclear localisation of Aly protein is dependent on comrfunction.

The reciprocal experiment revealed that Comr protein localisation depended on normal function of aly. We carried out immunofluorescence with anti-Comr antibody on aly⁵ and mia mutant testes. The anti-Comr staining of aly⁵ andcomr^Z1340 mutant primary spermatocytes was indistinguishable. A single stained spot, corresponding to the 120 kDa antigen was present in the mutant cells (Fig. 7D). mia mutant spermatocytes showed staining throughout the nucleus, in addition to the bright spot(Fig. 7G). By immunohistochemistry, the anti-Comr antibody staining ofaly⁵ and comr^Z1340 mutant spermatocytes were indistinguishable (data not shown), with only one small region of the nucleus stained. mia and can³testes probed with the anti Comr antibody had staining throughout the nuclei of primary spermatocytes, in addition to the single strongly stained spot(data not shown). In both immunostaining techniques, the staining intensity of the spot in the nucleus was less than in wild type, consistent with the reduction in level of the 120 kDa antigen seen in the western blotting. Thus,the nuclear localisation of Comr depended on aly, but did not depend on the normal function of the can-class meiotic arrest genes can andmia.

comr mutant spermatocytes are defective for transcription of a number of cell cycle and spermatid differentiation genes(Fig. 2). To investigate how Comr protein localisation in the nucleus is associated with mRNA transcription, we used an antibody specific to active RNA polymerase II phospho-C-terminal domain (P-CTD) in triple labelling experiments.Fig. 8 shows the relationship between Comr, DNA and active transcription in primary spermatocytes. Staining with anti P-CTD revealed that the regions of the nucleus with most active transcription were adjacent to, but not overlapping with, regions of visible(i.e. condensed) DNA (arrow in Fig. 8D). The active transcription was found in domains within the nucleus, not randomly distributed. Lower levels of detectable P-CTD colocalised with the more condensed DNA (arrowhead inFig. 8D). The highest levels of active transcription partially overlapped with high levels of Comr protein(arrows in Fig. 8F), although some regions of strong P-CTD staining were associated with weaker Comr staining. All of the regions where transcriptional activity was detected had at least some Comr protein present (arrow inFig. 8F), although P-CTD staining was not found in all regions containing Comr protein. Thus, Comr protein was not exclusively localised on the pol II-transcriptionally active chromatin, but all the chromatin where RNA polymerase II is transcriptionally active had at least some associated Comr protein.

In spermatogenesis, a major transition occurs as the mitotically amplifying population of spermatogonia cease mitosis and develop into primary spermatocytes. These primary spermatocytes become committed to undergoing the meiotic divisions, and then differentiating into spermatozoa. This change in cell behaviour is associated with a dramatic switch in the transcript profile:some genes are downregulated and many are upregulated or switched on for the first time. The meiotic arrest genes of Drosophila are crucial for regulating transcription in primary spermatocytes. Numerous genes require the activity of the meiotic arrest genes for their transcription. Earlier analysis showed that aly is different from the other meiotic arrest genescan, mia and sa, in that it is required for the transcriptional activation of more target genes, and for normal chromosome structure. Based on this, we proposed that aly acts upstream of thecan-class genes in primary spermatocytes. We have shown thatcomr is a novel meiotic arrest gene, that comr mutant spermatocytes fail to transcribe twine and mst87F as well as many other target genes, and that the cells arrest with abnormal chromatin morphology. This demonstration that the comr mutant phenotype is indistinguishable from that of aly supports the segregation of the meiotic arrest genes into aly and can classes. The chromosome morphology defect seen in both aly and comrmutant lines suggests that the pathway in which they act has a role in the maintenance of normal chromatin structure.

Comr and Aly are mutually dependent: Aly protein requires comr for its normal nuclear localisation, and Comr protein requires aly for its normal nuclear localisation. Thus, the phenotypes of the two mutants are identical, because ablation of one essentially knocks out the other by preventing its normal localisation. It seems likely that Aly and Comr proteins physically interact or are both essential components of a multi-subunit complex. Anchoring of either protein in the nucleus would then be dependent on the formation of this complex, loss of either would give the aberrant protein localisation patterns seen. To date we have been unable to detect any direct physical interaction between Aly and Comr proteins (J. J. and H. W.-C.,unpublished), so we would favour a hypothesis where Aly and Comr are in a complex with at least one other protein. Aly nuclear localisation is regulated; this regulation need not be solely dependent on Comr. For example,cytoplasmic Aly could be phosphorylated, and dephosphorylation could allow nuclear entry and therefore interaction with Comr. This protein complex would then be able to attach to chromatin.

aly, comr and can transcripts are all transcribed in a characteristic pattern in wild-type testes(Hiller et al., 2001;White-Cooper et al., 2000). High levels of transcripts are found in early primary spermatocytes, and the levels decrease as the spermatocytes mature. None of the transcripts are detectable as the cells enter the meiotic divisions. This pattern is entirely consistent with the role of these genes in activation of transcription of many new genes in early primary spermatocytes, and the continued high transcriptional activity throughout primary spermatocytes development.

It is interesting to note that both comr and alytranscripts appear to be expressed at low levels in tissues other than the testis. We were able to detect some comr transcript by RT-PCR in females and in all stages of embryos examined (0-1, 3-5 and 12-24 hours).aly mRNA was also detected at low levels in the older embryo samples(J. J. and H. W.-C., unpublished). However null mutants of both comrand aly are fully viable and female fertile, showing that the expression of these genes outside the testis is not essential. It is possible that comr and aly have a redundant function in these other tissues. A second Drosophila homologue of aly has been identified (White-Cooper et al.,2000), but there is no obvious comr homologue in the genome. It is attractive to postulate that the meiotic arrest genes are all under the control of the same transcription factor, so that they are all activated early in the primary spermatocyte programme of development.

The predicted comr protein is novel. While its predicted size is 68 kDa, Comr protein (from testes or expressed in mammalian tissue culture cells) migrates at about 100 kDa in SDS-PAGE. This aberrant mobility on SDS-PAGE gels may be due to the acidity of the protein retarding its migration. The low predicted pI of the protein may provide some clues as to its biochemical function. The protein is rather acidic throughout its length,as well as having a very acidic region near the C terminus. In this regard, it bears some similarity to the acidic histone chaperone protein nucleoplasmin,which is important for nucleosome assembly and remodelling during transcription (Chen et al.,1994; Earnshaw et al.,1980). It is possible that the acidic domain on Comr interacts with the basic histone proteins to alter chromatin structure.

aly encodes a homologue of the C. elegans SynMuvB genelin-9 (Beitel et al.,2000; White-Cooper et al.,2000). The SynMuvA and B genes act in two genetically redundant pathways to repress vulval cell fate and promote hypodermal cell fate in the vulval precursor cells (Chen and Han,2001; Fay and Han,2000; Ferguson and Horvitz,1989). The SynMuvB genes include subunits of the NURD histone deacetylase/nucleosome remodelling complex, and probably regulate genes involved in vulval formation by altering chromatin structure(Lu and Horvitz, 1998;Solari and Ahringer, 2000). By analogy, aly may activate such a NURD complex in primary spermatocytes. comr could act with aly as a regulator of the complex. Alternatively comr and maybe also aly could function as testis specific components of the NURD complex. In this model Comr(and Aly) proteins would be directly involved in nucleosome remodelling, Comr protein perhaps interacting with histones as postulated above.

The predicted Comr protein did not contain any sequence motifs with known DNA-binding activity. Nevertheless, Comr protein is found in cells in close association with the chromatin, suggesting that the chromatin localisation of Comr may be mediated by protein-protein, rather than protein-DNA,interactions. A candidate region for mediating such protein-protein interactions is the PB-1-like motif. This region of Comr is not similar enough to the PB-1 consensus to score a significant match, therefore it is unlikely that the domain is a true PB-1 domain, interacting with the PC motif. However this PB-1 like region of Comr could be responsible for mediating protein-protein interactions by binding to a motif similar to PC.

It has recently been shown that the meiotic arrest gene canencodes a testis-specific homologue of dTAF80, a subunit of the basal transcription factor TF_IID(Hiller et al., 2001). TF_IID consists of TATA-binding protein and associated factors,binds to the promoter region, interacts with transcriptional activator proteins and helps in recruitment of the RNA polymerase II holoenzyme complex to the transcription initiation site(Dynlacht et al., 1991;Goodrich et al., 1996;Roeder, 1996). The human and yeast homologues of can are also found in the histone acetyl-transferase (HAT) complex PCAF or SAGA(Grant et al., 1998;Ogryzko et al., 1998). Ascan is not required for all transcriptional activation in primary spermatocytes, it may be that spermatocytes have two TF_IID complexes, each with a different set of target genes. aly andcomr could function by altering the chromatin structure at the site of target promoters so that the testis specific, Can-containing,TF_IID complex can bind and activate transcription. This would make all the can-dependent transcripts also dependent on comr andaly, consistent with our observations. However this simple model cannot explain why some genes, namely cyclinB, twine andboule, require aly and comr but not thecan-class genes for their expression. Thesealy-class-dependent genes are all required for normal meiotic cell cycle progression in testes; however, they are not transcribed exclusively in primary spermatocytes. cyclinB is required for mitosis, so is expressed throughout development (Lehner and O'Farrell, 1990), twine is required for meiosis in the female germline so is also expressed in ovaries(Alphey et al., 1992), andboule transcripts have been found in cDNA libraries derived from heads (H. W.-C., unpublished) (Rubin et al., 2000). Their transcription in primary spermatocytes may depend on particular chromatin structure to facilitate binding of a spermatocyte specific transcription factor, which would act in conjunction with the conventional TF_IID complex. This, or a related, postulated specific transcription factor could also be required for transcription ofcan-class dependent genes by interacting with the testis specific TF_IID complex. Fig. 9 shows a model for how target promoters could be regulated, first by chromatin remodelling promoted by Aly and Comr, then by transcription factor binding and recruitment of TF_IID complexes.

When assessed using immunostaining, both Comr and Aly proteins persist until the G2-M transition of meiosis I, but become undetectable as the chromosomes condense in prometaphase I. At this point, transcriptional activity is shut down. The cause and effect relationship between transcription shut down and Aly/Comr disappearance events not clear. Perhaps transcription shuts down because Comr and Aly are degraded in response to the same cues that signal chromosome condensation. Alternatively, the proteins could become physically excluded from the DNA during chromosome condensation, and then degraded.

Transcription from a particular promoter can be viewed as a cycle of polymerase binding, initiation of transcription, promoter clearance and termination. During this cycle, the phosphorylation state of the CTD of RNA polymerase II changes. Unphosphorylated pol II is competent to enter the pre-initiation complex, as promoter clearing occurs the pol II becomes multiply phosphorylated on the CTD (Conaway et al., 2000; Dahmus,1996). At termination of transcription, RNA polymerase II is dephosphorylated and released, ready for another cycle. Hence, antibodies specific to phosphorylated CTD only label the pool of RNA polymerase II that is actively transcribing. The antibody used in this study (H5) specifically recognises phosphorylation of Ser2 in the heptapeptide CTD repeat(Patturajan et al., 1998).

We have demonstrated that the most active regions of mRNA transcription are not associated with strong DNA staining. Lower levels of transcriptional activity were found to be associated with this DNA. This is expected, as we would predict that active transcription occurs on the most decondensed, and therefore weakly staining, regions of DNA. The staining clearly shows that there are discrete domains within the nucleus with high levels of transcriptional activity, rather than the active regions of chromatin being randomly distributed within the nucleoplasm. Comr protein colocalised with DNA staining, and therefore it is not surprising that we found Comr protein associated with all regions of transcriptional activity. There was no clear correlation between the intensity of Comr staining and the presence or absence of P-CTD immunoreactivity. Within each nucleus some regions of high mRNA transcription were associated with high Comr, other regions were associated with low Comr levels. Not all Comr-positive regions of chromatin were transcriptionally active. Given this distribution how does Comr control the transcription in primary spermatocytes of some genes but not others? Comr protein seems to be present on more regions of chromatin than those that are actively transcribing, so it is unlikely that binding of Comr to chromatin regions is solely responsible for marking them for transcriptional activation. The Comr-containing complex must interact with another factor to give the promoter specificity we observe. For many promoters, that other factor may be the Can-containing TF_IID complex; the other factor for can-independent promoters remains unidentified. Because neither of thealy-class genes described to date encode direct DNA-binding proteins,it is likely that more genes of this class, which encode DNA-binding transcription factors, remain to be identified. These would be likely to contribute also to the promoter specificity of the meiotic arrest transcriptional activation pathway through sequence-specific DNA binding.

We thank the Wellcome Trust, the MRC and the Royal Society for funding this research; and Margaret Fuller, Luke Alphey and Myles Axton for helpful discussions and for critical reading of the manuscript. We also thank Margaret Fuller and Mark Hiller for sharing their unpublished data. We are indebted to Charles Zuker, Barbara Wakimoto, Dan Lindsley and Ed Koundakjian for their generation and sharing of mutant lines, includingcomr^Z1340. Deficiency chromosomes were kindly supplied by the Bloomington Drosophila Stock Centre. Comments from Amy MacQueen about the phenotype lead to naming of the gene.