ABSTRACT
Meiotic crossovers must be properly patterned to ensure accurate disjunction of homologous chromosomes during meiosis I. Disruption of the spatial distribution of crossovers can lead to nondisjunction, aneuploidy, gamete dysfunction, miscarriage, or birth defects. One of the earliest identified genes involved proper crossover patterning is mei-41, which encodes the Drosophila ortholog of the checkpoint kinase ATR. Although analysis of hypomorphic mutants suggested the existence of crossover patterning defects, it has not been possible to assess these in null mutants because these mutants exhibit maternal-effect embryonic lethality. To overcome this lethality, we expressed wild-type Mei-41 only after the completion of meiotic recombination, allowing embryos to survive. We find that crossovers are decreased more severely in null mutants, to about one third of wild-type levels. Crossover interference, a patterning phenomenon that ensures that crossovers are widely spaced along a chromosome, is eliminated in these mutants. Similarly, crossover assurance, which describes the distribution of crossovers among chromosomes, is lost. Despite the loss of interference and assurance, a third important patterning phenomenon – the centromere effect – remains intact. We propose a model in which the centromere effect is established prior to and independently of interference and assurance.
Introduction
Meiotic crossovers are subject to numerous mechanisms of spatial control to ensure genetic diversity and proper disjunction of homologous chromosomes (Wang et al. 2015). Crossover assurance is the phenomenon in which there is at least one crossover per bivalent, generating the “obligate chiasma” that ensures disjunction (Owen 1949). Crossover interference is the inhibition of crossover formation within intervals flanking sites of crossover precursors (Sturtevant 1913; Berchowitz and Copenhaver 2010). Together with crossover homeostasis, which buffers crossover formation from increases or decreases in potential crossover precursors (Martini et al. 2006), assurance and interference demarcate the minimum and maximum number of crossovers possible per meiosis. Modeling suggests that crossover assurance, interference, and homeostasis are the result of a single patterning process with varying degrees of plasticity depending on the meiotic environment (Wang et al. 2015). However, less is known regarding the centromere effect, a phenomenon wherein crossover formation is suppressed within pericentromic euchromatic regions (Beadle 1932; Mather 1939).
Perturbation of crossover control can be viewed in the context of the two-pathway paradigm, wherein crossovers created within the ‘Class I’ pathway use canonical meiotic proteins that result in crossover patterning characteristic of that species (Kohl and Sekelsky 2013). Alternatively, in some cases mutants that lack one or more of these meiosis-specific proteins default back to a more mitotic-like ‘Class II’ pathway. This switch from Class I to Class II is often associated with a significant reduction in crossover formation and abnormal crossover patterning that results in gamete aneuploidy (Argueso et al. 2004; Lu et al. 2008; Hatkevich et al. 2017). Therefore, teasing apart the relationship between pathway usage and crossover control phenomena may help elucidate evolutionarily conserved mechanisms resulting in proper crossover placement.
The Drosophila mei-41 gene, originally identified in a screen for meiotic mutants in 1972, encodes the ortholog of the DNA damage checkpoint kinase ATR (Baker and Carpenter 1972; Hari et al. 1995). Progeny from mei-411 females exhibited altered crossover distribution, suggesting a critical role for the protein in meiotic crossover patterning (Baker and Carpenter 1972). However, meiotic recombination has not been assayed in the complete absence of Mei-41 because null mutants exhibit maternal-effect embryonic lethality due to DNA replication checkpoint failure: Embryos that lack maternal Mei-41 fail to slow rapid nuclear cycles leading up to the midblastula transition, do not cellularize, and eventually degenerate (Sibon et al. 1999). Thus, alleles used in previous studies of meiotic recombination are either hypomorphic or separation-of-function (Laurençon et al. 2003).
To overcome the embryonic requirement for maternal Mei-41, we expressed Mei-41 under control of a promoter that turns on during oogenesis after recombination has been completed, generating a fertile mei-41 null mutant. Crossover and non-disjunction phenotypes are more severe in the mei 41 null mutant. With regard to crossover patterning, crossover interference and assurance are completely lost in mei-41 null mutants, but the centromere effect remains intact. Intriguingly, progeny from mothers lacking both Mei-41 and the presumed Drosophila Class I Holliday junction resolvase Mei-9 have phenotypes nearly identical to those of mei-41 single mutants, indicating a switch to the Class II pathway following the establishment of the centromere effect. We conclude that the centromere effect is established prior to, or separate from, the essential role of Mei-41 in the Class I crossover pathway and that the centromere effect is achieved independently of interference and assurance.
Materials and Methods
Drosophila stocks
Flies were maintained at 25° on standard medium. To overcome the maternal-effect embryonic lethality of mei-4129D null mutation (Sibon et al. 1999; Laurençon et al. 2003), wild-type genomic mei-41 was cloned into the pPattB, UASp::w vector (courtesy of Steve Rogers) via In-Fusion HD (Takara Bio USA, Inc., Mountain View, CA) and transformed into XL10-Gold ultracompetent cells (Agilent Technologies, Inc., Santa Clara, CA). This construct was injected via phiC31 integrase-mediated trans-genesis into the X chromosome landing site 2A (BestGene Inc., Chino Hills, CA). Integrants were crossed into a P{matα4::GAL4-VP16} background. w mei-4129D/y P{UASp::mei-41} w mei-4129D; matα4::GAL4-VP16/+ was used in all mei-41 null assays.
Double mutant stock creation used the above transgenic rescue in conjunction with appropriate null alleles. The mei41; mei-P22 double mutant genotype was y w mei-4129D/y PUASp::mei-41 w mei-4129D; mei-P22103 st/mei-P22103 BlmD2 Sb P{matα4::GAL4-VP16}. The mei-9 mei-41 double mutant genotype was y mei-9a mei-4129D/y P{UASp::mei-41} w mei-9a mei-4129D; P{ matα4::GAL4-VP16}/+. The mei-41 Blm double mutant genotype was w mei-4129D/y P{UASp::mei-41} w mei-4129D; st BlmD2 ry531 P{matα4::GAL4-VP16}/BlmN1 ry606 Sb P{UASp::Blm}.
Hatch rates
To test P{UASp::mei-41} rescue efficiency, 60 virgin females of appropriate genotypes were crossed to 20 isogenized Oregon-Rm males (courtesy of Scott Hawley). Adults were mated in grape-juice agar cages containing yeast paste for two days prior to collection. Embryos were collected on grape-juice agar plates for five hours and scored for hatching 48 hours later.
Meiotic assays
Meiotic crossovers were quantified by crossing net dppd-ho dp b pr cn / + virgin females of the appropriate mutant background to net dppd-ho dp b pr cn males. All six markers were scored in progeny from each genotype, with the exception mei-41; mei-P22. In that case, 731 XX females were scored for all six markers and 1023 XXY females and XY males were scored for net – b; eye color markers pr and cn were excluded because of the presence of a w mutation in the mothers. These data were pooled for a final n of 1754. Crossover density was calculated using Drosophila melanogaster reference genome release 6.12 with transposable elements excluded, as described in Hatkevich et al. (2017). Complete progeny counts are given in Supplemental Table S1.
Interference was assayed by crossing dp wgSp−1 b / + virgin mei-41 null and wild-type females to net dppd-ho dp b pr cn males. Complete progeny counts are given in Supplemental Table S2.
X nondisjunction was scored by crossing virgin mutant females of the appropriate genotypes to y sc cv v g f / Dp(1;Y)BS males. Exceptional progeny for X nondisjunction events originate from diplo-X and nullo-X ova, resulting in Bar-eyed females (XXY) and wild-type-eyed males (XO), respectively. Counts of scored exceptional progeny were multiplied by two to account for X nondisjunction progeny that do not survive to adulthood (XXX and YO).
mei-41; Blm double mutant morphology analysis
mei-41; Blm double mutant females reach adulthood at less than expected Mendelian ratios relative to sibling classes, scored over a five day period. Double mutants and wild-type virgins were mated to isogenizedw1118 males for two days in vials containing yeast paste, followed by ovary dissection in PBS buffer. Female morphology and ovary defects were photographed using the EOS Rebel T3i (Canon U.S.A., Inc., Long Island, NY) with an MM-SLR Adapter (Martin Microscope Company, Easley, SC).
Data Availability
The authors state that all data necessary for confirming the conclusions presented in the article are represented fully within the article. Drosophila stocks are available upon request.
Results
Post-germarium expression of mei-41 rescues embryoniclethality and creates a meiotic recombination null
Drosophila females homozygous for null mutations in mei-41 produce no viable progeny due to a requirement for maternally-deposited Mei-41 at the midblastula transition, resulting in cleavage-stage arrest (Sibon et al. 1999). Blm null mutants also exhibit maternal-effect embryonic lethality (McVey et al. 2007). To study meiotic recombination in Blm null mutants, Kohl et al. (2012) expressed wild-type Blm under indirect control of the alpha tubulin 67C (matα) promoter via the Gal4-UASp system. This promoter does not express until the early vitellarium (Sanghavi et al. 2013), by which time recombination should be compete. In support of this conclusion, crossover assays on surviving progeny of females with null mutations in Blm give similar results to those from embryos rescued by expressing UASp::Blm with the matα4::GAL4-VP16 driver (McVey et al. 2007; Kohl et al. 2012; Hatkevich et al. 2017).
We used the same system to overcome the maternal-effect lethality in embryos from mei-4129D homozygous null females (see Materials and Methods). To quantify the extent of maternal P{UASp::mei-41} rescue, we compared hatch rates of embryos from wild-type, mei-4129D, and P{UASp::mei-41} mei-4129D with and without P{matα4::GAL4-VP16} (Table 1). Embryos from females homozygous for mei-4129D with or without P}UASp::mei-41} but lacking P{matα4::GAL4-VP16} did not survive to hatching, whereas embryos from females with both components of the Gal4-UASp rescue system had a hatch rate of 52.8%. It is possible that the rescue is complete and that the remaining embryonic lethality is due to aneuploidy resulting from high nondisjunction in mei-41 mutants. Larvae that did hatch survived to adulthood, allowing for analysis of the crossover patterning landscape in a mei-41 null mutant. For simplicity, flies carrying this transgene system are denoted below as mei-4129D or mei-41 null mutants.
Crossover reduction in mei-41 null mutants
Drosophila mei-41 was initially characterized as a meiotic mutant by Baker and Carpenter in 1972 (1972). Hypomorphic mei-41 alleles resulted in an overall 46% decrease relative to wild-type controls, measured across adjacent intervals spanning the entirety of 2L and proximal 2R, about 20% of the euchromatic genome. Progeny from mei-4129D mothers had a significantly more severe phenotype, with a 67% reduction in crossovers summed across this region (Figure 1A). Given the many functions of Mei-41 in mitotically proliferating cells, we wanted to determine whether the remaining crossovers were meiotic or occurred within the pre-meiotic germline. Mei-P22 is the binding partner of Mei-W68, the Drosophila Spo11 ortholog, and is required to generate meiotic DSBs (Liu et al. 2002; Robert et al. 2016). In the absence of Mei-P22, resulting crossovers must be mitotic in origin and occur prior to meiotic recombination. Drosophila males hemizygous for mei-4129D do not display mitotic recombination in the pre-meiotic germline (LaRocque et al. 2007). Likewise, meiotic crossovers were completely abolished in mei-4129D; mei-P22103 double mutants (n = 1754). One vial had two female progeny that appeared to be either a double crossover in the adjacent b – pr and pr – cn regions, gene conversion of the pr mutation, or reversion of this mutation (an insertion of a 412 transposable element). Since these were in the same vial they likely represent a single pre-meiotic event. We conclude that the crossovers observed in the mei-41 null mutant females are meiotic in origin.
Crossover interference and crossover assurance are lost in mei-41 null mutants
Meiotic crossover control includes a phenomenon known as interference, which is a decreased likelihood of having two crossovers close to each other within the same chromosome arm (Sturtevant 1913; Berchowitz and Copenhaver 2010). While the strength of crossover interference differs between organisms, complete interference in Drosophila extends out to about 10 cM (Weinstein 1958). Baker and Carpenter (1972) reported that interference is reduced in mei-41 hypomorphic mutants. We determined the extent of this reduction in null mutants by analyzing two adjacent intervals on 2L (Figure 2A). Single and double crossovers were scored and interference (I) was calculated using the method of Stevens (1936). In Stevens’ definition, I = 1 indicates complete positive interference and I = 0 indicates no interference. Among progeny of wild-type females (n = 3325), there are significantly fewer double crossovers observed (5) than expected (59; p < 0.0001), demonstrating strong interference between these intervals (I = 0.915; Figure 2A). Compared to wild-type, mei-4129D mutants (n = 9740) show a significant reduction in interference (p < 0.0001; Figure 2A), with no significant difference between expected (23) and observed (22) (p = 0.83; I = 0.041; Figure 2A). Based on this, we conclude that interference is completely lost in mei-41 null mutants.
Meiotic crossover reduction in Drosophila leads to an increase in nondisjunction events, often due to a failure to form at least one crossover (known as the obligate chiasma) between homologs (Hawley 1988; Koehler et al. 1996). These observations imply the existence of a mechanistic phenomenon known as crossover assurance, where bivalents must achieve a minimum number of crossovers to create the meiotic spindle tension required for stable homolog orientation at metaphase I (McKim et al. 1993). X chromosome nondisjunction occurs at a frequency of less than 0.1% in wild-type Drosophila, and original hypomorphic alleles of mei-41 cause a significant increase, to 9-10% (Baker and Carpenter 1972). We crossed mei-4129D females to males carrying a dominant Y-linked BS mutation and scored progeny. NDJ was significantly increased relative to the rate in hypomorphic mutants (13.6%; p = 0.0018) (Supplementary Table 3).
Based on the loss of interference, the severe reduction in crossovers per meiosis, and the significant increase in X nondisjunction seen in mei-41 null mutants, we hypothesized that crossover assurance would be severely reduced, if not completely lost. If the reduction in crossovers on 2L is representative of the entire genome, then mei-41 null mutants average less than two crossovers per meiosis. It is therefore not possible to have full assurance, which would require a minimum of three (one per major chromosome) or five (one per major chromosome arm) crossovers. Nonetheless, assurance could manifest as the two crossovers being on different chromosomes more often than expected by chance. This would predict a decrease in double crossovers; since we observed no such decrease (Figure 2A) it suggests that crossover assurance is indeed lost.
We also assessed assurance by comparing the observed and expected frequency of meioses in which there were no crossovers (E0) on 2L. For mei-4129D mutants the expected E0 frequency (0.740, based on Poisson distribution) is similar to the observed E0 frequency (0.720, based on the equations of Weinstein (1936)). This differs significantly compared to wild-type females (p < 0.0001) (Figure 2B). Together, the significant increase in X chromosome nondisjunction, the loss of interference, and the E0 frequency indicate that crossover assurance is lost in mei-4129D mutants.
The centromere effect is retained in mei-41 mutants
A striking observation from both mei-411 and mei-4129D mutants is that despite significant reduction in medial and distal crossovers, a proportional reduction in the number of crossovers within the proximal regions is not seen. As in hypomorphic mutants, crossover reduction in mei-41 null mutants was more severe within the three distal intervals (>70% reduction in each) compared to the two proximal intervals (15% for b to pr and 29% for pr to cn, which spans the pericentric heterochromatin) (Figure 1B). This suggests that the centromere effect on recombination remains intact despite severe reduction or complete loss of crossover interference and assurance. To evaluate this hypothesis in mei-41 null mutants, we calculated CE, a measure of the centromere effect that is similar to I (Hatkevich et al. 2017), in the centromere-spanning pr - cn interval. As in wild-type females (CE = 0.89; (Hatkevich et al. 2017), in mei-4129D mutants there was a significant difference between expected and observed crossovers in this region (p < 0.0001), yielding a CE value of 0.79 (Figure 2C).
Crossovers in mei-41 mutants are not dependent upon the Class I meiotic resolvase Mei 9
The change in crossover distribution seen in mei-41 null mutants differs from patterning seen in Blm mutants, which encodes a helicase required for proper meiotic patterning and recombination through the Class I pathway (Hatkevich et al. 2017). Loss of Blm results in loss of interference, assurance, and the centromere effect. These crossovers are generated in the Class II pathway as they do not use the Class I resolvase Mei 9, suggesting that Blm helicase is required for shuttling meiotic DSBs into the Class I pathway early in the repair pathway, and that there is no patterning in the Class II pathway.
As mei-4129D null mutants lose crossover assurance and interference but retain the centromere effect, we hypothesized that loss of Mei-41 shifts meiotic recombination into the Class II path-way later than loss of Blm, after establishment of the centromere effect. To determine whether crossovers generated in mei-41 null mutants rely upon the Class I resolvase, we generated mei-9a mei-4129D double mutants and analyzed crossover patterning on 2L as described above. Similar to mei-4129D single mutants, double mutants displayed a 66% reduction in crossovers and exhibited a similar same distribution. (Figure 3A & 3B). Consistent with this, nondisjunction frequency in the mei-9 mei-41 double mutant (15.9%, p = 0.3424) was not significantly different to that of the mei-41 single mutant (Supplementary Table 2) and crossover assurance was lost (p = 0.488; Figure 22A). Most importantly, the centromere effect in double mutants (CE = 0.75) was similar to that of mei-41 single mutants, with a significant difference between expected and observed crossovers (p < 0.0001; Figure 2C). We conclude that crossovers generated in mei-41 mutants do not require the Class I Mei-9 resolvase, suggesting that loss of Mei-41 shifts meiotic recombination into the Class II pathway after or separate from the establishment of the centromere effect.
mei-41; Blm double mutants display partial synthetic lethality and low brood size
To further tease apart the establishment of crossover control mechanisms in the context of the two-pathway paradigm, we hypothesized that Blm acts earlier in the recombination pathway than Mei-41, since there is a complete loss of crossover patterning in Blm mutants compared to loss of only interference and assurance in mei-41 mutants. Therefore, mei-41; Blm double mutants should have a phenotype like that of Blm single mutants with respect to crossover patterning. However, while both mei-41 and Blm single mutants are fully viable, we recovered fewer mei-41; Blm double mutant females than expected, suggesting partial synthetic lethality. Of the few obtained, females exhibited underdeveloped abdomens and abnormal tergites (Figure 4A). Most lacked midgut and hindgut structures, displayed ovary epithelial sheaths that contained under-developed ovaries or lacked ovaries entirely, and died within 1-2 days of eclosion. Survivors retained underdeveloped ovaries at six days post-eclosion (Figure 4B), corresponding to a severe reduction in brood size and makinge crossover and nondisjunction analyses impractical. These synthetic phenotypes (reduced viability and developmental defects) are likely the result of a combination of mitotic defects seen in mei-41 and Blm single mutants, which include elevated spontaneous apoptosis (LaRocque et al. 2007; Trowbridge et al. 2007).
Discussion
We have demonstrated that the Gal4-UASp rescue successfully overcomes maternal-effect embryonic lethality of mei-41 mutants, allowing us to perform meiotic crossover patterning analysis in mei-41 null mutants. The crossover reduction in null mutants is more severe than that of the previously reported for hypomophic mutant (Baker and Carpenter 1972). Importantly, we found that crossover interference and assurance are abolished when Mei-41 is absent, yet the centromere effect remains largely intact. Removing the presumptive Class I meiotic resolvase (Mei-9) in a mei-41 null background resulted in phenotypes similar to those of the mei-41 single mutant, suggesting that meiosis in mei41 null mutants relies on alternative endonucleases to resolve recombination intermediates into crossovers. These crossovers might therefore be defined as being made through the Class II pathway (Kohl and Sekelsky 2013).
Crossovers also appear to be made through the Class II pathway in Drosophila Blm mutants (Hatkevich et al. 2017). In these mutants, crossover distribution is more-or-less random along and between chromosomes, perhaps reflecting DSB distribution. Notably, interference, assurance, and the centromere effect are all severely decreased or lost entirely. In contrast, in mei-41 mutants we find that interference and assurance are lost but a strong centromere effect is retained. If crossover interference and crossover assurance result from a single patterning process (Wang et al. 2015), it is not surprising to see loss of both. Retention of a centromere effect, however, suggests a mechanism that is separate from these other patterning phenomena.
The molecular function of Mei-41 that impacts crossover patterning is unknown. One parsimonious interpretation of our results is that the centromere effect is established prior to crossover interference and assurance Figure 5). Blm is required before any of these processes occurs, so loss of Blm results in loss of all three. Mei-41, however, is required later, after the centromere effects has been established but before interference and assurance are achieved. In mice, ATR localizes to unsynapsed chromosome axes (Keegan et al. 1996). Immunolocalization of Mei-41 has not been reported, but in Drosophila synapsis is not dependent on DSBs (McKim et al. 1998), and both synapsis and DSB formation appear to be normal in mei-41 mutants (Carpenter 1979; Joyce et al. 2011). The budding yeast ortholog of ATR, Mec1, is required to ensure inter-homolog bias during meiotic recombination (Grushcow et al. 1999). Use of the sister chromatid could disrupt the Class I pathway and reduce the number of interhomolog crossovers, resulting in part or all of the observed decrease. In mitotic DSB repair, mei-41 mutants have no observable defects in the early steps of SDSA, such as resection, strand invasion, and synthesis, but required for subsequent annealing and ligation (LaRocque et al. 2007). The molecular function of Mei-41 in mitotic DSB repair has not been elucidated, but it has been hypothesized that Mei-41 activates Marcal1, which then catalyzes annealing of complementary sequences (Holsclaw and Sekelsky 2017). It is possible that Mei-41 activates a protein that catalyzes 2nd-end capture during meiosis, and that loss of this activity prevents recombination from proceeding through the Class I crossover pathway. Assays for sister chromatid exchange and analyses of heteroduplex DNA may shed additional light on the mechanistic role(s) of Mei-41 during meiosis.
Perhaps the most interesting outcome of this work is that the centromere effect is established separately from crossover interference and crossover assurance. The beam-flm model postulates that the establishment of crossover interference relies on relief of axial mechanical stress outwards from sites of crossover designation, with crossover homeostasis buffering this spreading inhibitory region based on crossover precursor density (Kleckner et al. 2004; Zhang et al. 2014). Crossover assurance is therefore observable as the passive byproduct of proper crossover-designation resolution, suggesting that interference, assurance, and homeostasis result from a single patterning process (Zhang et al. 2014; Wang et al. 2015). We suggest that the centromere effect is achieved through an independent process. Whole-genome sequencing reveals that noncrossover gene events are as common in proximal regions as in other regions, suggesting that DSBs are made throughout the euchromatin (Comeron et al. 2012; Miller et al. 2016). This implies that the centromere effect is achieved by directing repair of proximal DSBs preferentially into noncrossover pathways or preventing them from entering the crossover pathway. Analysis of recombination in triploid females suggests that the centromere effect might be sensitive to the number of centromeres (Redfield 1932; Sturtevant 1951; Hartmann and Sekelsky 2017), but there is really nothing else known about mechanism. Additional experiments to understand the function of Mei-41 in meiotic recombination may also provide insights into the mechanism of the centromere effect.
Acknowledgements
We thank Nicole Crown, Michaelyn Hartmann, and Talia Hatkevich for comments on the manuscript. This work was supported by a grant from the National Institute of General Medical Sciences (NIGMS) to JS under award 1R35GM118127.