During meiosis, the micronuclei of the ciliated protist Tetrahymena thermophila elongate dramatically. Within these elongated nuclei, chromosomes are arranged in a bouquet-like fashion and homologous pairing and recombination takes place. We studied meiotic chromosome behavior in Tetrahymena in the absence of two genes, SPO11 and a homolog of HOP2 (HOP2A), which have conserved roles in the formation of meiotic DNA double-strand breaks (DSBs) and their repair, respectively. Single-knockout mutants for each gene display only a moderate reduction in chromosome pairing, but show a complete failure to form chiasmata and exhibit chromosome missegregation. The lack of SPO11 prevents the elongation of meiotic nuclei, but it is restored by the artificial induction of DSBs. In the hop2AΔ mutant, the transient appearance of γ-H2A.X and Rad51p signals indicates the formation and efficient repair of DSBs; but this repair does not occur by interhomolog crossing over. In the absence of HOP2A, the nuclei are elongated, meaning that DSBs but not their conversion to crossovers are required for the development of this meiosis-specific morphology. In addition, by in silico homology searches, we compiled a list of likely Tetrahymena meiotic proteins as the basis for further studies of the unusual synaptonemal complex-less meiosis in this phylogenetically remote model organism.
Meiosis is a special cell division in sexually reproducing organisms by which diploid chromosome sets are reduced to haploid sets owing to the disjunction of corresponding (homologous) chromosomes. In order to do so, chromosomes have to search for their homologous partners and establish physical links (crossovers) that define the pairs to be separated during the subsequent nuclear division. Crossovers (and their cytological manifestations, the chiasmata) are the consequence of the reciprocal exchange between homologous chromosomes. In the majority of the organisms studied to date, both the stable pairing of homologs and crossing over depend on the formation of DNA double-strand breaks (DSBs).
DSBs are induced by the meiosis-specific protein Spo11, which cuts DNA by a topoisomerase-like transesterase reaction by which Spo11 is covalently bound to the DNA ends at DSBs (Keeney et al., 1997; Bergerat et al., 1997). Spo11 is subsequently removed, allowing the activity of nuclease complexes to digest the 5′-to-3′ strands of the flanking DNA, rendering 3′ single-stranded overhangs on both sides of the DSB. These single strands are loaded, with the help of various cofactors, with Rad51 and Dmc1 protomers (for a review, see Shinohara and Shinohara, 2004). Rad51 and Dmc1 help the single strands to invade double-stranded homologous DNA regions (with a preference for DNA on the homologous chromosome). At this point, a decision is made as to whether this exchange will result in a crossover or a non-reciprocal exchange, a conversion (for reviews, see Hollingsworth and Brill, 2004; Bishop and Zickler, 2004; Whitby, 2005). In either case, the DSB will be repaired using the homologous sequence as a template to replenish lost DNA.
Hop2 (for homologous pairing 2; also known as TBPIP, and as PSMC3IP in mammals) functions in a complex with Mnd1 and contributes to processing of meiotic DSBs (for a review, see Neale and Keeney, 2006). The Hop2-Mnd1 complex was reported to have a dual role in vitro in stabilizing the Rad51- and Dmc1-single-stranded DNA (ssDNA) nucleoprotein filament and in enhancing the ability of the Rad51- or Dmc1-ssDNA nucleoprotein filament to capture or invade duplex DNA (Chi et al., 2007; Pezza et al., 2007; Ploquin et al., 2007). Budding-yeast Mnd1p was found to associate with chromatin independent of DSB formation, and does not colocalize with DSB sites or with Rad51p and Dmc1p complexes. Therefore, it was suggested that Mnd1 (together with Hop2) facilitates chromatin accessibility, which is required to allow strand invasion (Zierhut et al., 2004).
Tetrahymena thermophila is a ciliated protist with two cell nuclei. A polyploid macronucleus (MAC), which is transcriptionally active, serves as the soma, and a micronucleus (MIC), which is transcriptionally silent, represents the germ line. Only the MIC undergoes meiosis, whereas the MAC degenerates and a new MAC is reconstituted from the MIC after every sexual reproduction cycle (for a review, see Collins and Gorovsky, 2005). Meiosis in Tetrahymena is remarkable in many respects. It does not seem to employ a synaptonemal complex (SC), which is used by most organisms to stabilize nascent crossovers (Wolfe et al., 1976; Loidl and Scherthan, 2004). Most notably, the MIC undergoes a remarkable elongation to about 50 times its original diameter during meiotic prophase (Wolfe et al., 1976). Within this elongated MIC (the `crescent'), chromosomes are oriented in a polarized manner with all telomeres assembled at one end, resembling the conserved meiotic bouquet arrangement. At the same time, chromosome pairing takes place in the elongated MIC (Loidl and Scherthan, 2004).
In many organisms, the induction and recombinational repair of DSBs is not only a prerequisite to meiotic recombination but is also required for the stable pairing of homologous chromosomes (see Peoples-Holst and Burgess, 2005). Therefore, mutants in SPO11 and HOP2 suffer severe pairing defects. In Caenorhabditis elegans and Drosophila, however, homologous pairing is independent of Spo11-induced DSBs (Dernburg et al., 1998; McKim and Hayashi-Hagihara, 1998). In Tetrahymena, pairing takes place in the extremely elongated MICs, in which the alignment of corresponding chromosome regions is expected to be enforced by the shape of the nucleus together with the bouquet arrangement of chromosomes (Loidl and Scherthan, 2004). We wanted to know whether and to what extent the initiation and progression of recombination have an effect on pairing in this phylogenetically distant organism. To this end we identified meiotic recombination proteins in Tetrahymena and produced spo11- and hop2-null strains by MAC-targeted gene disruption, and studied the role of these genes in homologous chromosome pairing and other aspects of meiosis. We detected an unexpected requirement of Spo11p for the elongation of MICs and, indirectly, for the alignment of homologous chromosomes.
An inventory of conserved meiotic proteins in Tetrahymena
We screened for putative meiotic genes of Tetrahymena by generating clusters of orthologous proteins from Arabidopsis, the mouse, budding yeast, fission yeast and Tetrahymena [predicted from the sequenced T. thermophila macronuclear genome (Eisen et al., 2006); http://www.tigr.org/tdb/e2k1/ttg/] using OrthoMCL (Chen et al., 2006). Clusters with proteins from at least two species associated with Gene Ontology terms `meiosis' (GO:0007126) and/or `meiotic recombination' (GO:0007131) in the Gene Ontology Database (http://www.geneontology.org/) were selected. Independently, Tetrahymena proteins were subjected to reciprocal BLAST searches with the other proteomes and their similarities were assessed by the rank sum of reciprocal BLAST hits. Furthermore, the InParanoid program was used for the identification of orthologs and in-paralogs from pairwise comparisons of Tetrahymena and the other species (Remm et al., 2001). A list of the best-conserved Tetrahymena meiotic proteins is provided as supplementary material Table S1. As a criterion for a possible meiosis-specific function of Tetrahymena proteins, expressed sequence tag (EST) data from the Tetrahymena Genome Database (http://www.ciliate.org/) were incorporated (supplementary material Table S1).
Among the best-conserved putative meiotic genes were TTHERM_00627090 and TTHERM_01190440. In the Tetrahymena Genome Database, TTHERM_00627090 is annotated as a homolog of the conserved meiotic recombination gene SPO11 and ESTs are reported only from conjugating cells (http://www.ciliate.org). We confirmed the meiosis-specific expression of TTHERM_00627090 by reverse transcriptase (RT)-PCR (Fig. 1A), which supports its functional homology to SPO11 in other organisms. Tetrahymena does not have obvious SPO11 paralogs as do plants and other groups including the related apicomplexans, presumably owing to their loss after rounds of gene duplications preceding the diversification of eukaryotic lineages (Malik et al., 2007).
The protein family (Pfam) database (Sonnhammer et al., 1997) lists two Tetrahymena Hop2 family members, of which TTHERM_01190440p is the clear reciprocal best proteome blast hit of Hop2 proteins from the organisms tested (Fig. 1B). However, ESTs for the corresponding gene are reported from vegetatively growing and conjugating cells (http://db.ciliate.org/), raising the issue of whether TTHERM_01190440p is functionally equivalent to meiosis-specific Hop2. The second Hop2 member, TTHERM_00794620p, is a slightly more divergent paralog (Fig. 1B), of which ESTs have so far only been recovered from conjugating cells. In the following, we will designate the meiosis-specific HOP2 (TTHERM_00794620) as HOP2A and the other as HOP2B. RT-PCR confirmed the meiosis-specific expression of HOP2A and the ubiquitous expression of HOP2B (Fig. 1A).
Most Hop2 proteins possess a tripartite organization, with a 70- to 80-amino-acid (aa)-long segment of highest sequence conservation and similarity to the `winged helix' fold, a coiled-coil-forming segment, and a helix-rich C-terminal region required for efficient DNA binding (Enomoto et al., 2004). Both Tetrahymena Hop2 proteins share these features but, in addition, feature a predicted 100- to 200-aa-long compositionally biased N-terminal extension. Such an extension to the core sequence is also observed in other protist groups such as Plasmodium and Naegleria.
It is known that Hop2 acts in a complex with Mnd1 (for a review, see Neale and Keeney, 2006). In accordance with the presence of two Tetrahymena Hop2 homologs, two Mnd1 homologs were also found (supplementary material Fig. S1), one of which, TTHERM_00300660p (MND1-like A), is represented only by conjugative ESTs, and the other, TTHERM_00382290p (MND1-like B), by vegetative and conjugative ESTs. RT-PCR confirmed these expression patterns (Fig. 1A).
Fertility is strongly reduced in spo11Δ and hop2AΔ single mutants
We produced single-knockout strains for SPO11 and HOP2A by macronuclear gene disruption in each of two different mating types to study the course of meiosis in the absence of these two important meiotic recombination proteins. The loss of macronuclear copies of the genes was confirmed by Southern hybridization (supplementary material Fig. S2). When we attempted to produce knockouts of HOP2B, we were unable to maintain clones at paromomycin concentrations above 1 mg/ml. This suggests that cells in which a substantial proportion of macronuclear HOP2B copies are knocked out are not viable. In fact, we noticed the persistence of a significant amount of HOP2B genes by Southern hybridization (supplementary material Fig. S2). Therefore, Hop2Bp is very probably essential for vegetative growth.
The fertility of spo11Δ and hop2AΔ single mutants was determined by testing the viability of meiotic products. For this, conjugating pairs were singled in small drops of growth medium and the emergence of colonies in these drops was scored. In the wild-type control, 76% of conjugants produced sexual progeny that showed vigorous growth and 13% showed weak growth; only 11% showed no growth (n=123 conjugants tested), indicating that the majority of meioses were normal. For the hop2AΔ strains, 20/265 (8%) conjugants grew into colonies, two of which died-out after 2 days and could not be tested further. The others remained viable upon transfer to paromomycin-containing medium, meaning that they had retained their old MACs (that carry the neo4 resistance gene) and hence were not meiotic products but vegetative progeny of the parental strains. This indicates that practically all cells that underwent meiosis produced dead progeny and that deletion of HOP2A confers sterility.
In the case of the spo11Δ mutation, 107/263 (41%) conjugating pairs produced viable offspring. However, all of these carried parental MACs. It is conceivable that many of the mutant cells either underwent defective meiosis while retaining their old MACs or stopped conjugation and meiosis. It remains to be determined at which step and how cells desist from producing sexual progeny and return to a vegetative state while retaining their old MACs. Those cells that completed meiosis did not produce viable progeny.
Timing of meiosis and morphology of meiotic nuclei in hop2AΔ and spo11Δ single mutants
Normal meiotic development of MICs has been divided into six stages (Sugai and Hiwatashi, 1974) (Fig. 2A,B). Whereas, prior to meiosis, chromatin distribution is homogeneous within the MIC, stage I is characterized by an inhomogeneous distribution within the round MIC. Stage-II MICs are egg- to spindle-shaped and, during stage III, they become torch-shaped. During stage IV, MICs reach their maximal elongation. Elongated MICs are referred to as crescents. Stage V is characterized by the shortening and widening of the crescents, which adopt a filamentous texture. At stage VI, these structures disintegrate and distinct entities emerge, which can sometimes be distinguished as five bivalents (Fig. 3). Completion of meiotic prophase occurs just over 4 hours after mixing (Martindale et al., 1982). Another paper (Kaczanowski et al., 1985) has reported an even shorter duration. The appearance of bivalents is followed by a first and a second meiotic anaphase.
Centromeres (as delineated by immunostaining of the centromeric H3 histone variant Cna1p) assembled in a limited region at the periphery of premeiotic and stage I MICs (Fig. 4A) (Cervantes et al., 2006; Cui and Gorovsky, 2006). This arrangement might be due to the Rabl orientation [the localization of centromeres and telomeres at opposite poles, reflecting the arrangement of chromosome arms at the preceding anaphase/telophase (Cowan et al., 2001)]. During elongation of the MIC, the centromere cluster transiently resolved but, at stage IV, centromeres assembled at one end of the elongated MIC (Fig. 4A). The two ends of the elongated MIC can be discriminated by the weak DAPI (4′,6-diamidino-2-phenylindole) staining of one tip. It is this tip where all telomeres are clustered (Loidl and Scherthan, 2004) and the centromeres occupy the opposite end.
In the hop2AΔ mutant, the early meiotic development of MICs was normal (Fig. 2A). Elongation of MICs occurred with wild-type dynamics (Fig. 2B), the centromeres became transiently scattered and finally assembled at the tip of the stage-IV MIC (Fig. 4B). During stage V, distinct chromatin threads developed inside the MICs. However, this was not followed by the emergence of condensed bivalents, but chromatin masses remained more entangled. In favorable nuclei, around ten individual structures could be counted (Fig. 3), suggesting failure of crossover formation and/or a defect in chiasma structure that results in the separation of bivalents into univalents. Chromosomes that are not joined in bivalents are expected to segregate randomly during the first meiotic division. Indeed, meiotic divisions were often clearly asymmetric (Fig. 3), which indicates the separation of unequal numbers of chromosomes. In addition, asymmetric meiosis II configurations were frequently found (Fig. 3), which arise because of the separation of interkinesis nuclei with unequal chromosome numbers. Despite these anomalies, there was no notable delay in the progression of meiotic stages, which suggests the absence of a checkpoint-dependent delay in meiosis (Hochwagen and Amon, 2006) in Tetrahymena (see also Song et al., 2007). Chromosome fragments (which would indicate the presence of unrepaired DSBs – see below) were not observed at any stage. Altogether, cytological evidence suggests that, in the hop2AΔ mutant, bivalents are not formed but that single and intact chromosomes are irregularly separated in the first meiotic division.
In the spo11Δ mutant, there was only limited elongation of meiotic MICs and full crescents did not form. Development of crescents was normal up to stage II. Stage-III MICs lacked the typical torch shape (Fig. 2A) and did not further elongate to reach stage IV. As in the wild type, the arrangement of centromeres was polarized in early-meiotic MICs and was lost as soon as the MICs started to elongate; however, the centromeres failed to reassemble at one pole (Fig. 4C). Thus, whereas in wild type and in the hop2AΔ mutant, stage-IV MICs are the most prevalent single stage at t=180, t=210 and t=240 minutes after mixing, aberrant stage-III MICs transiently arrested during these time points in the spo11Δ mutant. These aberrant nuclei then progressed to stage-V morphology with about a 30-minute delay (Fig. 2B). At stage VI, when wild-type conjugants displayed condensed bivalents, the chromosomes in the spo11Δ mutant remained relatively long and entangled. Although they could not be resolved into individual structures, they probably represent univalents (Fig. 3). As in the hop2AΔ mutant, both meiotic divisions were often asymmetric (Fig. 3).
The end-point of normal early development of sexual progeny prior to re-feeding are cells (exconjugants) with a single MIC and two new MACs (Coyne et al., 1999). In total, 94% (n=181) of wild-type cells that had undergone meiosis were of this type at 48 hours after mixing. For spo11 and hop2A single mutants, 55% (n=159) and 18% (n=123) of cells, respectively, showed this normal development. The remaining cells were arrested at earlier stages or showed aberrant numbers of nuclei. They often failed to eliminate the old MAC or a MIC, which might be a consequence of a dysfunctional MAC derived from aberrant meiosis in the mutants.
γ-H2A.X and Rad51p signals appear transiently during hop2AΔ meiosis
The presence of the phosphorylated histone H2A.X variant (γ-H2A.X) is an indicator of the formation of meiotic DSBs (Mahadevaiah et al., 2001; Song et al., 2007). Likewise, the recombination protein Rad51p is a marker for meiotic DSBs undergoing repair (see Introduction). To study DSB turnover in the hop2AΔ mutant, we tested the occurrence of γ-H2A.X and Rad51p by immunofluorescence. In wild type, γ-H2A.X staining begins at early stage II, continues through stage V and disappears at stage VI (Song et al., 2007) (Fig. 4D). Rad51p foci emerge soon after the meiotic MIC begins to elongate and are maintained beyond the stage of maximal MIC elongation (see Loidl and Scherthan, 2004). During later stages, up to the first or the second division, homogeneous Rad51p labeling of the nuclei indicates the persistence of unbound Rad51p (Loidl and Scherthan, 2004) (Fig. 4G).
In the hop2AΔ mutant, γ-H2A.X staining appeared similar to wild type and was maintained through stage V. Later stages did not display signals (Fig. 4E). Similarly, numerous Rad51p foci were present in stage-IV (Fig. 4H) and stage-V MICs, whereas later stages displayed only homogeneous staining (not shown). Together with the absence of chromosome fragments (see above), the turnover of γ-H2A.X and Rad51p suggests the formation and efficient repair of meiotic DSBs.
Cytological evidence for the lack of DSBs in spo11Δ meiosis and the requirement of DSBs for full crescent formation
To test whether meiotic DSBs were formed, we immunostained mutant meiotic MICs for γ-H2A.X. Whereas, in wild-type (Fig. 4D) and hop2AΔ (Fig. 4E) meiosis, numerous γ-H2A.X patches were present in crescents, no γ-H2A.X staining was observed in the spo11Δ mutant (Fig. 4F). No Rad51p foci were formed in meiotic nuclei of the spo11Δ mutant either. Instead, the MICs displayed homogeneous Rad51p staining (Fig. 4I). This is probably due to the expression and dispersed distribution of Rad51p within nuclei but its failure to accumulate in nucleofilaments in the absence of DSBs. These observations are consistent with the lack of DSBs, which is expected in the absence of Spo11p.
To determine whether Spo11p per se or the occurrence of DSBs is required for crescent formation, conjugating spo11Δ cells were treated with 100 μg/ml cisplatin. Cisplatin induces DNA cross-links, the excision of which creates DSBs. Cisplatin-generated DSBs can stimulate meiotic recombination (Hanneman et al., 1997; Sanchez-Moran et al., 2007). Cells were exposed to cisplatin at 2 hours after mixing the strains, which roughly corresponds to the time at which there is the maximal number of cells being in early stage II (Martindale et al., 1982), when DSBs are believed to be induced (Song et al., 2007). Cisplatin-treated cells were harvested at 4 or 5 hours after mixing. Cisplatin treatment restored crescent formation (Fig. 2A) and cisplatin-treated cells displayed massive γ-H2A.X patch formation in MICs but also numerous weaker foci in the MACs (Fig. 4F). Thus, the presence of DSBs is sufficient to trigger the elongation of MICs.
Homologous pairing is moderately reduced in spo11Δ and hop2AΔ single mutants
We employed fluorescence in situ hybridization (FISH) to study the pairing behavior of homologous chromosomal loci (Loidl and Scherthan, 2004). Pairing was determined by the merging of FISH signals into a double dot or a single dot. Two sites, an intercalary region (A) and a region close to the telomere of a chromosome (B), were evaluated (Fig. 5). In the wild type, pairing continuously increased for both regions throughout meiotic prophase (Fig. 6 and supplementary material Table S2). The frequency of nuclei with pairing at region B rose from 30% in stage-I MICs to 89% in stage-IV MICs. In the hop2Δ mutant, the level of pairing achieved in stage IV was somewhat lower (69%).
In the spo11Δ mutant, the maximal pairing (52%, occurring at aberrant stage III) was higher than in stage I in all genotypes, indicating that some pairing can take place in the absence of DSBs (Fig. 6). To test whether this pairing was dependent on the clustering of telomeres, we performed FISH with a telomeric probe (Fig. 5). Both in the wild type and in the mutant, telomeres were only weakly labeled but seemed to be concentrated at one side of stage-II MICs. However, whereas, in wild type, telomere clusters became increasingly distinct as MIC elongated, they were no longer visible, possibly due to dispersal, in aberrant stage-III mutant MICs. Thus, it remains unclear whether the residual pairing in the incomplete crescents of the spo11Δ mutant is promoted by the bouquet-like clustering of telomeres.
Basic meiotic recombination proteins are conserved in Tetrahymena
Tetrahymena possesses close homologs to many genes that have previously been identified as playing a role in the formation and processing of meiotic DSBs (supplementary material Table S1). However, there is little evidence for the presence of genes encoding proteins involved in SC formation. Using in silico homology searches, we have not yet found homologs or even weak homologs of the Saccharomyces cerevisiae SC proteins Zip1, Red1 and Zip2, or related proteins in other species. However, these proteins are notoriously poorly conserved between SC-possessing organisms. More importantly, there was also no Tetrahymena homolog found for Hop1, Mer3, Msh4 or Msh5. Hop1 is the best-conserved protein among those found in SCs or SC-related structures, with representatives from S. cerevisiae (Hollingsworth and Byers, 1989; Hollingsworth et al., 1990), Schizosaccharomyces pombe (Lorenz et al., 2004), C. elegans (Couteau and Zetka, 2005; Martinez-Perez and Villeneuve, 2005) and plants (Caryl et al., 2000; Nonomura et al., 2004), and possibly also mammals (Chen et al., 2005). Mer3, Msh4 and Msh5 are the most conserved members of the so-called ZMM (or SIC) group of proteins, the role of which (besides in the interference-prone crossover pathway) is in the nucleation of the SC (Lynn et al., 2007). The lack of Hop1, Mer3, Msh4 and Msh5 corroborates previous reports (Wolfe et al., 1976; Loidl and Scherthan, 2004) on the absence of an SC or of SC-related structures in Tetrahymena.
Normal SCs were found in Eimeria tenella (del Cacho et al., 2005), which belongs to the subphylum Apicomplexa and is related to the ciliates within the alveolates (Baldauf, 2003), and we have observed residual SC-like structures in the ciliate Stylonychia (J.L., unpublished result). Thus, the ciliates seem to represent a group within which SCs were reduced in the course of evolution, resembling the situation in ascomycetous fungi, which has members (S. pombe and Aspergillus) in which SCs are missing and others (S. cerevisiae) in which they are present (for a review, see Loidl, 2006).
Hop2 and Mnd1 are similar in sequence, and Tetrahymena possesses two Hop2 and two Mnd1 homologs
Hop2 and Mnd1 are meiosis-specific proteins that function in a stable heterodimeric complex (for a review, see Neale and Keeney, 2006). Interestingly, they show a striking resemblance to each other in their general architecture. Typically, a 70- to 80-aa-long highly conserved N-terminal-to-central region is followed by a coiled-coil-forming segment of ca. 60-80 aa and a C-terminal helix-rich part of a similar size. Independent analysis of the best-conserved segment in Hop2 and Mnd1 protein families shows its resemblance to the superfamily of `winged helix' DNA-binding domains (supplementary material Fig. S1; for additional information see http://mendel.imp.ac.at/SEQUENCES/hop2mnd1/). More strikingly, sequence-analysis evidence suggests that Hop2 and Mnd1 proteins are similar in sequence and therefore potentially homologous protein families (see supplementary material Fig. S1). Likewise, a weak sequence similarity has been previously suggested for budding-yeast Hop2 and Mnd1, and it has been proposed that these proteins might originate from a common ancestor (Shinohara and Shinohara, 2004).
Interestingly, the Hop2-Mnd1 system seems to have undergone duplication in the evolutionary history of Tetrahymena, because both protein families are represented by two homologs with distinct expression patterns in this species. Just as for HOP2, there is a ubiquitously and a meiotically expressed version of MND1 (Fig. 1), which raises the possibility that a meiotic and a ubiquitous Mnd1p-Hop2p complex exists.
Whereas one Hop2p version (Hop2Ap) was dispensable for vegetative growth, HOP2B could not be knocked out completely, with about 30-50% of the ca. 45 macronuclear copies of the gene remaining wild type (see supplementary material Fig. S2). When conjugation was induced in cells carrying a reduced dosage of Hop2Bp, they produced dead sexual progeny (data not shown), possibly because these partial hop2B-knockdown cells might not undergo successful meiosis. This could mean that a higher Hop2Bp dosage is required for meiosis than for vegetative growth and that Hop2Bp might have a specific meiotic function in addition to its essential function in vegetative growth.
The requirement of Hop2Bp for vegetative growth is consistent with evidence from Arabidopsis that its sole Hop2 and Mnd1 versions are involved in the vegetative DNA-damage-repair pathway (Domenichini et al., 2006).
DSBs are formed and repaired in the absence of Hop2Ap
HOP2 is a conserved gene from S. cerevisiae, S. pombe (meu13+), Arabidopsis (AHP2) and vertebrates, but is not found in C. elegans or Drosophila, in which homology recognition is regulated differently (Saito et al., 2004). Where present, it is required for the recombinational repair of meiotic DSBs (Schommer et al., 2003; Petukhova et al., 2003; Tsubouchi and Roeder, 2003; Ploquin et al., 2007). In budding yeast, hop2-null mutant cells fail to repair DSBs and to pair chromosomes homologously, and therefore arrest in meiotic prophase (Leu et al., 1998). In an Arabidopsis hop2 (ahp2) mutant, chromosome fragments at anaphase I indicate a failure to repair DSBs (Schommer et al., 2003). In the mouse (Petukhova et al., 2003) and in Arabidopsis (Schommer et al., 2003), there is no homologous pairing in the absence of Hop2. These defects are consistent with the biochemical function of the Hop2-Mnd1 complex in stabilizing Rad51 and/or Dmc1 presynaptic filaments and enhancing their ability to capture duplex DNA (Ehmsen and Heyer, 2007) in those organisms in which stable homologous contacts depend on recombination (Peoples-Holst and Burgess, 2005). Fission-yeast Hop2 (Meu13) seems to be different in that it plays a relatively subordinate role in recombinational repair of DSBs (Nabeshima et al., 2001).
In the Tetrahymena hop2A mutant, γ-H2A.X staining appeared and disappeared with wild-type dynamics, suggesting that the numerous DSBs whose presence is revealed by an estimated >100 Rad51p foci become processed or repaired. Moreover, extensive fragmentation of chromosomes that would be expected if DSBs remained unrepaired, was not observed. However, from diplotene to metaphase I, univalents were formed. Therefore, the hop2AΔ mutant might efficiently repair DSBs via the sister chromatid. This would be consistent with evidence that, in vivo, the Hop2-Mnd1 complex prefers Dmc1 over Rad51 (reviewed by Sheridan and Bishop, 2006; Neale and Keeney, 2006) and would, therefore, be less important for Dmc1-independent sister recombination (see Sheridan and Bishop, 2006). The relatively mild cytological phenotype displayed by hop2AΔ could possibly also be explained by its partial redundancy with its ubiquitously expressed HOP2B paralog (see above). The timely progression of meiosis in the hop2A mutant is reminiscent of meiosis in the presence of unphosphorylated H2A.X, where repair of meiotic DSBs is impaired (Song et al., 2007). The ability of this mutant to carry out both meiotic divisions led Song et al. to the conclusion that a recombination checkpoint is either weak or nonexistent in Tetrahymena cells (Song et al., 2007).
A high degree of homologous pairing is attained in the absence of Spo11p and Hop2Ap
In a variety of organisms, the induction and recombinational repair of DSBs was found to be required for stable homologous pairing (see Peoples-Holst and Burgess, 2005), and mutants defective in these processes suffer severe pairing defects. Nevertheless, in Tetrahymena, we observed a high residual level of associated or fused FISH signals in the absence of Spo11p and Hop2Ap. The arrangement of chromosomes inside the elongated (or, in the case of spo11Δ, incompletely elongated) MIC, with their telomeres clustered at one pole, resembles the conserved meiotic bouquet (see Scherthan, 2001). The spatial constraints within this narrow tube might enforce the roughly parallel alignment of chromosome arms, limiting homologous regions to a similar latitude in the nucleus (Lorenz et al., 2003; Barzel and Kupiec, 2008). Although the initial centromere-telomere polarization is not maintained beyond stage II in the spo11Δ mutant, it could be sufficient for the degree of pairing observed.
Although a course homologous alignment might be achieved in the spo11Δ and hop2AΔ mutants by the chromosomal bouquet arrangement, it might not be sufficient for the precise pairing of homologous chromosomal regions on a molecular scale. Even in wild-type meiosis, fully extended MICs sometimes displayed homologous FISH signals that were separated along their length axis (Fig. 5C). This suggests that, in addition to the bouquet, another mechanism might contribute to the stable pairing of homologous chromosomal regions, paralleling the situation in S. pombe. There, by virtue of the elongated bouquet (`horsetail') nucleus, homologs are predisposed to alignment even in the absence of Spo11 (Rec12)-dependent recombination (Nabeshima et al., 2001), but intimate and stable pairing is achieved only by recombination (Ding et al., 2004).
DSBs are required for the shaping of meiotic nuclei in Tetrahymena
The meiotic MICs of Tetrahymena are lined by intranuclear microtubules (Wolfe et al., 1976) and the inhibition of microtubule formation by nocodazole prevents both MIC elongation and bivalent formation (Kaczanowski et al., 1985). Thus, it was proposed that the elongation of the MIC promotes chromosome pairing or another step in the formation of chiasmata (Kaczanowski et al., 1985; Loidl and Scherthan, 2004). Meiotic DSBs are formed as soon as MICs start to elongate (Song et al., 2007). Therefore, DSBs do not depend on the extensive elongation of the MIC; in fact, as we showed here, quite the contrary is the case. Whereas we could rule out that the presence of the Spo11 protein is a precondition for MIC elongation because the induction of DNA lesions by cisplatin is sufficient, we do not know at present how DSBs, or possibly other DNA lesions, dictate this morphological process. It is conceivable that checkpoint-like regulation is involved to ensure that subsequent steps in meiosis only ensue if DSBs are formed. If this were the case, the lack of DSBs would be expected to retard the progression of meiosis. Indeed, at the level of precision allowed by our 30-minute-interval time-course experiments, a slight delay was noted (Fig. 2B). Although little can be said about the possible signaling pathway that triggers MIC elongation, it is clear that H2A.X phosphorylation [which is central to the DSB response leading to DNA-damage signaling and repair – see Nussenzweig and Paull (Nussenzweig and Paull, 2006) and literature citations therein] is not involved, because MICs are elongated in its absence (Song et al., 2007).
In any case, via their role in the development of the extremely elongated crescent, DSBs contribute to the alignment of homologous chromosomes.
Materials and Methods
Total RNA was extracted using TRIzol reagent (Invitrogen) from wild-type (B2086 and CU428) strains and residual genomic DNA was eliminated using the Turbo DNase kit (Ambion). cDNA was synthesized from 5 μg of total RNA by using a RevertAid H Minus First Strand cDNA Synthesis kit (Fermentas) with a random hexamer as a primer. cDNA from 12.5 ng total RNA was amplified by PCR (94°C for 20 seconds, 50°C for 30 seconds, 68°C for 1 minute). The following primers were used to amplify cDNAs of the indicated genes: HOP2A, 5′-TGAGGCTGCACTTAGTTAGGCTG-3′ and 5′-CCTTGACGGCTTCTTCACAAGTTC-3′; HOP2B, 5′-GCACCCTTTTTAGCAGCTCCACC-3′ and 5′-ACGATCAAGCTGAAGATACAGG-3′; SPO11, 5′-TTATGGATTGAAACAAACTCTGC-3′ and 5′-TGGCGATTAAGGCATCCAGC-3′; TTHERM_00300660, 5′-TCACTAAGCTAATTATCCCAAGCAT-3′ and 5′-TCATCTCTGTCATCTGATGTTCT-3′; TTHERM_00382290, 5′-AGAATTAGAGGATGAGCTAGAG-3′ and 5′-CATACATTATCGAGATCAGCTGG-3′; RPL21, 5′-AAGTTGGTTATCAACTGTTGCGTT-3′ and 5′-GGGTCTTTCAAGGACGACGTA-3′.
Somatic gene disruption
Tetrahymena thermophila (previously known as Tetrahymena pyriformis) wild-type strain B2086 and isolate CU428 were used for macronuclear gene disruption.
For the design of knockout constructs, see supplementary material Fig. S2. The HOP2A disruption construct was made as follows: a genomic region flanking the 5′ end of the HOP2A gene was amplified by PCR using total genomic DNA of CU428 as a template and HOP2KO5FW (5′-GGACTCGAGAAGATTCATGGTCTACTTGACTC-3′, XhoI recognition sequence is underlined) and HOP2KO5RV (5′-GTTCACTAGTGGATCCCCAATTACTTCTTCGTCAAGG-3′, SpeI and BamHI recognition sequences are underlined) as primers. A genomic region 3′ of the HOP2 gene was also amplified by PCR with primers HOP2KO3FW (5′-GGATCCACTAGTGAACTTGTGAAGAAGCCGTCAAGG-3′, BamHI and SpeI recognition sequences are underlined) and HOP2KO3RV (5′-CGCTCTAGAGCTATTAAATTTGTTAACTTATAC-3′, XbaI recognition sequence is underlined). The 5′ and 3′ PCR products were then connected by overlapping PCR with HOP2KO5FW and HOP2KO3RV. BamHI and SpeI sequences were made at the overlapping site. The connected PCR product was cloned into XbaI- and XhoI-digested pBlueScript SK(+) vector. Finally, the neo4 cassette, which confers paromomycin resistance in Tetrahymena (K.M., unpublished; the nucleotide sequence of the cassette is available as DDBJ/EMBL/Genbank EU606202), was introduced into the BamHI and SpeI sites. The HOP2A disruption construct was excised from the vector backbone by XbaI and XhoI, and used for transformation.
The SPO11 disruption construct was made as was the HOP2A disruption construct except that primers SPO11KO5FW (5′-GGACTCGAGCTAAATATATTATTCAAACCAACTTATGCTC-3′) and SPO11KO5RV (5′-GCTTACTAGTGGATCCGTGTCTTTGGAGTATTTGAGC-3′) were used to amplify a 5′-flanking sequence of the SPO11 gene, and SPO11KO3FW (5′-CACGGATCCACTAGTAAGCTGGATGCCTTAATCGCC-3′) and SPO11KO3RV (5′-GCCTCTAGACACAGTTTTCATCCATTTAAAGCTCC-3′) to amplify a 3′-flanking sequence of the SPO11 gene (restriction-enzyme recognition sites are underlined).
The HOP2B disruption construct was made solely by PCR. A genomic region flanking the 5′ end of the HOP2B gene was amplified by PCR with primers HOP2BKO5FW (5′-AGTAATACAATACTCATAAGAC-3′) and HOP2BKO5RV (5′-GTCTATCGAATTCCTGCAGCCCGTTTAAGCATCAGAATCCG-3′, the underlined sequence is complementary to the 5′-arm of the neo4 cassette). A genomic region 3′ of the HOP2B gene was also amplified by PCR with primers HOP2BKO3FW (5′-CTGGAAAAATGCAGCCCGTATTGCTTGTAATTAGCACTG-3′, the underlined sequence is complementary to the 3′-arm of the neo4 cassette) and HOP2BKO3RV (5′-AAGAACAATATGTGAACACC-3′). Three pieces of DNA – the 5′-flanking sequence, the SmaI-digested neo4 cassette and the 3′-flanking sequence – were then connected and amplified by overlapping PCR using HOP2BKO5FW and HOP2BKO3RV. The PCR product was directly used for transformation.
The constructs were introduced into the corresponding macronuclear loci of B2086 and CU428 cells by biolistic transformation. The transformants were exposed to increasing concentrations of paromomycin to gradually select for cells in which all macronuclear copies of the wild-type allele were replaced by the deletion construct because of phenotypic assortment (Cassidy-Hanley et al., 1997).
Culture was performed at 30°C according to standard methods (reviewed by Orias et al., 2000). Cells were made competent for conjugation by starvation in 10 mM Tris-HCl (pH 7.4) for at least 16 hours. Conjugation and meiosis were induced by mixing starved cultures of cells of different mating types at equal densities. Samples of conjugating cultures were drawn at the indicated time points and subjected to cytological preparation.
A cisplatin stock solution of 2 mg/ml starvation medium was prepared by shaking the mixture for several hours at room temperature. Stock solution was added to a final concentration of 100 μg/ml to conjugating cells 2 hours after mixing and the cells were cultured for another 2-3 hours in the presence of cisplatin. After this time, the cisplatin was washed out by centrifugation and the cells were resuspended in starvation medium. Immediately thereafter, the cells were fixed by one of the methods described below.
Cytological preparation and staining
For FISH (see below), 5 ml of cell suspension were gently centrifuged (3 minutes, 350 g) and the pellet was resuspended in 1 ml Carnoy's fixative (methanol:chloroform:acetic acid, 6:3:2). After 1 hour at room temperature, the cells were again centrifuged and resuspended in 500 μl of 70% ethanol. Several drops of fixed cells were applied to a clean slide and air dried.
The preparation protocol for chromosome staining with DAPI (4′,6-diamidino-2-phenylindole) and for immunostaining of Rad51p followed Loidl and Scherthan (Loidl and Scherthan, 2004). In short, a suspension of conjugating cells was fixed by the addition of formaldehyde and Triton X-100 (final concentrations of 4% and 0.5%, respectively). After careful mixing the cells were left for 30 minutes at room temperature, then centrifuged and the pellet resuspended in 1:10 volume of 4% formaldehyde + 3.4% sucrose. 80 μl of this mixture were spread on a clean slide and air-dried. For DAPI staining, dry slides were incubated for 5 minutes in 1 × phosphate buffered saline (PBS) and mounted under a coverslip in Vectashield anti-fading agent (Vector Laboratories, Burlingame, CA, USA) supplemented with 1 μg/ml DAPI.
For immunostaining of γ-H2A.X and Cna1p, 5 ml of conjugating cell culture were mixed with 20 μl of partial Schaudin's fixative (saturated HgCl2, ethanol 2:1) and kept at room temperature for 5 minutes, followed by two washes in methanol. Cells suspended in methanol were dropped on a slide and air-dried for 30 minutes (see Song et al., 2007).
For immunostaining, slides were incubated in 1 × phosphate buffered saline (PBS) and 1×PBS + 0.05% Triton X-100 for 10 minutes each. Primary antibodies against Rad51p (1:50 mouse monoclonal antibody, Clone 51RAD01, NeoMarkers, Fremont, CA), Cna1p (1:200 rabbit anti-Cna1p) (Cervantes et al., 2006), or γ-H2A.X [1:200 purified mouse monoclonal anti-H2A.X phosphorylated (Ser139) antibody Clone 2F3, BioLegend, San Diego, CA] were applied under a coverslip for 3 hours to overnight at room temperature. The coverslip was removed and the preparations were rinsed with 1×PBS, post-fixed with ice-cold 96% ethanol for 30 seconds and incubated in 1×PBS + 0.05% Triton X-100 for 10 minutes. FITC-conjugated anti-mouse antibody or Cy3-conjugated anti-rabbit antibody was applied under a coverslip for ∼2 hours at room temperature. Finally, the slides were incubated twice for 10 minutes in 1×PBS and mounted under a coverslip in Vectashield supplemented with DAPI as a DNA-specific counterstain.
Fluorescence in situ hybridization (FISH)
For the delineation of an intercalary chromosomal region (region A), a compound FISH probe was prepared by PCR with primer pairs TetFW6 (5′-CTTAATTTCCCTCCACCACACTCAGAC-3′), TetRV6 (5′-TTAACTCTGAGACATCATCAGGGAAG-3′), and TetFW7 (5′-AGCTAATAATTATCTCCCTCATTGCCTACCTGC-3′), TetRV8 (5′-TCATTTCAGGAATAGCCAGTTCCCAACC-3′). Together, they comprised a ∼13.8-kb probe to a locus in the vicinity of the one studied by Loidl and Scherthan (Loidl and Scherthan, 2004). A FISH probe for a distal chromosomal locus (region B) was prepared by PCR-amplification of two ∼8.7-kb-long genomic DNA fragments. The primers used were TetFW2 (5′-ATGACCCTGAAGACATGC-3′), TetRV2 (5′-TACCTTCTTGCTGATGGC-3′) and TetFW3 (5′-TCTACCTCAGAGAGTTGG-3′), TetRV3 (5′-AGCAATTCTTTCTCCGGC-3′). The samples were pooled to obtain an ∼17.6-kb contiguous probe for FISH. The PCR products were purified with Micropure EZ (Millipore) and labeled with Cy3 or digoxigenin by nick translation. For telomere FISH, 5′-plus 3′-biotinylated 24-mer oligonucleotides homologous to the telomere (G4T2)n and subtelomere (G4T3)n repeats were synthesized (MWG Biotech AG, Ebersberg, GER) and pooled (Loidl and Scherthan, 2004).
Preparations (see above) were incubated for 1-3 hours in 4×SSC with 0.1% Tween 20. Denaturation of the slides was performed in 70% formamide, 2×SSC, pH7 for 2 minutes at 68°C. The slides were then dehydrated in ice-cold 90% ethanol and air-dried. The DNA probe was dissolved in 8 μl of a solution consisting of one part dextran sulfate, five parts 4×SSC and six parts formamide, heated to 95°C for 5 minutes to denature the DNA and put on ice. 8 μl of denatured probe were applied to a slide, covered with an 18×18-mm coverslip and sealed with rubber cement (Marabu, Hamm, Germany) under a coverslip. After another round of denaturation on the slide for 10 minutes at 80°C, hybridization was allowed to proceed for 36-48 hours at 37°C. After hybridization, slides were washed three times for 5 minutes in 1×SSC at 37°C. Digoxigenin-labeled probe was detected using FITC-conjugated anti-digoxigenin antibodies; biotinylated telomere probe was detected using ExtrAvidin-FITC. Finally, slides were embedded in Vectashield supplemented with DAPI.
Preparations were evaluated using an epifluorescence microscope equipped with single-band-pass filters for the excitation of blue (DAPI), green (FITC) and red (Cy3) fluorescence. z stacks of images were recorded with a cooled CCD camera using MetaVue software (Universal Imaging, Downingtown, PA), deconvolved with AutoDeblur (AutoQuant Imaging, Watervliet, NY) and projected in two dimensions using ImageJ (Wayne Rasband, N.I.H.; http://rsb.info.nih.gov/ij/) software. Images of an object taken with different filters were assigned false colors and merged.
Testing the viability of sexual progeny
Conjugating-cell pairs were mouth-pipetted and released into single droplets of growth medium 6-10 hours after mixing and kept for 2 days at 30°C. Conjugants that did not develop colonies after 2 days were scored as having defective meiosis or development. Colonies that had formed after this time were transferred to multiwell plates and, after another day of growth, aliquots were transferred to medium containing 100 μg/ml paromomycin and 1 μg/ml CdCl2. The old MAC carries the neo4 paromomycin resistance gene, whereas the new MAC that is formed after meiosis does not. Colonies that survived 2 days of paromomycin treatment were scored as parental (i.e. having returned to vegetative growth before completing meiosis or after defective meiosis). Thus, only the conjugants that produced paromomycin-sensitive colonies were counted as having undergone meiosis successfully.
We are grateful to Harmit Malik for the antibody to Cna1p. We gratefully acknowledge the help of Maria Siomos (GMI, Vienna) with the preparation of the manuscript, and the technical help of Lucia Aronica and Christian Pflügl. This work was supported by grant FWF P17329 to J.L. K.M. is a Junior Group Leader at IMBA, funded by the Austrian Academy of Sciences.
Supplementary material available online at http://jcs.biologists.org/cgi/content/full/121/13/2148/DC1
- Accepted April 14, 2008.
- © The Company of Biologists Limited 2008