Regulated chromatin domain comprising cluster of co-expressed genes in Drosophila melanogaster.

Recently, the phenomenon of clustering of co-expressed genes on chromosomes was discovered in eukaryotes. To explore the hypothesis that genes within clusters occupy shared chromatin domains, we performed a detailed analysis of transcription pattern and chromatin structure of a cluster of co-expressed genes. We found that five non-homologous genes (Crtp, Yu, CK2betates, Pros28.1B and CG13581) are expressed exclusively in Drosophila melanogaster male germ-line and form a non-interrupted cluster in the 15 kb region of chromosome 2. The cluster is surrounded by genes with broader transcription patterns. Analysis of DNase I sensitivity revealed 'open' chromatin conformation in the cluster and adjacent regions in the male germ-line cells, where all studied genes are transcribed. In contrast, in somatic tissues where the cluster genes are silent, the domain of repressed chromatin encompassed four out of five cluster genes and an adjacent non-cluster gene CG13589 that is also silent in analyzed somatic tissues. The fifth cluster gene (CG13581) appears to be excluded from the chromatin domain occupied by the other four genes. Our results suggest that extensive clustering of co-expressed genes in eukaryotic genomes does in general reflect the domain organization of chromatin, although domain borders may not exactly correspond to the margins of gene clusters.

INTRODUCTION

It has been suggested that eukaryotic genome is functionally divided into domains that contain all regulatory elements necessary for full, correct and positionally independent expression of resident gene(s) (1,2). These domains are evident at the level of local chromatin structure, where transcriptionally active genes are associated with ‘open’ and acetylated chromatin, whereas silent genes are embedded in compacted and hypermethylated chromatin [reviewed in (3–5)]. The assumption that not only single genes such as chicken glyceraldehyde-3-phosphate dehydrogenase (6), human apoB (7), and chicken lysozyme (8,9) genes, but also multigenic loci may be regulated at the level of common chromatin domains, originated from the data obtained on several intensively studied gene clusters. The examples include, but are not limited to, the cluster of β-globin genes in chicken, mouse and human (10–17); growth hormone genes (18,19), IL-4/IL-13 genes (20) and PRM1-PRM2-TNP genes in human (21); ovalbumin genes in chicken (22); and heat-shock genes in Drosophila (23). All these clusters comprise paralogous genes that most likely have originated from ancestral genes through local duplications. It is not clear whether the inferences from the studies on these models can be extrapolated to the organization of entire genome, where clusters of paralogous genes occur rather rarely.

Recently, a number of reports have demonstrated non-random clustering of co-expressed genes on chromosomes. First observations of this phenomenon, that we are aware of, date back to 1991 (24), but in 2002 an influx of publications based on the analysis of whole-genome transcription data from different organisms indicated that mechanisms of transcriptional co-regulation, that operate with chromatin domains, are common from yeast to higher eukaryotes [reviewed in (25)]. Moreover, according to the data of Spellman and Rubin (26) over 20% of all Drosophila genes are clustered on chromosomes according to their expression patterns and thus may share common chromatin domains. However, direct evidence that would link the observed gene clusters to the chromatin domains was still missing. To address this issue, we thoroughly characterized the cluster of non-homologous testes-specific genes, and analyzed the chromatin structure in the region. The cluster of five testes-specific genes in the cytological region 60D1-2 of Drosophila melanogaster chromosome 2 includes new genes Crtp, Yu and CG13581, along with previously identified Pros28.1B (27) and CK2βtes (aka Ssl) (28,29). Analysis of chromatin sensitivity to DNase I revealed that four out of five genes in the cluster occupy shared chromatin domain which appears to be in the ‘closed’ configuration in somatic tissues (embryos and larval brain) and in ‘open’ configuration in larval testes. This finding of the regulated chromatin domain that encompasses cluster of non-homologous co-expressed genes implies that the ‘domain’ hypothesis may be applicable to a large portion of the eukaryotic genome.

MATERIALS AND METHODS

Isolation and sequencing of cDNA clones

The λZapII cDNA library (Stratagene) custom made from testes of D.melanogaster was provided by Dr Tulle Hazelrigg. About 6 × 105 phage plaques were screened on the nitrocellulose lifts with the 32P-labeled probes indicated in Figure 1, yielding numerous phage corresponding to the genes Crtp, Yu and CG13581. After in vivo excision from λZapII, both strands of cDNA inserts were sequenced using the Sequenase 2.0 kit (United States Biochemicals). Corresponding genomic regions were subcloned from the cosmid clone #9 (29) into pBlueScriptII SK− vector and also sequenced. The following fragments of the cosmid clone #9 were used as the probes for screening the library: probe a, a mixture of 880 bp BamHI–BamHI and 1461 bp BamHI–HindIII fragments; probe b, 1113 bp AvrII–PstI fragment; probe c, 3103 bp NsiI–NsiI fragment; and probe d, 308 bp fragment amplified by PCR using the primers CTCGAATTCGGACCCAGCACTTTTGCATTCCCG and CTCAAGCTTTGACTCGCGGTGGAACCACCCATA.

Figure 1

Structure of the 60D1-2 region including the cluster of five testes-specific genes, and surrounding genes with broader expression pattern. Exon–intron structure and the location of genes from the region is according to the GadFly (release 3.1) with addition of the Yu transcription unit, according to our data. Testis-specific genes are in black. Gene CG13590 (dashed) is predicted in Drosophila genome annotation, but its existence is not supported by our data.The horizontal bars (labeled a through d) indicate the probes used for testes cDNA library screening and for in situ hybridizations.

Northern analysis

For developmental northern analysis, 30 μg of total RNA isolated by TRIzol (Invitrogen) extraction from adult or larval testes, ovaries, embryos, larvae, pupae, gonadectomized adult males or females, and from cell culture, were fractionated by electrophoresis in denaturing formaldehyde-agarose gel and transferred by blotting onto the HyBond-N membrane (Amersham). For northern analysis of mutants, total RNA was isolated from testes of the bam (30) and aly (31) homozygous adult males. Total RNA isolated from the testes of the Df(1)w , y strain with normally proceeding spermatogenesis was used as a positive control. Hybridizations and washes were performed according to standard protocols (32). 32P-labeled antisense riboprobes were synthesized with the T7 RNA polymerase and [α-32P]UTP (3000 Ci/mmol) on the linearized plasmid templates, using the pBlueScriptII SK− T7 promoter. For the templates generated by PCR, the T7 promoter sequence was embedded in one of the PCR primers. Plasmid templates were as follows: full-size Crtp cDNA #321 (29) lacking poly(A) tail; Yu-specific probe, 1113 bp AvrII–PstI fragment of cosmid #9 (29) cloned into the XbaI–PstI digested pBlueScriptII SK−; CG13581-specific probe, 308 bp PCR fragment cloned in the EcoRI–HindIII digested pBlueScriptII SK− (primers used: CTCGAATTCGGACCCAGCACTTTTGCATTCCCG and CTCAAGCTTTGACTCGCGGTGGAACCACCCATA); and probe c, 3103 bp NsiI–NsiI fragment of cosmid #9 cloned in both orientations into the PstI site of pBlueScriptII SK−. PCR-generated templates included 184 bp CK2βtes-specific probe, (primers used: CGATAGCCAGACACTGGAGGTAGTCT and GAATTAATACGACTCACTATAGGGAGAGTCCCCCCTTTCGTACTTTAATCG); 364 bp Pros28.1B-specific probe (primers used: GGCCACCCTGATCCATTCTG and GAATTAATACGACTCACTATAGGGAGACCAATCAAGGGAAACCATCTCG); and 738 bp anon60Da-specific probe, (primers used: CCACCATCCGGCGAATGAAG and GAATTAATACGACTCACTATAGGGAGACCTTCGCTTTATCGGGCACA). In all cases, DNA of cosmid #9 was used as a PCR template. As a control for RNA loading, the same filters were hybridized with the probe specific to constitutively expressed gene Rp49 (33).

RNA in situ hybridization

For RNA in situ hybridization the same antisense probes were used as for the northern analysis. Antisense RNA digoxigenin-labeled probes were synthesized using T7 RNA polymerase and templates as described above, and hybridized with testes according to conventional protocols (34) with some modifications. Testes were manually dissected, fixed for 1 h on ice in phosphate-buffered saline (PBS) containing 4% paraformaldehyde, and treated with Proteinase K (50 μg/ml) for 8 min. Prehybridization was performed at 60°C in the HS buffer (50% formamide, 5× SSC, 0.1% Tween-20, 100 μg/ml salmon sperm DNA and 50 μg/ml heparin). Hybridization was performed at 60°C overnight in the HS buffer, and was followed by washes at 60°C: HS buffer for 1.5 h; 2× SSC, 0.1% Tween-20 for 30 min; and 0.2× SSC, 0.1% Tween-20 for 30 min. Blocking was performed in PBS containing 0.1% Tween-20 and 0.3% Triton X-100. Incubation with anti-DIG-alkaline-phosphatase-conjugated antibodies (Roche Diagnostics) was performed for 1 h in the same solution, followed by mounting in glycerol/PBS (9:1). Samples were observed using the Leica MZ9-5 microscope.

RT–PCR analysis

Total RNA was extracted from manually dissected adult testes, ovaries and heads, from larval salivary glands and brains, and from 2 to 10 h embryos of the laboratory strain Df(1)w , y of D.melanogaster using the TRIzol reagent (Invitrogen), and purified on the RNeasy columns (Qiagen) according to the manufacturer's protocols. Reverse transcription of 1 μg samples was performed with the OmniScript enzyme (Qiagen) in the presence of oligo(T) primer, reaction products were treated with RNAse H, and diluted 10-fold with water. Mock-reactions run in the absence of reverse transcriptase were used as negative controls. An aliquot of 2 μl of diluted samples served as templates for 25 μl real-time PCRs, which were performed in the ABI 5700 Sequence Detector using the SYBR Green chemistry (Applied Biosystems). The constitutive transcript of the Actin5C gene was used as a cDNA template loading reference. Primer sequences are presented in Table 1.

Table 1

Primers used for real-time RT–PCR and for chromatin analysis

Amplicon	Upper primer	Lower primer
A8	AATGCTCGGCGGAACTCGGTAATG	CGGCGACTGCTGAAGTTGTAAGGA
A10	TCAATCGCCGCAAACATTCCCACA	CACAAAGCGGTCACACAATAATCC
A12	ACAGCAGACCACCTCCTTCAGCAA	ATCCTTGATCCCAGTGCCCTCTGT
A15	GCATATCATGTGCTGCTGTGGAAC	TTCGACGCAATGGTAGTTATTACA
A16	AAACTGACGGTGTACTGGTAAATAA	CTGCCATCTGGGGACTATCTACTAA
A17	CGGCTCCACGAACGTGACATTTAT	TTCCACTTCCGGCTATTCGATTTA
A17.5	ACCCTTGCCGAATCAACAACATTAC	CCGCTCGTCCGCCAACTTTGTGTCC
A18	GGATCGGGCTCAGTACAAGTGTCT	GCTTCCGCTCCAATTTCTGATACG
A19	TTTGAGCAGGCGGTCGAGATCATGG	GGCGACCGTAGGAGTATCCGTATAG
A21	CTTTCTGTATTTCGATGGGGTCTCT	CCGCGTCCTGTTACTGTTCCTGTTC
A24	GTGTAGGTGGAGCAGGAAATCGCAA	GTAATCAGCAGGAAAAATGTGTAAA
A25	CCTCCCATTGCTCTGCGTGTAGTT	GCCACCCGCACTTGCCTATTAGTT
A27	AACCCCCGTCGCCGTCCAAACTAA	TTCGAGGTCTGCCAACATTTTCAC
A30	CTTCTCCAGCTTCTTCTAATCTATC	TGTACTTGTCGTCTGGGCTGCGTTC
A32	TTCCCGTCCTGCCCATCCAGCATA	TCTTTGCCTGCCCGCTGACCCATA
A36	CACTCGCCAAAGGAGCTTAGAGCA	GTCACAGATGAACCCGCCAAAGAA
A37	GCAGGTGCCTCTTTGTCTGGTCTT	GTGTGTGTTAGCGGGCATTTTGTT
Actin5C	AGTTGCTGCTCTGGTTGTCG	CGTAGGACTTCTCCAACGAGG
Crtp	CACCAGGCGCTCATTTACAG	TGGGACAGGAGGCGACACAG
CK2βtes	GTCGTGTCCCAGGAGCATCG	CAGCGTGGACCGAATGGTGT
Pros28.1B	TCCGAAGGGGTTCCACTGTG	GTGATTTGCCCGCTTTACAC
CG13581	GCCGGTCTATCCCTTTTCCA	TCGTCTGGGCTGCGTTCGTA
CG13589	AAATCGTGGGTGGTTGTTCACTACT	CTTATACCCATTGGCCTTCTTCATC
CG13579	TTGAGGGCCTCTCCGAGAAGTACTG	CTGCTGCGGCTGGTGGATCTG
CG13582	ATTGAATCGTGTTGGGAGCAGAAGT	GAAAAGCCCGAAAGCTGTAGACGAA

Chromatin sensitivity to DNase I

The assay was performed essentially as described by Kramer et al. (35), with some modifications. Dechorionized 2–4 h embryos and manually dissected larval testes and larval brains were homogenized in the freezing storage buffer (50 mM Tris–HCl, 5 mM MgCl2, 10 mM NaCl and 25% glycerol, pH 7.5) and frozen at −80°C. After thawing, homogenized tissues were permeabilized in 0.05% NP-40 for 5 min on ice, then pelleted by centrifugation at 1500 g for 8 min at 4°C. Pellets were resuspended in the DNase I-buffer (40 mM Tris–HCl, 0.4 mM EDTA, 10 mM MgCl2, 10 mM CaCl2 and 0.1 mg/ml bovine serum albumin) and treated with 0.5 U/ml DNase I (Promega). Two digestion timepoints (10 and 15 min at room temperature) were analyzed and similar results were observed. Reactions were stopped by adding EDTA to 20 mM, genomic DNAs were purified on DNeasy columns (Qiagen) according to the manufacturer's protocol and eluted in 200 μl. An aliquot of 2 μl of the eluates served as templates for 25 μl real-time PCRs, which were performed as described above. Each reaction was repeated three to six times, and each DNase I-digestion was repeated twice (for embryos and brains) or four times (for testes). Overall, each data point presented in Figure 3 is an average of 6–12 repeated measurements. Ct values for each reaction are presented in Supplementary Table S1. Data were analyzed using the Excel 2001 (Misrosoft) and SPSS 10.1 (SPSS) software packages. Primer sequences are presented in Table 1.

Figure 3

Profiles of chromatin DNase I-resistance across the 60D cluster region and surrounding sequences. Chromatin resistance to DNase I [as normalized relative yield (NRY) for each amplicon] is plotted on the vertical axis; the length of 60D1-2 genomic region (in kb) is plotted on the horizontal. Positions of the genes in the region are shown at bottom. Circles (black for the regulated chromatin domain, and white for the rest of the region) indicate average NRY for each amplicon, the grey area corresponds to the calculated 95% confidence interval. Upper panel: in larval testes, the entire region shows nearly uniformal low resistance to DNase I typical for the ‘open’ chromatin. In contrast, in larval brains (middle panel) and in embryos (lower panel) regulated chromatin domain that contains the genes CG13589 through Pros28.1B shows significantly higher resistance to DNase I, indicative of ‘closed’ chromatin configuration.

Analysis of the chromatin assay data

To quantify chromatin sensitivity to DNase I-digestion, we first measured the relative quantitative PCR (QPCR) yield for each amplicon as the ratio of yields observed with DNase I-treated sample (‘output’) versus untreated sample (‘input’). Results were adjusted to account for the differences in amplicon length, as described in (35). The regions flanking the 60D cluster genes show high accessibility to DNase I in all tissues (low relative QPCR yields), similar to the constitutively expressed (36) gene Actin5C (Table 3). We therefore assumed that these highly accessible regions possess the ‘open’ chromatin structure characteristic of actively transcribed genes, and chose the level of resistance to DNase I in these regions as a reference. For each tissue, the average relative QPCR yield observed for the amplicons A8 through A12 (corresponding to CG13579) and A37 (corresponding to CG4589) was calculated, and the relative QPCR yields observed for individual amplicons were normalized against this value to produce the normalized relative yields (NRYs) shown in Figure 3. As a result, ‘open’ chromatin regions show NRYs close to 1.0, while ‘closed’ chromatin regions, more resistant to DNase I, show significantly higher NRY values.

Table 3

Results of the chromatin assay

Amplicon name	Position in scaffold (bp)a	Larval testes				Larval brains				Embryos
		Average NRYb	Nc	SD	95% CId	Average NRYb	Nc	SD	95% CId	Average NRYb	Nc	SD	95% CId
A8	108405.5	0.54	12	0.14	0.08	0.77	6	0.07	0.05	1.01	12	0.28	0.16
A10	110492.5	0.62	12	0.25	0.14	0.69	6	0.09	0.07	0.76	12	0.11	0.06
A12	112175	1.24	12	0.20	0.11	1.16	6	0.21	0.17	1.29	12	0.23	0.13
A15	114444	0.83	12	0.26	0.14	1.46	6	0.25	0.20	1.67	12	0.20	0.11
A16	115868.5	0.98	12	0.89	0.50	3.22	6	1.02	0.82	3.32	6	0.85	0.68
A17	116704	1.42	12	1.11	0.63	3.82	6	0.89	0.71	2.08	12	0.62	0.35
A17.5	117459.5	1.32	12	0.37	0.21	2.22	6	0.56	0.45	2.02	6	0.34	0.28
A18	118236.5	1.26	12	0.60	0.34	2.15	6	0.71	0.57	2.01	12	0.38	0.22
A19	118939	1.34	12	0.39	0.22	2.69	6	0.64	0.51	2.16	6	0.42	0.33
A21	121367	0.95	12	0.33	0.19	3.09	6	0.94	0.75	3.04	6	0.58	0.46
A24	123583	0.75	12	0.32	0.18	2.62	6	0.31	0.25	2.67	6	0.75	0.60
A25	124541.5	0.84	12	0.36	0.21	1.42	6	0.19	0.15	1.49	6	0.36	0.29
A27	126990	0.76	12	0.29	0.16	0.99	6	0.10	0.08	0.96	6	0.10	0.08
A30	130001.5	1.24	12	0.60	0.34	2.26	6	1.30	1.04	1.99	6	0.97	0.77
A32	131511	1.57	12	0.66	0.37	2.15	6	0.62	0.49	1.43	6	0.29	0.23
A36	135594	1.52	12	0.75	0.42	1.88	6	0.15	0.12	1.67	6	0.20	0.16
A37	136941.5	1.57	12	0.47	0.27	1.37	6	0.40	0.32	0.84	6	0.25	0.20
Actin5C		0.75	6	0.25	0.20	0.62	6	0.22	0.18	0.58	6	0.15	0.12

aPositions of the center points of amplicons in the AE003464 scaffold are shown.

bAverage normalized relative yield (NRY) is proportional to efficiency of amplification from the DNase I-digested chromatin template.

cNumber of real-time PCRs used to produce the average NRY.

dCalculated 95% confidence interval.

The plot data points (NRY versus genomic position) for each tissue were fitted into a panel of simple curve models using the SPSS 10.1 software package. The subset of data points corresponding to the regulated chromatin domain (amplicons A15 through A25) was analyzed against the rest of the data (treated as the second subset) to evaluate the significance of differences in NRY between the two subsets, also using SPSS 10.1.

HSS mapping using indirect end-labeling

Approximately 0.6 g of 0–18 h embryos were homogenized in 0.3 M sucrose-A1 buffer [60 mM KCl, 15 mM NaCl, 15 mM Tris–HCl (pH 7.5), 5 mM MgCl2, 0.1 mM EGTA, 0.5 mM DTT, 0.5% Triton X-100 and Complete protease inhibitor cocktail (Roche Diagnostics GmbH)], filtered through two layers of Miracloth (Calbiochem) on the step gradient of 0.3 M sucrose-A1/0.8 M sucrose-A1. Nuclei were pelleted by centrifugation at 3300 g for 5 min at 4°C and resuspended in 0.4 ml of DNase I-digestion buffer [60 mM KCl, 15 mM NaCl, 15 mM Tris–HCl (pH 7.5), 3 mM MgCl2, 0.1 mM CaCl2, 0.5 mM DTT and 0.3 M sucrose]. Four 0.1 ml aliquots were digested by 0, 0.05, 0.15 and 0.45 U of DNase I (RQ1, Promega) for 5 min at 37°C. Digestion was stopped by addition of 2 μl 0.5 M EDTA and 10 μl 30% Sarcosyl. Proteinase K (40 μg) was then added and samples were incubated at 50°C overnight. DNA was phenol and chloroform extracted and then ethanol precipitated. Isolated DNA was digested by HindIII and BamHI, separated in 0.75% agarose gel overnight, blotted onto the HyBond-XL membrane (Amersham) and hybridized with 32P-labeled 839 bp ClaI–HindIII fragment adjoining the promoter region of CG13581 gene. Hybridization and washes were performed according to standard protocols (32). Filter was exposed with the film for 2 weeks.

RESULTS

Our initial observation was that two testes-specific genes, CK2βtes (28,29) and Pros28.1B (27), are located <1 kb apart from each other in the chromosomal cytological region 60D1-2. Both genes have constitutively expressed paralogs in the Drosophila genome, namely the CK2β (29) and Pros28.1 (27) genes. We hypothesized that duplicated copies of CK2β and Pros28.1 may have integrated into a testes-specific chromatin domain, and this event in part shaped the path for evolution of these copies into testes-specific CK2βtes and Pros28.1B. This suggestion, which also implies that other testes-specific genes pre-existed in the region before integration of duplicated copies, prompted us to examine the transcription patterns and chromatin structure in the vicinity of the CK2βtes and Pros28.1B.

Three more testes-expressed genes are located in the CK2βtes/Pros28.1B region

To search for another transcription units in the genomic regions adjacent to CK2βtes and Pros28.1B genes, the hybridization probes were prepared from the cosmid clone #9 (29) covering several kilobases upstream and downstream from these genes (Figure 1) and used for screening of the testes cDNA library. While each of the probes a, b and d revealed numerous positive plaques, probe c gave no hybridization signals when 6 × 105 phage plaques were screened. Strand-specific RNA probes for the fragment c were hybridized with northern blots and gave no detectable signal in embryos, larvae, pupae, adult flies, testes, ovaries and heads (data not shown). We conclude that region corresponding to the probe c is not transcribed in Drosophila and may harbor regulatory functions, e.g. carrying a boundary element (see below).

Partial sequencing of five cDNAs detected by the probe a showed that they were identical in the nucleotide sequence, differing only by the 5′- and 3′-truncation points. Sequencing of the longest cDNA and of corresponding genomic region (DDBJ/EMBL/GenBank accession no. AF197938) from cosmid clone #9 revealed 2235 bp open reading frame (ORF) interrupted by 52 bp intron. Encoded 86 kDa polypeptide shows weak (20% amino acid identity) similarity with the human smooth muscle caldesmon across the whole length, hence it was named Crtp (caldesmon-related testes protein). However, despite the name it is not likely that Crtp is related to caldesmon in function, because the level of identity between the proteins in the actin-, myosin-, calmodulin- and tropomyosin-binding domains essential for caldesmon does not exceed the average 20%.

Sequencing of two cDNA clones detected by the probe d (DDBJ/EMBL/GenBank accession no. AY078077) revealed ORF region corresponding to the gene CG13581 predicted in the annotated Drosophila genome database. The encoded 23 kDa Arg/Ser-rich polypeptide has no significant similarities in protein databases, and lacks any known protein domain signatures.

Three poly(A)-containing cDNAs detected by the probe b and corresponding genomic region were also sequenced (DDBJ/EMBL/GenBank accession no. AY078076). These cDNAs represent an intronless transcription unit Yu that lacks obvious ORFs beginning with AUG, but may encode several peptides of <100 amino acid residues starting with the unconventional GUG or CUG initiator codons. Alternatively, Yu transcript may be a precursor of miRNA, a representative of the small ∼22 nt RNA molecules that are involved in the regulation of expression of protein-coding genes [reviewed in (37)]. Further experiments are necessary to examine which of these possibilities is realized in vivo.

Thus, three testes-expressed genes are located in close vicinity to CK2βtes and Pros28.1B. We further asked whether these genes share additional aspects of transcription with previously characterized CK2βtes and Pros28.1B, such as testes specificity and expression in primary spermatocytes.

Crtp, Yu and CG13581 are testes specific

Northern analysis demonstrates the presence of Crtp, Yu, CK2βtes, Pros28.1B and CG13581 transcripts exclusively in adult or larval testes (Figure 2A). No signals were detected in the RNA samples isolated from embryos, whole larvae, pupae, gonadectomized males and females, ovaries, heads and cell culture. For each gene tested, the size of the transcripts matched the size of the corresponding cDNA. The observed lack of the hybridization signal in whole larvae or pupae is probably due to excessive dilution of testes transcripts by the RNA from other body parts, since Drosophila male larvae and pupae do contain developing testes where most of the male germ-line-specific genes are already transcribed [reviewed in (38)]. Testes specificity of transcription of Crtp, CK2βtes, Pros28.1B and CG13581 was confirmed by real-time RT–PCR analysis (Table 2).

Figure 2

Transcriptional analysis of genes from the 60D1-2 region. (A) Developmental northern analysis shows testis specificity of expression of genes from the 60D cluster. Total RNA sources are indicated on top; culture corresponds to the Schneider-2 cells; m.carc. and f.carc. represent gonadectomized males and females, respectively; ad.testes and lar.testes represent adult and larvae testes, respectively. 32P-labeled antisense RNA probes were used. Hybridization with the antisense RNA probe for the constitutively expressed gene Rp49 served as the loading control. RNA in situ hybridization of DIG-labeled CG13581 (B) or Crtp (C) probes with whole-mount adult testes show onset of transcription in spermatocytes; transcripts persist until late post-meiotic stages of spermatogenesis. Arrow marks tip of testis occupied by stem cells and spermatogonia, where no hybridization was detected. (D) Northern analysis of the transcription of genes from the 60D cluster in the bam and aly mutants shows universal requirement for the bam function, but diverse response to the aly deficiency. Total RNA was isolated from testes of males with the genotype indicated on top; the Df(1)w (2), y males with normally proceeding spermatogenesis were used as positive control.

Table 2

Gene expression profiles assayed by the real-time RT–PCR analysis

Gene name	Adult headsa	Adult testes	Adult ovariesa	Embryosa	Larval brainsa	Larval salivary glandsa
Crtp	0.16	1.00	0.01	0.09
CK2βtes	0.09	1.00	0.04	0.01
Pros28.1B	0.06	1.00	0.00	0.00
CG13581	0.05	1.00	0.00	0.00
CG13589	0.26	1.00	0.51	0.00	0.00	2.31
CG13579	5.01	1.00	0.95	1.25
CG13582		1.00		0.00	0.68	192.67

aAmounts of transcripts in tissues are shown, relative to the amount of transcripts in adult testes.

We also determined transcription patterns of other genes in the region, including CG4589 (anon60Da) and CG13590 (for gene locations see Figure 1). We detected anon60Da transcripts by northern analysis at all stages of Drosophila development, although they were noticeably more abundant in testes (Figure 2A). No transcripts were detected for the predicted gene CG13590 at any stage of development by northern analysis or by real-time RT–PCR (data not shown). The gene CG13589, located next to CG13590, appears to be transcribed in heads, testes, ovaries and salivary glands according to our real-time RT–PCR data, although no significant transcription was observed in embryos and larval brains (Table 2). The gene CG13579, located adjacent to CG13589, is transcribed at similar levels in heads, testes, ovaries and embryos as determined by real-time RT–PCR (Table 2). We also analyzed transcription of CG13582 which is located next to CG4589, and found it to be expressed at similar levels in testes and larval brains and up-regulated in salivary glands, but silent in embryos (Table 2). Therefore, the five genes (Crtp, Yu, CK2βtes, Pros28.1B and CG13581) form cluster of testes-specific genes (hereafter called the 60D cluster) flanked by more broadly expressed CG13589 and CG13579 on one side, and CG4589 (anon60Da) and CG13582 on the other side.

Simultaneous onset of transcription of all genes in the 60D cluster during spermatogenesis

All stages of male germ line cell differentiation are represented in a single adult testis of Drosophila [reviewed in (38)]. The primary spermatogonial cells, born at the apical tip of testis from the stem cells, undergo four rounds of mitotic divisions to produce cysts of 16 interconnected spermatogonia. Maturing cysts of spermatogonia are gradually displaced from the tip as they soon enter the meiotic program and become primary spermatocytes. After extended G2 phase associated with robust gene expression, primary spermatocyte undergo meiotic divisions, producing haploid spermatids. Since spermatocyte cysts continue to be displaced away from the apical tip as they grow, the cells enter meiotic division near the start of the coiled part of the testis.

Figure 2B and C show the results of RNA in situ hybridization of DIG-labeled CG13581 and Crtp probes to whole-mount adult testes. The pattern of hybridization with other genes comprising the 60D cluster was in general the same (data not shown). No labeling was detected at the apical tip occupied by stem cells and spermatogonia. Transcription of each gene starts in spermatocytes and transcripts disappear at the late post-meiotic stages of spermatogenesis. This transcription pattern resembles that of other male germ-line-specific genes, such as the β2-tubulin gene (39), the don juan gene (40) and the Sdic gene (41).

Transcription of genes comprising the 60D cluster is diversely affected in the bam and aly mutants

We analyzed the transcription of genes comprising the 60D cluster against the background of mutations bag-of-marbles (bam) and always early (aly), which affect spermatogenesis in Drosophila. Testes of the bam homozygous males are filled with amplified mitotically dividing spermatogonia, with no spermatocytes present (30). Figure 2D demonstrates the absence of transcripts of all testes-specific genes from the 60D cluster in bam mutants. This is consistent with our RNA in situ hybridization data indicating that transcription of these genes is initiated in spermatocytes but not in spermatogonia (Figure 2B and C).

The meiotic arrest gene aly controls transcription of numerous genes in testes (31). It encodes a chromatin-associated protein suspected of having a function in chromatin remodeling (42). The transcription of the four genes from the 60D cluster (Crtp, Yu, Pros28.1B and CG13581) appears to be aly-dependent, as no hybridization signals were detected in the RNA isolated from testes of homozygous aly mutant males (Figure 2D). Surprisingly, transcripts of the fifth gene of the cluster (CK2βtes) are even accumulated in the aly mutant testes. Since CK2βtes is located in the same chromatin domain as the Crtp, Yu and Pros28.1B genes (see below), it seems unlikely that the aly protein is involved in ‘potentiation’ (35) of chromatin domains during spermatogenesis.

Regulated chromatin domain in the 60D cluster region

To study the organization of chromatin in the 60D cluster region, we examined the accessibility of chromatin to the DNase I using a modified PCR-based nuclease resistance assay (35). The segment of genome covered by analysis included the 60D cluster region and surrounding sequences. Seventeen ∼0.5 kb PCR amplicons (named according to their positions on the x-axis of Figure 3) were designed to provide a dense coverage across this 29 kb region of interest. Larval testes that were used for the assay contain mostly spermatocytes (where the 60D cluster genes are expressed), and also some spermatogonia and somatic cells [reviewed in (38)]. Larval brains and embryos were chosen as examples of somatic tissue where the 60D cluster genes are silent. Cells from larval testes, larval brain and embryos were permeabilized with detergent. Part of the cell suspension was set aside, and later used to estimate the amount of ‘input’ DNA. The rest of the suspension was incubated with DNase I. Then, both undigested and partially digested DNA samples were isolated and the quantity of DNA that escaped cleavage by DNase I was estimated by real-time PCR. Since cutting the DNA renders it an ineffective template for PCR, the less accessible is the chromatin region to DNase I, the higher is the yield of PCR product in the DNase I-treated sample (‘output’). The PCR yield was measured relative to the sample not treated with DNase I (‘input’), and normalized against the values observed for constitutively transcribed genes CG13579 and CG4589 flanking the region. The resulting NRYs were plotted along the region. The data (Figure 3) shows variability of chromatin structure of the genomic segment including amplicons A15 through A25 in different tissues.

As shown in the Figure 3, upper panel (see also Table 3), in larval testes the chromatin exhibits nearly uniform high sensitivity to DNase I across the entire region studied. The distribution fits well to a linear model (P = 0.019) with the minor slope of 0.022 NRY U/kb and intercept at 0.62 NRY U, i.e. the line very similar to a constant of 1.0. This indicates that in larval testes, the chromatin in the entire region of the 60D cluster and neighboring genes is in the ‘open’ configuration, similar to the chromatin over Actin5C gene.

In contrast, much greater variation in chromatin sensitivity to DNase I across the region was observed in embryos (Table 3) (Figure 3, lower panel). An ∼10 kb segment containing testes-specific genes Crtp, Yu, CK2βtes and Pros28.1B from the 60D cluster, as well as the non-cluster gene CG13589, (amplicons A15 through A25, marked black in Figure 3) showed higher resistance to DNase I, while the rest of the region still demonstrated NRY values not so different from the 1.0 reading representative for the ‘open’ chromatin. The average NRY for the DNase I-resistant segment is 2.28, this is about two times higher than the average NRY of 1.25 for the rest of the region. The high significance of this difference is supported by non-parametric tests (the Mann-Whitney U-test and Kolmogorov-Smirnov two-sample test showed P values of 0.002 and 0.012, respectively). For comparison, in larval testes the average NRY for the amplicons A15 through A25 is 1.08, which is not different from the average NRY of 1.13 observed for the rest of the region (0.95-fold difference, P values for the two aforementioned tests are 0.85 and 0.59). The distribution of chromatin resistance to DNase I in embryos is no longer supported by the linear model (P = 0.78), the best fit was observed for the quadratic (P = 0.031) and cubic (P = 0.039) models which can accommodate the ‘hill’ of resistance to DNase I in the middle of the region.

Analysis of chromatin sensitivity to DNase I in larval brains (Table 3) and (Figure 3, middle panel) revealed the pattern almost identical to that observed in embryos. DNase I-resistant 10 kb segment containing testes-specific genes Crtp, Yu, CK2βtes and Pros28.1B from the 60D cluster, as well as the non-cluster gene CG13589, (amplicons A15 through A25) showed average NRY of 2.52, highly significantly different from the average NRY of 1.41 observed for the rest of the region (1.79-fold difference, P values for the the Mann-Whitney U-test and Kolmogorov-Smirnov two-sample test are 0.007 and 0.054, respectively). The distribution of chromatin resistance to DNase I in brains is also not supported by the linear model (P = 0.77), the best fit was observed for the quadratic (P = 0.06), cubic (P = 0.04) and S-shaped curve (P = 0.05) models.

Altogether, the data are indicative of a regulated chromatin domain that includes testes-specific genes Crtp, Yu, CK2βtes and Pros28.1B from the 60D cluster, as well as the non-cluster gene CG13589.

DISCUSSION

Eukaryotic genome has been intensively studied in efforts to unravel the relationship between its structure and function. One attractive hypothesis is that chromosomes are organized in independent chromatin domains that contain individual genes or gene clusters, and these domains play the regulatory role, being either permissive or repressive towards gene expression (1,2). The ‘domain’ hypothesis is supported mainly by the data from thoroughly investigated models such as the β-globin (10–17), growth hormone (18,19), interleukin (20) and some other gene clusters. These clusters are composed of paralogous genes, which are probably the products of local gene duplications. Because of the implied mechanism of formation of these clusters, occurrence of the related genes in a common chromatin domain may merely be an automatic consequence of gene duplications within the pre-existing domain. Whether or not the resulting structure is employed in co-regulation of the paralogous genes in the cluster, it is not clear that similar mechanism(s) operates to regulate gene expression elsewhere in genome.

Recently, we (43) and others [reviewed in (25)] have revealed the abundant presence of clusters of co-expressed genes in the genomes of multicellular eukaryotes, leading to a tempting suggestion that these clusters outline chromatin domains. For example, according to our data (43), about one-third of the tissue-specific genes form clusters in Drosophila genome. Spellman and Rubin (26) have shown that more than 20% of Drosophila genes (both constitutively and tissue-specifically expressed) are clustered according to their expression patterns. Here, we present the evidence for the regulated chromatin domain that encompasses cluster of non-homologous co-expressed genes. Considering that majority of the clusters of co-expressed genes consist of non-homologous genes (26,43–45), this finding implies that the ‘domain’ hypothesis may be applicable to the substantial portion of eukaryotic genome.

We have analyzed in detail the 15 kb genomic region of Drosophila genome where five non-homologous testes-specific genes are localized next to each other, thus forming a non-interrupted cluster of co-expressed genes (the 60D cluster). All genes in the 60D cluster are activated simultaneously in spermatocytes, as shown by RNA in situ hybridizations and by northern analysis of bam mutants that are arrested at the spermatogonial stage. On each side, the cluster is flanked by genes with broader tissue specificity of expression.

Our analysis of chromatin structure indicates that four testes-specific genes in the cluster share a regulated chromatin domain. While sensitivity to DNase I in larval testes was high across the entire region that included the 60D cluster and adjacent sequences, a segment showing high resistance to DNase I was observed in larval brains and in embryos. Sensitivity to DNase I is a conventional gauge for the compactness of chromatin. The actively transcribed genes are associated with the DNase I-sensitive, unfolded chromatin, while the ‘closed’ chromatin structure, typical for the repressed genes, is more resistant to DNase I (6–8,11,12,14,16,21–23). Therefore, our interpretation of the observed patterns of sensitivity to DNase I is that the chromatin of the 60D cluster and surrounding genes is ‘open’ in testes (where all these genes are expressed), but a segment encompassing most of the 60D cluster is ‘closed’ in somatic tissues represented by embryos and brain. This segment apparently represents a regulated domain of repressed chromatin, and it includes the genes CG13589, Crtp, Yu, CK2βtes and Pros28.1B. All these genes, except for CG13589, are transcribed in the male germ line only. Chromatin structure of the domain becomes ‘open’, i.e. ‘potentiated’ (35), before or at the primary spermatocyte stage, at which transcription of the cluster genes is initiated.

It is quite intriguing that while the regulated chromatin domain does contain most of the testes-specific genes from the 60D cluster, the borders of the domain and of the cluster are not identical. On one side, the gene CG13589, which is included in the regulated domain, is not a member of the 60D cluster of testes-specific genes. Our RT–PCR data show that CG13589 is transcribed at low level in testes, ovaries and heads, but not in embryos and larval brains. Thus, some aspects of tissue specificity of CG13589 appear to be similar to that of the genes in the 60D cluster (namely, expression in testes and absence of expression in embryos and in larval brains). The shared chromatin domain may, at least partially, underlie this similarity. This indicates that genes included in common chromatin domain ought to show significantly overlapping, but not necessarily identical transcription patterns.

On the other side of the 60D cluster we observed that a continuous stretch of co-expressed genes on the chromosome may occupy more than one chromatin domain. The border of regulated domain is clearly positioned to the left of CG13581, thus leaving CG13581 outside of the domain shared by other testes-specific genes of the cluster. ‘Closing’ of chromatin over CG13581 was observed in somatic tissues, indicating the possible presence of a second domain of repressed chromatin in the region, however, the effect was not very significant. Whether or not the chromatin domain containing CG13581 is a regulated one, it is distinct from the domain that contains other genes of the 60D cluster, being separated by a constitutively DNase I-sensitive region that contains amplicon A27 (Figure 3). It is important to mention here that the constitutive DNase I sensitivity of the amplicon A27 is not due to the presence of DNase I-hypersensitive site (HSS). We have studied this by Southern analysis of the 5 kb fragment between the genes Pros28.1B and CG13581 (encompassing amplicon A27) in embryos, using indirect end-labeling. We found nine ‘weak’ sites with increased sensitivity to DNase I, however, none of these mapped within A27 (Supplementary Figure S1).

This ‘border’ region may represent a fixed domain boundary established, e.g. by insulator sequence [reviewed in (46–48)], although other types of boundary elements were described (49,50). Taken together, our studies indicate that while clusters of co-expressed genes do outline chromatin domains, the exact domain pattern can not be deduced solely from the expression data.

Shared chromatin domains provide the tool for co-regulation of genes with similar expression patterns. For clusters of paralogous genes, possible mechanism of creation of multigenic domains is the internal ‘stretching’ due to local gene duplications within domains. In case of clusters of non-homologous co-expressed genes, shared domains could be created by integration of genes into pre-existing domains. This could partially canalize the evolution of newly integrated genes by pre-setting general aspects of their transcription patterns.

Finding of the regulated chromatin domain implies that there are factors that control its repressive state in somatic tissues, and potentiation in male germ line as it was shown, e.g. for the cluster of neuron-specific genes (51). One potential candidate for such role would be the gene aly, which is required for expression of wide spectrum of genes during male gametogenesis, and has been implicated in regulation of chromatin structure (42). We analyzed the effect of aly deficiency on expression of testes-specific genes in the 60D cluster, and found that one gene (CK2βtes) does not require the aly function for its transcription. CK2βtes is located in the middle of regulated chromatin domain, being surrounded by other testes-specific genes that depend on aly heavily. Failure to ‘open’ the regulated domain due to the aly mutation would have similar effect on all genes included in the domain. Therefore, the Aly protein is probably involved in more downstream steps of transcriptional regulation in testes.

The results presented here suggest that extensive clustering of co-expressed genes observed recently in eukaryotic genomes does in general reflect the domain organization of chromatin, although exact domain borders may not correspond to the margins of gene clusters. ‘Opening’ or ‘closing’ of chromatin domains would outline tissue specificity of expression of genes included in these domains. The chromatin organization then provides a tool for coordinated regulation of multiple genes. Pattern of ‘open’ and ‘closed’ chromatin domains along the chromosomes, if maintained through cell divisions, can shape the transcriptome of the cell lineage. Alternatively, modifications of this pattern during development may lead to the coordinated changes in the expression of gene batteries, thus establishing new cell lineages.

SUPPLEMENTARY MATERIAL

Supplementary Material is available at NAR Online.