Abstract
Although in Saccharomyces cerevisiae the initiation of meiotic recombination, as indicated by double-strand break formation, appears to be functionally linked to the initiation of synapsis, meiotic chromosome synapsis in Drosophila females occurs in the absence of meiotic exchange. Electron microscopy of oocytes from females homozygous for either of two meiotic mutants (mei-W68 and mei-P22), which eliminate both meiotic crossing over and gene conversion, revealed normal synaptonemal complex formation. Thus, synapsis in Drosophila is independent of meiotic recombination, consistent with a model in which synapsis is required for the initiation of meiotic recombination. Furthermore, the basic processes of early meiosis may have different functional or temporal relations, or both, in yeast andDrosophila.
In the classical view of meiosis, homologous chromosome synapsis, as indicated by the formation of an elaborate ribbonlike structure called the synaptonemal complex (SC), was thought to be the first and primary event of meiotic prophase, essential for the initiation of meiotic recombination (1). Studies in Saccharomyces cerevisiae, however, have created a different view of the meiotic process in which the initiation of recombination, as evidenced by a double-strand break (DSB), precedes the initiation of synapsis (2,3). Three lines of evidence support this view of early meiotic prophase in yeast. First, the initiating event of meiotic recombination, the formation of a DSB, appears before SC formation (4). Second, meiotic mutants that either fail to create DSBs or to process DSBs to make single-stranded tails prevent the formation of a mature SC (2). Third, some mutants allow high levels of meiotic recombination but prevent the production of a mature SC (5). These data are consistent with a model in which single-stranded DNA generated by a DSB carries out a homology search required for synapsis and SC formation. In contrast, synapsis is not an absolute prerequisite for either the initiation (6) or completion of meiotic recombination (7).
To assess the relation between synapsis and the initiation of recombination in Drosophila oocytes, we examined both recombination and SC formation in oocytes homozygous for either of two null-recombination mutations. The mei-W68 andmei-P22 (8) mutants prevent the initiation of meiotic recombination as defined by four independent assays: (i) reduction or elimination of meiotic gene conversion; (ii) elimination of meiotic crossing over, as assayed by measuring either intragenic crossing over or the frequency of meiotic crossing over along entire chromosome arms; (iii) lack of double-strand DNA breaks that persist into metaphase or anaphase I; and (iv) failure to produce either early or late recombination nodules (RNs).
To assay the effects of the mei-W68 andmei-P22 mutations on meiotic crossing over, we examined intragenic recombination at the rosy locus (9). No gene conversion events or intragenic crossovers were observed among the progeny of mei-W68 or mei-P22 females (Table1 and Fig.1). Compared to controls, the frequency of both intragenic exchange and simple gene conversion was reduced by a factor of at least 30 to 40 and, most likely, was eliminated. A small reduction in gene conversion frequency was also observed in themei-W68/+ females, suggesting a dosage effect.
Gene conversion at the rosy locus (genetic map position 52.0). A schematic of the parental chromosomes and the three products of intragenic recombination that yieldrosy + recombinants. The distances between the loci are not drawn to scale. The recombinants were classified as convertants or crossovers on the basis of the flanking mutations. In the mei-P22 experiment, the flanking markers werekar, an eye color mutant mapping 0.3 cM to the left, andcv-c, a wing vein mutant mapping 2.1 cM to the right ofry. In the mei-W68 experiment, Ace(52.5) or red (53.6) replaced cv-c(54.1). The two rosy alleles used in this study, ry 531 andry 606, were chosen because they are at opposite ends of the rosy gene, 3780 nucleotides and 0.012 cM apart (29). The average length of conversion tracts in Drosophila is 885 base pairs (32), and therefore co-conversion events are expected to have a minimal effect on our experiments.
Intragenic recombination at the rosylocus in mei-P22, mei-W68, and control females.
The effect of these mutations on crossing over was also assayed by more conventional means. In both mei-W68 andmei-P22 mutant females, the frequency of crossing over along the entire X and second chromosomes was reduced to less than 0.5% of normal. Moreover, the few crossover events that were observed tended to be recovered in clusters of identical recombinants among the progeny of single females, suggesting that they resulted from mitotic and not meiotic recombination events (10). The failure to observe meiotic recombination events in progeny of mei-W68and mei-P22 mutant oocytes is consistent with our failure to observe either early or late RNs in the oocytes of these females [(11) and see below].
Recombination events may be initiated in mei-P22 andmei-W68 oocytes, but then redirected into a pathway that results in sister-chromatid exchanges. We tested this possibility by examining the effects of these mutations on sister-chromatid exchange assayed by ring-chromosome loss. Sister-chromatid exchange within a ring chromosome results in the formation of a dicentric chromosome that is not transmissible. No excess of ring loss was observed inmei-P22 females compared with +/+ or+/mei-P22 sisters (12).
The above data demonstrate that meiotic recombination events involving either homologous chromosomes or sister chromatids do not occur in mei-W68 and mei-P22 mutant oocytes. The ablation of gene conversion further demonstrates that mature recombination intermediates are also not formed. Based on three lines of evidence we can also rule out the possibility that DSBs are created in mei-W68 and mei-P22 mutant oocytes but left unrepaired. First, cytological analysis of meiotic chromosomes during metaphase and anaphase I from both mei-W68 andmei-P22 mutant oocytes showed no evidence of chromosome fragmentation or of gapped chromosomes (13), as would be expected if the DSBs were not repaired (14). Second, there was no excess of nullo-X oocytes relative to diplo-X oocytes among the progeny of either type of mutant female, and thus chromosome loss is not common (15). Finally, these females were no less fertile than expected, on the basis of expected frequencies of zygotic death due to aneuploidy, suggesting that there was no excessive dominant lethality from broken chromosomes in the oocytes. Although we cannot rule out the possibility that DSBs are made but are rapidly repaired in a manner that leaves no genetic trace, all available data argue that recombination is not initiated in these mutants.
Saccharomyces cerevisiae mutants exhibiting recombination defects as severe as those observed here do not make SC. To compare the effects of such a recombination defect in Drosophilaoocytes, we examined SC formation in mei-W68 andmei-P22 oocytes. SC formation was analyzed by the reconstruction of serial sections examined by transmission electron microscopy. All of the pachytene nuclei from a singlemei-W68 germarium plus several nuclei from two other germaria were reconstructed (Fig. 2A) (16). Three mei-P22 pachytene nuclei, two from a single germarium and one from another germarium, were completely reconstructed and 10 nuclei from several germaria were thoroughly examined (Figs. 2B and 3). Formation of the central and lateral elements and transverse filaments of the synaptonemal complex was normal in both mei-W68 and mei-P22 females. The width of the central region in euchromatic SC for bothmei-W68 and mei-P22 was equal to that of wild-type (109 ± 8 nm) (17). SC is continuous along each of the X, 2R, 2L, 3R,3L, and 4th chromosome arms (Fig.3), and the age-adjusted euchromatic lengths of the SCs for the X and 2L are equal to those of wild type. There was no evidence for failed or nonhomologous synapsis. Consistent with the recombination phenotype, however, we failed to observe either early or late RNs in the mutants (11).
Transmission electron microscopy of ovaries from females homozygous for mei-W68 andmei-P22. (A) Section from a mei-W68female (16) and (B) a section from amei-P22 female (33). The complete SC is visible in either mutant, including lateral elements (le), transverse elements (te), and central elements (ce). Bar, 100 nm.
(A to D) Consecutive sections from mei-P22 females. (E) A tracing generated by superimposing all the sections from this nucleus, with each chromosome arm represented by a different color.
Meiotic progression in these mutants also appeared to be normal. We observed the same progression of SC shortening and thickening as in wild type. Three mei-W68 germaria were examined in detail (16), and in all respects the timing of developmental events was indistinguishable from that in wild type (17). In particular, there was no delay in reaching full pachytene. Inmei-W68 germaria there was an average of 1.5 ± 1.7 16-cell cysts in pre-pachytene (16), compared with 1.3 ± 0.9 and 2.0 ± 1.8 in two different wild-type samples (18).
We conclude that, in Drosophila melanogaster oocytes, SC can form in the absence of detectable recombination events. Moreover, classical cytogenetic studies argue that inDrosophila oocytes it is synapsis that is required to initiate recombination, and not vice versa. For example, in D. melanogaster, Caenorhabditis elegans, maize and a variety of other organisms, heterozygosity for a translocation breakpoint substantially reduces crossing over for large regions surrounding that breakpoint (19). In yeast, however, similar rearrangements have little or no effect on the frequency of meiotic recombination in the vicinity of the breakpoint (20). Similarly, although small regions of ectopic DNA undergo recombination with their normally located homologous regions in yeast (7), the same is not true in Drosophila (21). Even large translocated regions only very rarely recombine or synapse (or both) with homologous intervals in Drosophila(22), worms (23), and mice (24).
We interpret these data to mean that large-scale synapsis is required to initiate meiotic recombination in Drosophila. Meiotic chromosome synapsis in yeast, which appears to require only a homology search done at a “gene-by-gene” level (25), may be representative of organisms with relatively small and noncomplex genomes. Indeed, such a strategy might well be disastrous in organisms with large amounts of dispersed repetitive DNA or with large dispersed gene families. We imagine that in higher organisms synapsis is required before the initiation of exchange both to prevent recombination events between homologous DNA sequences on nonhomologous chromosomes and to facilitate the nonrandom positioning of exchanges along the arms of meiotic chromosomes.
