Endogenous pararetroviral sequences in tomato (Solanum lycopersicum) and related species

LycEPRV identification, isolation and sequence analysis

Tomato EPRVs were originally detected by DNA blot analysis using a 5.5 kb DNA fragment of NsEPRV (Nicotiana sylvestris EPRV), one of three EPRV families in tobacco [1, 2], to probe DNA prepared from various species of Solanum. The resulting banding pattern was complex, with numerous strong and weak bands superimposed on a background smear (Fig.1). This pattern is reminiscent of that observed with Nicotiana species [1] and suggests a dispersed organization of multiple copies of a related EPRV family. Judging from the hybridization intensity, the relative copy number of the elements detected by the NsEPRV probe was similar in all five Solanum species tested. The banding pattern in S.lycopersicum strongly resembled that in S.cheesmaniae and S.pimpinellifolium, whereas notable differences were observed in S.habrochaites and S.peruvianum (Fig.1).

To analyze the tomato EPRV sequences in more detail, a genomic λ-library was constructed from cultivated tomato (S.lycopersicum "MicroTom"; [29]) and the wild relative S.habrochaites. Both λ-libraries were screened with the 5.5 kb fragment of NsEPRV. Five positive clones were isolated and partly sequenced for S.lycopersicum and nine for S.habrochaites. Each clone contained EPRV-like DNA and flanking plant genomic sequences (Fig.2A, Table 1).

EPRV-like sequences from both species were AT-rich (65.4–78.4%) and were most similar to EPRVs in Nicotiana, revealing up to 83% sequence identity to endogenous Tobacco vein clearing virus (TVCV; [2]), NsEPRV [1], and NtoEPRV (N. tomentosiformis EPRV; the second EPRV family in tobacco; [3]). Similar to the Nicotiana EPRVs, four open reading frames (ORFs) were identified (Fig.2A): coat protein (CP), cell-to-cell movement protein (MP), polyprotein (POL) and transactivator protein (TAV). The POL domain revealed 80 to 90% identical nucleotides, compared to MP (75 to 91%) and TAV (63 to 95%). Only one clone contained a full CP sequence that showed 65 to 94% sequence identity to fragments of CP sequences from other clones. The identity between DNA sequences derived from the same species (S.lycopersicum or S.habrochaites) was generally not higher than between species. Thus, in the subset of clones analyzed, no species-specific clusters of identity were identified and sequences within one species are as divergent as between species. We therefore assigned these sequences to a single family termed LycEPRV (Lyc opersicon endogenous pararetrovirus).

The putative amino acid sequence identities of the coding regions ranged from 60 to 87% identity for MP, 72 to 89% for POL and 48 to 91% for TAV (CP shares 39 to 85% identity to various fragments). However, all of the cloned protein-coding regions are either truncated or harbour several frameshifts and stop codons and can therefore be considered translationally defective, a feature also found with Nicotiana EPRVs. Nine of the clones contained parts of the putative non-coding intergenic region (IGR) of the virus. The IGR was less conserved compared to the ORFs except for a 272 to 282 bp box (Fig.2B) which revealed up to 86 to 92% sequence identity on the nucleotide level. The conserved 272 to 282 bp box has an overall identity of up to 70% with its counterpart in S.tuberosum, SoTu [4] and 80% to the IGR of Nicotiana EPRVs with several highly conserved motives. Some IGR sequences contained short (27 to 104 bp) AT-rich structures of low complexity (Lh2, Lh5, Le4, Le5) while others revealed short (12 to 24 bp) direct repeats which were not conserved between the different IGRs (Lh2, Lh5, Lh7, Le5). Some clones (Lh7, Lh2, Lh3) contain a conserved 12 bp motif complementary to the 3'end of the tRNA^Met (5'-TGGTATCAGAT/GC-3') 50 to 60 bp upstream of this box as well as a putative polyadenylation signal (5'-AATAAA-3') and a putative TATA box (5'-TATAAA-3') at a distance of 130 to 140 bp and 150 to 160 bp upstream, respectively.

All of the cloned LycEPRV sequences were truncated and flanked either by plant DNA unrelated to EPRVs or by rearranged (fragmented, inverted or otherwise partly duplicated) EPRV regions that appeared to be out of context when compared to the TVCV-like consensus structure (Le4, Le5, Lh3, Lh7; Fig.2A). Nearly all LycEPRV junctions analysed adjoin transposable elements, most frequently retrotransposon LTRs or related sequences (see Table 1 and Fig.2A). Clones from S.habrochaites revealed homologies to members of the PCRT1 family, a Ty3-gypsy (Metaviridae) element that is dispersed throughout the pericentric heterochromatin of S.lycopersicum (AY850394; [30]). The LTRs of PCRT1 partly correspond to the repetitive families TGRII and U30, the latter of which comprises more than 4000 copies in the S.lycopersicum genome [30, 31]. The junctions between EPRV and PCRT1 sequences were verified for three clones by PCR amplification from genomic DNA (Lh2, Lh4 and Lh7, data not shown), confirming that the LycEPRV sequences are indeed physically joined to plant DNA while these sequences could not be amplyfied in S.lycopersicum.

We reconstructed a general structure from the alignments of several incomplete sequences (upper bar in Fig.2A). The coding region closely resembles that of the tobacco elements (NsEPRV, NtoEPRV) in size with 1779 bp for CP, 1293 bp for MP, 1933 bp for POL which overlaps with TAV (1279 bp) forming a coding region of 6221 bp. The intergenic region varies between 1606 to 1680 bp for different clones, summing up to a total length of approx. 7900 bp (7827 to 7901 bp) for a putative full copy of LycEPRV. The 140 kb sequence of a BAC clone (AC171732) that was submitted only recently (November 2006, note added in revision) revealed a single LycEPRV copy. A single stretch of 6125 bp of this sequence corresponds to the putative LycEPRV coding region and reveals the same order of the four ORFs as reconstructed from the λ-clones. The coding region is flanked by altogether 1542 bp homologous to the IGR on both sides and reveals only one internal stop codon. The nucleotide sequence of this copy contains 84–96% identical nucleotides compared with the λ-clones and 76–92% homology to TVCV. Approximately 2.7 kb upstream of this LycEPRV copy sequences homologous to the LTR of PCRT1a could be identified.

Fluorescent in situhybridization (FISH)

To analyze the chromosomal distribution of LycEPRVs, we performed FISH on root tip metaphase chromosomes and pollen mother cells at meiotic prophase of S.lycopersicum and S.habrochaites. By mixing several probes covering most of the LycEPRV (LycEPRV-Sl; Table 2), we were able to observe several weak LycEPRV-Sl hybridization sites with signal strength of several magnitudes lower than that observed with the control 45S rDNA probe. Sites were visible in varying number near the centromeres of most S.lycopersicum chromosomes (Fig.3A, B): there were four to six chromosomes with a stronger signal, four chromosomes showing very weak signals (arrows) and no signal in the NOR region. Similar results were obtained with extended pachytene chromosomes demonstrating that the EPRV signals were located mainly in the DAPI positive pericentromeric heterochromatin or intercalary chromocentres (Fig.3D, E arrowheads), but rarly in the euchromatin. The weak, but in cases distinct signals of varying size and arrangements indicate that probably only few copies of LycEPRV-Sl are integrated in each cluster, that they might not contain all parts of the probe used or that sequences are only partly conserved. The FISH data (Fig.3A–C) support the results from Southern hybridization (Fig.1) and cloning as well as sequencing data derived from λ-clones (Fig.2) and the BAC clone AC171732 indicating that LycEPRV-Sl are probably not arranged in perfect tandem arrays, are truncated and frequently degenerated.

Pooled probe	Derived from region	Derived from λ-clone	Length
LycEPRV-Sl	CP/MP	Le4 (position: 2882 – 4182)	1300 bp
	MP	Le3 (position: 2365 – 3185)	820 bp
	TAV	Le2 (position: 1260 – 2365)	1100 bp
	IGR (box)	Le5 (position: 2203 – 3207)	1000 bp
LycEPRV-Sh	CP	Lh7 (position: 1576 – 2455)	880 bp
	MP	Lh6 (position: 1186 – 1862)	680 bp
	TAV	Lh4 (position: 3617 – 4706)	1090 bp
	IGR (box)	Lh7 (position: 7133 – 7716)	600 bp

FISH of LycEPRV-Sl in combination with the retroelement sequence U30 on metaphases (Fig.3C) and pachytene chromosomes (see Additional file 1) showed signal from both sequences near the centromeres. The signal of the U30 probe covered a larger area of the centromeric heterochromatin while the LycEPRV-Sl hybridization signal appeared to be nested within the U30 hybridizing regions. The U30 signal, as the LycEPRV-Sl signal, was absent from the NOR regions (Fig.3C) as has been previously reported [32]. FISH of LycEPRV-Sh (Table 2) on metaphase chromosomes of S.habrochaites showed similar, but not identical hybridization patterns to LycEPRV-Sl on S.lycopersicum in the pericentromeric region of most chromosomes (Fig.3G–I). However, the signal strength seemed to be more variable between chromosomes (Fig.3I); again, there was no hybridization detected to the NOR region (Fig.3G).

DNA methylation analysis

Cytosine methylation of LycERPVs in S.lycopersicum and S.habrochaites was investigated using methylation-sensitive restriction enzymes and DNA blot analysis. Previous work on EPRVs in Nicotiana has shown that the isoschizomer pair HpaII/MspI (recognition sequence CCGG), which is normally used to study CG methylation in animals, is not informative because of frequent CHG methylation in plants, that inhibits both HpaII and MspI, in these sequences [15]. We therefore focused on enzymes sensitive to CHG and CHH methylation: ScrFI-BstNI (C^mCNGG or CCWGG, respectively) reports on CHG methylation while Sau3AI-NdeI (GAT^mC) reports on methylation in potentially non-symmetrical cytosines, depending on the sequence context. The first enzyme in each isoschizomer pair is methylation-sensitive. Following a predigestion with XbaI, an additional digest was performed with either the methylation-sensitive or -insensitive enzyme from a particular isochizomer pair. Southern blots of electrophoretically separated DNA were hybridized to two different probes each (Fig.4). One was the 1.3 kb fragment (probe E1) of the CP/MP reading frame of a cloned S.lycopersicum EPRV copy (Le1), the other one was derived from a S.habrochaites clone (Lh7) and comprises 580 bp of the IGR including most of the 273 bp box (probe H7).

For both species, the methylation-sensitive ScrFI cleaved little beyond the XbaI predigest whereas methylation-insensitive BstNI digested substantially more, indicating the presence of CHG methylation of LycEPRV sequences (Fig. 4A, B). Little difference between coding regions and IGRs was observed. Hybridization of both the Sau3AI-and NdeI digested DNA with the CP/MP probe (E1) revealed substantial cleavage compared to the XbaI predigestion, suggesting little asymmetrical CHH methylation within the coding EPRV sequences (Fig.4C). Reprobing of the same blot with the IGR probe (H7) revealed a similar pattern, although smaller bands in the NdeI digests were more emphasized (Fig.4D). This suggests that asymmetrical methylation of the intergenic region is low but slightly stronger than in coding regions. The sequence of the cloned LycEPRV sequences did not reveal striking differences in the relative number of CHG and CHH residues between IGR and coding regions.

Expression analysis

Even though the LycEPRVs sequenced are defective and unable to encode intact viral proteins, one or more full-length copies could exist and potentially be pathogenic if activated under stress conditions. To test this possibility, we made inter-specific crosses with the aim of provoking a genome stress and then examined the hybrids for symptoms of virus infection. Four different interspecific crosses were made between different wild species (S.pimpinellifolium, S.habrochaites, S.cheesmaniae and, S.peruvianum) and S.lycopersicum ("MicroTom"). The phenotype of 7–27 individuals per cross resembled the phenotype of the wild parent rather than the dwarf cultivar of S.lycopersicum ("MicroTom"). Their hybrid nature was confirmed by SSR marker analysis (LE 20592; [33]) to exclude selfed offspring.

No typical symptoms of virus-induced diseases could be detected at any time during the development of the hybrids that were grown in a greenhouse for a full year and trimmed frequently. In addition, hybridization of undigested, genomic DNA of selected individuals to probe E1 and H7 (coding region and IGR, respectively) failed to demonstrate episomal virus DNA since all individuals lacked the expected three bands for the linear, circular or supercoiled episomal DNA species (Fig.4, first lane each).

The cytosine methylation of the interspecific hybrids was analysed in comparison to parental genomes of each cross. In all cases the methylation pattern of the hybrid individuals resembled that of their parents: CHG and CHH methylation in the LycEPRV coding regions as well as in the IGRs could be observed (Fig.4). The unchanged methylation pattern and the absence of any virus-induced disease symptoms in the interspecific hybrids suggest that active virus was not produced by endogenous virus sequences under the conditions tested.

Interestingly, despite the inability to induce active virus in hybrids and the presence of cytosine methylation LycEPRVs appeared to be transcribed to some extent in healthy plants. The NCBI EST sequence databases contain transcripts from S.lycopersicum, S.habrochaites and S.pennellii with high similarity to our sequenced LycEPRVs from S.lycopersicum and S.habrochaites. More than 30 EST homologies were distributed over all four EPRV ORFs and the intergenic region. The cDNAs were derived from different tissues including flowers, red or green fruits, seeds, trichomes and shoot meristems as well as from suspension culture, callus tissue or crown galls (Fig.5A, Table 3). This suggests widespread transcription of sequences closely related to LycEPRVs in healthy tomato plants and related wild Solanum species not only under stress but also under normal growing conditions.

To further study the transcriptional activity of LycEPRVs in S.lycopersicum, S.habrochaites and an inter-specific hybrid, RT-PCR was performed using the conserved primer pairs CP/MP and TAV1/TAV2 amplifying parts of the coding region and IGR1/IGR2 for the conserved box within the intergenic region (Fig.5A, B). Fragments of the expected size were amplified in all individuals (Fig.5B) and DNA sequence analysis revealed high sequence similarities to the respective LycEPRV regions. Twenty-one cDNA sequences and six genomic sequences of the TAV region comprising 761 to 806 bp each were aligned. Many turned out to be identical or nearly identical (> 98% sequence identity) on the nucleotide level whereas others diverged up to 30 to 37% (63 to 70% identity, Fig.5C). Taking into account the error-prone activity of reverse transcriptase, highly similar or identical transcripts appear to be derived from identical or corresponding EPRV copies present in both species. Nevertheless the transcripts are generally derived from more than one copy in each genome since diverging sequences are falling into at least five different clusters in S.lycopersicum, into four in S.habrochaites and six in the hybrid. None of the cloned genomic fragments of the corresponding region was matched with 100% sequence identity (97 to 99%). Many (62%) of the cDNA sequences are translationally defective, i.e. contain frameshifts and stop codons in their putative amino acid sequence. Similarly nine cDNA sequences and one genomic fragment of the IGR were analysed, which revealed higher homogeneity, but still fall into more than one cluster (Fig.5D).

Short RNA analysis

Given the absence of viral disease symptoms in plants constitutively expressing LycEPRV transcripts, we tested whether homologous short RNAs – which might be indicative of RNA-mediated silencing – were present in healthy plants. Northern blots containing short RNA fractions from leaf material of S.lycopersicum, S.habrochaites and an interspecies hybrid as well as flowers of S.lycopersicum were hybridized to RNA probes derived from the LycEPRV intergenic region and the TAV region, respectively. For the IGR probe a cDNA sequence homologous to the conserved 272 bp box served as a template. A mix of three different clones was chosen for TAV since this region is more heterogeneous. Signals could be detected in the two parental species and the hybrid with both probes and in both sense and antisense orientations. A distinct band of ~21 nucleotides in length and several bands ranging from 22–25 nucleotides in length were detected in all samples analysed. Generally, the flower-derived fraction produced the strongest signals (Fig.6).

To assess whether the short RNAs were derived from an RNAi/Post-Translational Gene Silencing (PTGS) pathway, and hence might contribute to viral defense, we analyzed short RNAs in plants infected with heterologous RNA viruses, exploiting their ability to counteract RNA silencing by encoded proteins that suppress PTGS [17, 34, 35]. Potato virus Y (PVY, Potyvirus) expresses HCPro, which is known to prevent the maintenance of RNA silencing and binds to siRNAs preventing the formation of the siRNA-initiated RISC assembly [39, 40]. Tomato bushy stunt virus (TBSV, Tombusvirus) encodes p19, which forms homodimers and prevents the strand separation of 20–22 nt siRNA duplexes. This is a prerequisite for their integration into the RNA induced silencing complex (RISC; [[36, 37], rev. in [38]]. Plants infected with either PVY or TBSV revealed increased amounts of the 21–22nt LycEPRV short RNA fraction compared to mock infected individuals and plants harvested before starting the infection procedure (Fig.7). The accumulation of the smaller sized short RNAs homologous to both the intergenic region of LycEPRVs (IGR) and part of the coding region (TAV) could be observed in the cultivars "MicroTom" as well as in "Moneymaker". The phenomenon is consistent with a formation of the LycEPRV short RNAs in the RNAi/PTGS pathway.

Plant material and DNA isolation

Seeds of Solanum lycopersicum L. (syn. Lycopersicon esculentum Mill.) "MicroTom" were provided by Dr. A.A. Levy, Rehovot, Israel. S.pimpinellifolium L. (syn. Lycopersicon pimpinellifolium (L.) Mill.) IPK genebank accession LYC 1835, S.cheesmaniae (L.Riley) Fosberg (syn. Lycopersicon cheesmaniae L.Riley) IPK genebank accession T 675, S.peruvianum L. (syn. Lycopersicon peruvianum (L.) Mill.) IPK genebank accession T 353 and S.habrochaites S. Knapp & D.M. Spooner (syn. Lycopersicon hirsutum Dunal) IPK genebank accession T 436 were procured from the „Institut für Pflanzengenetik und Kulturpflanzenforschung“ (IPK) in Gatersleben, Germany. S.lycopersicum ("Moneymaker") lines were obtained from D. Scharf, Frankfurt University. Plants were grown in the greenhouse. Genomic DNA was isolated from leaves with the DNeasy Plant Maxi kit (Qiagen) following the manufacturer's instructions.

λ-library and sequencing

Two genomic DNA libraries were prepared from Solanum lycopersicum ("MicroTom") and S.habrochaites using the λ-FIX II system (Stratagene) according to the protocols provided by the supplier. The libraries were screened with a subcloned 5.5 kb NotI-HindIII fragment of NsEPRV clone V6 corresponding to the approximate NsEPRV nucleotide positions 2–7.5 kb [1]. Λ-DNA was isolated using the Lambda Midi Kit (Qiagen) and sequenced with fluorescent chain terminators (ABI PRISM 3100 system). For analysis of DNA sequences the software programs BLAST [54] and CLUSTAL [55, 56] were used, homology searches employed public domain sequence databases (GenBank, EMBL, DDBJ, SwissProt, PDB, PIR, PRF). GenBank/EMBL/DDBJ accession numbers for sequences reported in this paper are DQ273220–DQ273264.

Southern hybridization

For Southern hybridization 1 to 2 μg of genomic DNA was sequentially digested with XbaI and an additional enzyme of the appropriate isoschizomer pair, fractionated on 1.5% agarose gels and transferred onto nylon membranes (Hybond N, Amersham) using standard techniques. Fragments amplified from clone Le1 with primers Le1-L: 5'GGAGGTATGACCA CGGATATAA 3'/Le1-R: 5'CCTGGTGCTAACTCTATTCCTG 3' (probe E1) and from clone Lh7 with primers Lh7-L: 5'GCAAGATATATCAGAAAGATTCC 3'/Lh7-R: 5'CCTTAGGATGGCATAGTCTG 3' (probe H7), respectively, were radiolabelled with α-[P³²]dATP (Amersham) by random priming and hybridized onto Southern blots at 65°C in 6 × SSC overnight and washed at 65°C in 1 × SSC (saline sodium citrate)/1%(w/v) SDS (sodium dodecyl sulphate).

RT-PCR and cDNA cloning

Total RNA was isolated from leaf material using the RNeasy Plant Mini kit (Qiagen) and enriched for polyA⁺ RNA using the Oligotex mRNA Mini kit (Qiagen). First strand DNA was produced by Revert Aid H Minus M-MuLV Reverse Transcriptase (Fermentas) according to standard protocols and used in PCR reactions with the following primer pairs: CP: 5'CWTGTTAYAAYTGYGGAAARWTAGGAC 3'/MP: 5'TTTCWATRGGNGTATCT ATTCCTTCTC 3' and TAV1: 5'RMWDNTANHAGTCAGCAGCATGAC 3'/TAV2: 5' CATHRHYTGATCTCKTDHATARTA 3' for the coding region (annealing temperature: 50°C) and IGR1: 5'CWYTTAAGWTYATGAGTAGCTAWATTAATTTATTCCTG 3'/IGR2: 5' CCTCAAMTYTGTTTAMTCCCCTAAACGG 3' (annealing temperature 56°C) for the intergenic region (Fig.5B). An actin sequence spanning an intron was amplified in parallel to detect genomic DNA contaminations using the primer pair ActL: 5'GTTGCTATTCAGGCTGTGCT 3'/ActR: 5'TCTTTTCAATGGAGGA GCTG 3' (annealing temperature: 50°C). Reactions (50 μl) contained 50 pmol of each primer, 1.5 mM MgCl₂, 150 μM dNTPs, 0.25U Taq Polymerase and ~50ng 1st strand DNA. PCR products were gel purified and cloned into the pGemT vector (Promega). In order to discriminate between different copies, cloned fragments were HinfI restricted and separated on agarose gels. Fragments producing different restriction patterns were sequenced.

Short RNA extraction and hybridization

RNA enriched for the low-molecular weight fraction (10 to 100nt) was isolated from leaves and flowers, samples of 50 μg per lane were separated on a 15% polyacrylamide gel containing 7 M urea and transferred onto nylon membranes (Hybond N⁺, Amersham) following the protocols described in [57]. The blots were hybridized with RNA probes of both orientations derived from the cloned cDNA fragments of IGRcLe-8 (DQ273223) for the intergenic region and from pooled TAVcLe-4, TAVcLe-8, TAVcLe-19 (DQ273225, DQ273229, DQ273228) for the TAV region. Hybridization conditions and probe preparation were following [57], omitting the probe fragmentation step.

Heterologous virus infection

For mechanical transmission trials, plants at the six leaf stage were inoculated with leaf extracts from S.lycopersicum infected plants with Potato virus Y (PVY) strain PVY-NTN [58] or with Tomato bushy stunt virus (TBSV) strain TBSV-type [59], respectively. The virus strains were obtained from the Department of Plant Protection Virology, University of Bari, Italy. Infected leaves were ground in 0.1 M phosphate buffer (pH 7.2) with 0.2% DIECA and the extract was rubbed on celite-dusted plants. The virus spread to younger leaves after 4–6 weeks post inoculation was verified by ELISA using TBSV and PVY detection kits (LOEWE, Germany). An ELISA sample was taken as positive when its OD value was at least three times higher than the negative control values. All determinations were run in duplicate.

Fluorescent in situhybridization (FISH)

Root tips from seedlings or plants growing in pots were treated with 0.02 M 8- hydroxyquinoline, fixed in ethanol: glacial acetic acid (3:1), digested with proteolytic enzymes, and dissected in 60% (v/v) acetic acid. Chromosome preparations were either made by squashing [60] or spreading [61]. Flower buds were fixed untreated and anthers were dissected and the stage of meiosis determined to be pachytene, before they were processed as above.

The ribosomal probe (clone pTa71), contains a 9 kb EcoRI fragment of the repeat unit of 25S-5.8S-18S rDNA from T. aestivum [62]. Part of the dispersed middle repetitive tomato sequence U30 [31] was amplified and cloned from S.lycopersicum (DQ273250). Mixtures of four probes each for S.lycopersicum and S.habrochaites (LycEPRV-Sl, LycEPRV-Sh) were selected (Table 2). PCR amplified inserts of clones were labelled with biotin 16-dUTP (Roche) or digoxigenin 11-dUTP (Roche) by random priming (Bioprime & Random primer kit; Invitrogen).

In situ hybridization followed [60]. The hybridization mixture consisted of 50 to 100 ng/slide of each probe, 50% (v/v) formamide, 2 × SSC, 10% (v/v) dextran sulphate, 0.12% (w/v) SDS, 0.12 mM EDTA (ethylene-diamine-tetra-acetic acid) and 1 μg/μl salmon sperm DNA. After overnight hybridization, slides were washed in 20% (v/v) formamide/0.1 × SSC at 42°C, giving a hybridization stringency of 85%. Hybridization sites were detected by streptavidin conjugated to Alexa 594 (Molecular Probes) or FITC (fluorescein isothiocyanate) conjugated anti-digoxigenin antibody (Roche) in 4 × SSC, 0.1% (v/v) Tween-20, 5% (w/v) BSA (bovine serum albumin). Preparations were stained with DAPI (4'-6-diamidino-2-phenylindole) and analysed on an Axioplan 2 epifluorescence microscope (Zeiss) with single band pass filters equipped with a cooled colour CCD camera (Optronics, model S97790). FISH and DAPI images were overlaid using the RGB channels of Adobe Photoshop CS and CS2 software; DAPI images were sharpened using the Gaussian deblur function and colour balance and processing of the FISH signal was achieved using only those function that treat all pixels equally. For the pachytene overlay figures (Figs. 3E and 3F) the captured colour images were converted to gray image, enhanced and overlaid: DAPI images were left B&W and the FISH signals were falsely coloured red and green, respecively. Each hybridization experiment was at least carried out twice and for each probe eight to twenty cells were analysed.