Abstract
In Drosophila, meiotic recombination is initiated by the formation of programmed DNA double-strand breaks (DSBs), which occur within the context of the synaptonemal complex (SC). To better understand the role of the SC in mediating recombination we created an in-frame deletion mutant in c(3)G (deleting amino acids L340 to N550, denoted as c(3)GccΔ1), which encodes the major transverse filament protein of the SC. Although c(3)GccΔ1 oocytes assemble ribbon-like SC and exhibit normal DSB formation, the euchromatic SC precociously disassembles into fragments that persist until mid to late pachytene in both c(3)GccΔ1 heterozygotes and homozygotes. Centromeric SC, however, is unaffected in both genotypes. Thus, c(3)GccΔ1 is a separation-of-function mutant that establishes different functional and structural requirements between euchromatic and centromeric SC. Our data also demonstrate that the chromosome arms differ in their sensitivity to c(3)GccΔ1-induced perturbations in the SC. The X chromosome is distinctly sensitive to these perturbations, such that euchromatic pairing and crossing over are altered in c(3)GccΔ1 heterozygotes and severely reduced in c(3)GccΔ1 homozygotes. On the autosomes, crossovers are shifted to centromere-proximal regions and crossover interference is defective in both c(3)GccΔ1 homozygotes and heterozygotes. However, only c(3)GccΔ1 homozygotes display a progressive loss of euchromatic pairing in distal autosomal regions, suggesting that discontinuity in the euchromatic SC—rather than failed pairing—might cause the altered crossover distribution. These phenotypes reveal that different chromatin states or regions have differing requirements to maintain both the SC and homologous pairing. Furthermore, c(3)GccΔ1 is the first mutant in Drosophila to demonstrate that the SC appears to facilitate the regulation of recombination frequency and distribution differently on each chromosome.
Author Summary Chromosome segregation errors during meiosis are the leading cause of miscarriage and birth defects in humans. To prevent these errors from occurring, meiotic cells have evolved multiple mechanisms to ensure that each gamete receives exactly half the number of chromosomes. During meiosis I, this is accomplished by forming a crossover between homologous chromosomes, which is facilitated by a large protein complex called the synaptonemal complex (SC). The SC is assembled between homologous chromosomes during early prophase I, and it is unclear how the SC regulates the position and number of crossovers each homolog receives. To better understand the role of the SC in mediating recombination, we created an in-frame deletion mutant in Drosophila melanogaster in the gene encoding the C(3)G protein, the major transverse filament protein of the SC. Although mutant oocytes assemble ribbon-like SC, the SC along the chromosome arms precociously disassembles in early meiosis. Surprisingly, the SC around the centromeres is unaffected in these mutants, suggesting that the requirements for SC formation may differ depending on where the SC is located along the chromosomes. Our data also demonstrate that the chromosome arms differ in their sensitivity to the mutant-induced perturbations of the SC in both crossing over and homolog pairing.
INTRODUCTION
At the beginning of meiotic prophase, homologous chromosomes must identify each other, pair, and initiate recombination. Meiotic recombination is critical for the proper segregation of homologous chromosomes during meiosis I. The failure to properly exchange genetic information will frequently result in missegregation of the chromosomes leading to eggs and sperm with the incorrect number of chromosomes. Indeed, errors in meiotic chromosome segregation are the leading cause of miscarriage and aneuploidy in humans, which can result in chromosomal disorders such as Down syndrome and Turner syndrome (reviewed in [1]).
To initiate recombination, a series of programmed DNA double-strand breaks (DSBs) are formed by the topoisomerase-like enzyme Spo11 [2]. Most organisms make a large excess of DSBs—in some cases 30 times more than necessary—since only a small fraction of DSBs will be repaired into crossovers [reviewed in 3]. Some organisms appear to use DSBs to identify homology, and it is thought that creating more DSBs than necessary helps to ensure that homologs are able to properly pair and that each chromosome receives at least one crossover [reviewed in 4].
Neither DSBs nor crossovers are randomly distributed along chromosome arms, but rather seem to be placed in a fairly ordered manner. Fine-scale genomic analysis of the nonrandom distribution of DSBs in mice and yeast has indicated that the location of DSBs involves a complicated series of factors working together on both a local DNA sequence level and on a global chromatin architecture level [5, 6]. Moreover, the nonrandom distribution of the resulting crossovers varies depending on the organism. Some organisms, like Drosophila, place most of their crossovers in the middle of the chromosome arms, while other organisms, like grasshoppers, planarians and nematodes, position crossovers at the distal regions of chromosome arms [reviewed in 7] [8]. This suggests that there may be multiple mechanisms that can determine the position of DSBs along the chromosome arms and that only certain DSBs can become crossovers.
In Drosophila, approximately one crossover forms per chromosome arm [9, 10]. The tight regulation of crossover number is controlled by multiple processes, including crossover interference, the centromere effect, and region-specific constraints [reviewed in 11]. Once a crossover has formed, crossover interference functions to suppress the formation of additional crossovers nearby. Similarly, the centromere effect and regional constraints function to prevent the formation of crossovers in or next to specific chromatin states, such as the centromere and heterochromatin. All of these processes function over megabases of DNA to constrain the crossover number in each meiosis to an average of six crossovers across the three crossover-competent chromosomes in Drosophila. Exactly how interference, the centromere effect, and region-specific constraints function is not well understood.
Double-strand break formation and crossing over in Drosophila occur within the context of full-length synaptonemal complex (SC) (Fig 1A) [15, 16]. The SC, a large protein complex that assembles between paired homologous chromosomes, is required for the formation of crossovers. It is therefore possible that the SC may play a role in regulating the crossover fate of DSBs. Additionally, the SC connects two homologs along their entire lengths, making it an ideal candidate structure for monitoring DSB formation and crossover maturation occurring at different places along the chromosomes. Indeed, several studies have strongly suggested the SC plays a role in mediating crossover interference [17, 18].
The synaptonemal complex is arranged in three parts: lateral elements (LEs), a central region (CR), and a central element (CE) (Fig 1B). The LEs assemble on each homolog alongside the axis components, which consist of cohesin and cohesin-like proteins that are thought to establish the meiotic chromatin architecture [reviewed in 19, 20]. The LEs interact with the CR proteins, which occupy the space between the two homologs. Transverse filament proteins span the distance of the CR, thereby connecting the LEs of each homolog. In the middle of the CR is the CE, which consists of proteins that are thought to help stabilize the transverse filaments.
Genetic analyses have identified multiple components of the Drosophila SC, and recent advances in superresolution microscopy have allowed SC components to be precisely positioned within the overall SC structure (Fig 1B). The Drosophila SC forms two layers that are mirror images of one another [21]. C(3)G, the major transverse filament protein of the Drosophila SC [22], forms a homodimer, with the C-terminal ends of each subunit positioned in the LE and the N-terminal ends in the CE [23]. The N terminus of one homodimer is then thought to interact with the N terminus of another homodimer to span the distance between the two homologs, thus connecting the two homologous LEs [23]. It is currently unclear exactly how C(3)G interacts with the LE. An additional CR protein, Corolla, was recently identified [24]. Corolla, resides within the CR as two parallel tracks and has been shown by yeast two-hybrid analysis to interact with the CE protein Corona (CONA) [21, 24, 25], which also forms two parallel tracks within the CE [21, 26]. C(3)G, Corolla, and CONA depend on each other in order to assemble the SC, and all three proteins are required for the formation of crossovers [22, 24-26].
The first step toward dissecting the functional anatomy of the C(3)G protein involved creating in-frame deletion mutations, which showed that the N- and C-terminal globular domains of C(3)G are required to properly assemble the SC and for robust crossover formation [27]. Here, we describe the effects of a mutant, c(3)GccΔ1, with an in-frame deletion of a coiled-coil domain. To properly characterize this mutant, we created a CRISPR/Cas9 allele, eliminating any of the common concerns regarding levels of expression that are inherent when using a randomly inserted transgene. (A transgene in-frame deletion mutant, similar to the one in this manuscript, has been previously created, but was never extensively analyzed [22]).
c(3)GccΔ1 removes 211 amino acids from the largest coil-coiled domain of C(3)G which, as a homozygote, shortens the width of the SC by ∼40 nm and thus reduces the distance between the homologs. Both heterozygotes and homozygotes of c(3)GccΔ1 are able to assemble the SC, display normal DSB formation, and affect neither the persistence of SC at the centromere nor centromere clustering. However, euchromatic SC disassembles prematurely in both genotypes, resulting in fragmented or discontinuous SC along the length of the chromosomes. This premature SC disassembly affects both homolog pairing and recombination, but to different degrees on each chromosome. c(3)GccΔ1 is the first Drosophila mutant to illustrate a chromosome-specific regulation of recombination, where each chromosome has different requirements for the SC in maintaining homolog pairing and facilitating recombination.
RESULTS
Novel loss-of-function allele of c(3)G affects the width of the SC
A large deletion in the yeast transverse filament protein Zip1 decreases the distance between the LEs, suggesting that Zip1 is directly involved in determining the width of the SC [28]. Since Zip1 and C(3)G are functional homologs of each other, we postulated that a similar deletion in C(3)G may alter the Drosophila SC in a similar manner to the Zip1 deletion in yeast. To that end, we created a CRISPR/Cas9-mediated in-frame deletion in the Drosophila c(3)G gene (see Methods) that removes 211 amino acids from the second coiled-coil domain in C(3)G (Fig 1B). We call this mutant c(3)GccΔ1.
Based on the 1.485-angstrom size of an amino acid in a coiled-coil, the mathematically calculated decrease in SC width expected by deleting 211 amino acids from a coiled-coil configuration is ∼63 nm. However, when using single-molecule superresolution techniques such as stochastic optical reconstruction microscopy (STORM), which require primary and secondary antibodies, the size of an antibody complex (20–30 nm) will introduce uncertainty into the precise positioning of the structure being imaged. To account for this, we added the size of the primary and secondary antibody complex, which recognizes the C-terminus C(3)G, to the calculated decrease in SC width resulting in an expected decrease of ∼43–53 nm in c(3)GccΔ1 homozygotes.
Using STORM on intact germaria with an antibody that recognizes the C-terminal end of C(3)G, we observed two tracks of the C(3)G C-termini, suggesting that the SC was able to assemble both C(3)G homodimers (Fig 2). Additionally, when we measured the distance between these two tracks, referred to as the SC width, it was decreased by ∼40 nm in c(3)GccΔ1 homozygotes (Fig 2; wild type, 119 nm ± SE 1.0; c(3)GccΔ1, 78.6 nm ± SE 0.3; P<0.0001), which closely matched the expected decrease in SC width. This suggests that, like Zip1, C(3)G is responsible for determining the distance between the LEs in Drosophila.
STORM analysis of c(3)GccΔ1 heterozygotes, in which flies express one c(3)GccΔ1 version and one wild type (WT) version of C(3)G, showed an intermediate change in SC width (Fig 2; 107 nm ± SE 0.4; P<0.0001 when compared to WT). This suggests that c(3)GccΔ1 heterozygotes incorporate a combination of c(3)GccΔ1 and WT versions of C(3)G into the SC. One possible way to accomplish this heterogeneity would be to alternate the C(3)G tetramers such that every other tetramer is the c(3)GccΔ1 mutant. In this case, we would expect the profiles on either C-terminal side of C(3)G to be thicker. However, this is not likely to be the case because the thicknesses of the C-terminal C(3)G signal are nearly the same for the c(3)GccΔ1 heterozygotes and homozygotes (WT: 62 nm; c(3)GccΔ1 heterozygote: 49 nm; c(3)GccΔ1 homozygotes: 43 nm). It is also possible that c(3)GccΔ1 and WT C(3)G monomers are able to dimerize and create mixed C(3)G homodimers, or that the N termini of c(3)GccΔ1 homodimers can interact with the N termini of WT homodimers in the CE of the SC to create a mixed tetramer. Both possibilities would result in the intermediate SC width seen in c(3)GccΔ1 heterozygotes. Given the heterogeneous nature of STORM data, it is impossible to distinguish between these possibilities. Further studies with alternative methodologies will be needed to gain a more detailed understanding of this ultrastructure.
Both c(3)GccΔ1 heterozygotes and homozygotes initiate SC assembly at the centromeres during the pre-meiotic mitotic divisions (region 1) and at the onset of meiosis in zygotene/early pachytene (region 2A) assembled ribbon-like SC with normal kinetics (Fig 3). Moreover, further analysis of the SC in region 2A by measuring the intensity of the antibody staining of Corolla between the mutants and WT showed that although the SC appeared to be ribbon-like in the mutants the intensity was decreased compared to WT (Fig S1). This suggested either that both c(3)GccΔ1 heterozygotes and homozygotes may not be assembling as much SC as in WT or that both mutants have altered the tripartite structure of the SC. However, based on our ability to resolve two C(3)G C-terminal tracks in the STORM data and on Corolla being able to assemble into ribbon-like SC in the deconvolution microscopy data (Fig 3, Figure S1), we do not believe that the decrease in SC intensity indicates a lack of tripartite SC. All of the known components of the SC in Drosophila are co-dependent on each other to be able to assemble, thus if the structure of the SC was failing in these mutants then we would expect both C(3)G and Corolla to be unable to assemble into ribbon-like SC [24, 25].
c(3)GccΔ1 affects the persistence of SC along the chromosome arms but not at the centromere
Previous studies have shown that c(3)G null mutants initiate DSBs, but at reduced levels compared to WT [16, 29]. This suggests that some DSBs will only form in the presence of SC. In c(3)GccΔ1 heterozygotes and homozygotes, DSB formation appears to be normal. To identify DSBs, we used an antibody that recognizes the phosphorylated form of the histone variant H2AV (referred to as γH2AV), and this phosphorylation mark is one of the first events that occurs following the formation of DSBs [30]. At zygotene/early pachytene (region 2A) all 16-cells within the cyst will induce DSBs, and for this assay we scored only DSBs in the meiotic cells, were identified by the presence of the SC [16]. Nurse cells also induce DSBs, but they do not assemble the SC and were excluded from this analysis [16].
In an otherwise WT background, both c(3)GccΔ1 heterozygotes and homozygotes displayed near WT levels of γH2AV foci in zygotene/early pachytene (region 2A) (Fig S2; WT= average 6 ± SE 0.6; c(3)GccΔ1 homozygotes= average 6 ± SE 0.6, P=0.77; c(3)GccΔ1 heterozygotes= average 7 ± SE 0.8, P=0.69). By early pachytene (region 2B), the number of γH2AV foci was reduced to ∼1 focus in WT (average 1 ± SE 0.3) and c(3)GccΔ1 heterozygotes (average 1 ± SE 0.2, P=0.06). c(3)GccΔ1 homozygotes displayed a wider range of γH2AV foci with an average of 3 DSBs (± SE 0.5) in early pachytene (region 2B). This was statistically different from WT (P=0.036), but it is unclear if this indicates a delay in DSB repair or the initiation of more DSBs. By mid pachytene (region 3) all γH2AV foci were absent in both c(3)GccΔ1 homozygotes (P=0.54) and heterozygotes (P=0.17). Thus, while c(3)GccΔ1 homozygotes may have some anomalies in DSB repair or late-DSB initiation in early pachytene, by mid pachytene the repair of these breaks follows similar to WT timing.
In WT, euchromatic SC initiates disassembly in mid/late pachytene (stages 5–7) and the SC is fully disassembled from the chromosome arms by stages 8–9. Interestingly, both c(3)GccΔ1 homozygotes and heterozygotes initiated premature disassembly of euchromatic SC as cells progressed into early and mid pachytene (Fig 3). c(3)GccΔ1 homozygotes began to initiate euchromatic SC disassembly in early pachytene (region 2B), whereas c(3)GccΔ1 heterozygotes started disassembling euchromatic SC in mid pachytene (region 3). Furthermore, this pattern was recapitulated when the antibody intensity of Corolla was quantified in each mutant at each stage of meiosis (Fig S1). Thus, the presence of one WT copy of C(3)G in a c(3)GccΔ1 heterozygote delays the onset of premature SC disassembly but cannot fully rescue the defect, demonstrating the semi-dominant nature of the c(3)GccΔ1 mutant. In both homozygotes and heterozygotes, the fragmented SC persisted until stages 5–7 (mid/late pachytene) and was fully disassembled by stages 8–9, similar to when euchromatic SC is normally disassembled from the chromosome arms (Fig 3; Fig S1). Surprisingly, c(3)GccΔ1 mutants were able to maintain SC at the centromeres, suggesting that the SC at the centromere is unaffected by this deletion.
In Drosophila, centromeres begin to homologously pair and initiate the assembly of centromeric SC during the four premeiotic mitotic divisions (region 1) (Fig 1A, Fig 3)[13]. At the transition from the premeiotic divisions to meiotic prophase (region 2A) the paired centromeres cluster together into 1–3 masses, and this process is dependent on SC assembly [12, 14]. Since c(3)GccΔ1 homozygotes and heterozygotes were able to maintain the SC at the centromeres with WT kinetics, we assayed centromere clustering to determine if this SC was functional. To do this, we counted the number of centromere foci in nuclei containing SC using the centromeric nucleosome Centromere identifier (CID) as a marker for the centromeres. In all of the stages analyzed, centromere clustering was unaffected by the presence of c(3)GccΔ1 (Fig S3, Table S1). Also, c(3)GccΔ1 did not display a dominant effect on centromere clustering when assayed as a heterozygote (Fig S3, Table S1). For c(3)GccΔ1 homozygotes, we do observe an increase in the number of nuclei with 3 centromeric foci in early pachytene. However, in the c(3)G68 null mutant, most of the nuclei have 4 or more CID foci at this stage, which does not occur in the c(3)GccΔ1 mutants [14]. Thus, these data demonstrate that centromeric SC is not dependent on the deleted amino acids in c(3)GccΔ1 to function in clustering centromeres.
c(3)GccΔ1 causes defects in both recombination and euchromatic pairing on the X chromosome
Previous studies have shown that in most (but not all) organisms the SC is required for the formation of crossovers [reviewed in 19, 31]. In Drosophila, Gowen showed that flies heterozygous for the c(3)G null mutation are indistinguishable from WT with respect to recombination on all three crossover-competent chromosomes, demonstrating that one copy of C(3)G is sufficient for normal recombination [32]. Moreover, flies homozygous for the c(3)G null mutation are completely unable to repair DSBs into either crossovers or noncrossover gene conversions on all three crossover competent chromosomes [16, 22, 32-34]. Therefore, the question of how these breaks are repaired remains open [34].
Unlike c(3)G null mutants, c(3)GccΔ1 does initially assemble ribbon-like SC, which prematurely disassembles into fragments as the cells progress through early to mid pachytene. To determine if the SC assembled in c(3)GccΔ1-bearing oocytes is able to promote crossing over, we assayed for the presence of crossovers in both c(3)GccΔ1 homozygotes and heterozygotes. Interestingly, c(3)GccΔ1 homozygotes and heterozygotes displayed different phenotypes with regard to recombination on the X chromosome (Fig 4A; Table 1). Both genotypes were able to from at least some crossovers, suggesting that the region deleted in c(3)GccΔ1 is not required to promote crossover formation per se. However, the crossovers made in c(3)GccΔ1 heterozygotes displayed an altered distribution along the length of the X chromosome, while c(3)GccΔ1 homozygotes severely reduced the frequency of crossing over on the X.
In WT, the majority of crossovers are known to occur in the middle one-third of the chromosome arms and are inhibited near the centromeres and telomeres. Indeed, in the WT control for our X chromosome recombination assay, most of the crossovers occurred between the markers vermillion (v) and forked (f), which reside in the medial one-third of the X chromosome (Fig 4A; Table 1 and S2). However, in c(3)GccΔ1 heterozygotes, the crossover distribution was shifted toward the centromere-proximal region of the chromosome, such that most crossovers occurred between the forked and yellow+ (y+) markers that encompass the centromere. This crossover distribution shift resulted in a corresponding decrease in crossovers in the distal chromosome regions, between the markers scute (sc) and crossveinless (cv). Also, the overall map length of the X chromosome was slightly decreased in c(3)GccΔ1 heterozygotes (51.2 cM in c(3)GccΔ1 heterozygotes vs. 63 cM in WT; Table 1). Thus, c(3)GccΔ1 heterozygotes appear to alter the distribution of crossovers, without greatly changing the total number of crossovers on the X.
In contrast, c(3)GccΔ1 homozygotes severely reduced crossing over on the X chromosome (Fig 4A; Table 1). The total map length of the X chromosome decreased by more than 80%, to 11.8 cM from 63.0 cM in WT, with the majority of the reduction in the medial and distal intervals. However, even with this reduction the distribution of crossovers seems to display the same centromere-proximal shift as in c(3)GccΔ1 heterozygotes.
A more comprehensive approach for understanding these reductions in exchange lies in the application of Weinstein’s method to calculate the frequency of bivalents that underwent zero (E0), one (E1), or two (E2) crossover events [36]. This method accounts for the fact that a single recombination event between two homologs will produce two single crossover and two noncrossover chromatids (only one of which is recovered) and that a double crossover event (depending on whether it is a two-strand, a three-strand, or a four-strand double) will produce some combination of noncrossover, single crossover, or double crossover chromatids (again, only one of which is recovered). In WT, the frequency of noncrossover bivalents (or E0 bivalents) was 18.9%, while in c(3)GccΔ1 homozygotes it was 79.2%—a four-fold increase in the frequency of an oocyte receiving a noncrossover bivalent (Table 1). The remaining ∼20% of X chromosomal bivalents in c(3)GccΔ1 homozygotes did undergo a single crossover event, creating single crossover (or E1) bivalents and the vast majority of these single crossovers occurred in centromere-proximal intervals (Fig 4A).
The presence of SC between homologous chromosomes is thought to keep them paired throughout pachytene, which promotes crossing over between two homologous chromosomes. Since the SC is prematurely disassembled in c(3)GccΔ1, it is possible that homologous chromosome pairing may be lost before crossovers can be established, which may explain the altered crossover distribution. We analyzed homolog pairing using euchromatic FISH probes at distal (3C3–3C7) and proximal (15C1–15D6) regions on the X chromosome (Fig 4B; Table S2 and S3). A locus was considered paired if the distance between the two FISH probe foci was less than 0.75 µm and unpaired if foci exhibited a distance 0.75 µm or greater.
In c(3)GccΔ1 heterozygotes, both distal and proximal probes were paired in nearly 100% of the oocytes in zygotene/early pachytene (region 2A) (Fig 4B, Table S3). However, the distal probe was progressively less paired as the cells progressed to early pachytene (region 2B) and mid pachytene (region 3). The proximal probe remained nearly 100% paired throughout these regions. Thus, the centromere-proximal shift in recombination was correlated with a loss of distal euchromatic pairing and a maintenance of proximal euchromatic pairing on the X chromosome in c(3)GccΔ1 heterozygotes.
Unlike c(3)GccΔ1 heterozygotes, c(3)GccΔ1 homozygotes displayed a progressive loss of euchromatic pairing at both proximal and distal loci on the X chromosome, suggesting that the X chromosomes may be unpaired by mid pachytene (Fig 4B, Table S3). Based on our FISH assay, the proximal locus was able to maintain pairing more effectively than the locus detected by the distal probe, which may explain why the few crossovers observed in c(3)GccΔ1 homozygotes occur in the centromere proximal region of the X chromosome. Indeed, we propose that the loss of homolog pairing caused by a progressive loss of X chromosome SC is the reason for the significant decrease in X recombination observed in c(3)GccΔ1 homozygotes. If we are correct in this asserion, then the observation that homolog pairing is not drastically reduced in c(3)GccΔ1 homozygotes until early pachytene (region 2B) suggests that crossover designation on at least the X chromosome may be occurring either multiple times throughout pachytene or later in pachytene than was previously thought.
Autosomal crossover distribution is shifted to centromere-proximal regions
Meiotic mutants that decrease recombination on the X chromosome also display similar decreases in recombination on the autosomes [22, 27, 29, 37, 38]. It was therefore surprising that both c(3)GccΔ1 homozygotes and heterozygotes displayed robust recombination on the autosomes with the map lengths of each autosome increasing by more than 10 map units, suggesting that each chromosome received more crossovers than normally would occur in WT (Fig 4C,E; Tables 2 and 3). Indeed, to our knowledge, c(3)GccΔ1 is the first mutant in Drosophila to exhibit such dramatic differences in terms of its effects on recombination on the X chromosome compared to the autosomes. Moreover, the biological significance of finding that the X chromosome may be more sensitive to c(3)GccΔ1-induced changes in the SC than the autosomes may be that the various chromosome arms differ in their requirements for proper SC structure and function.
Although overall recombination on the autosomes was robust, the crossover distribution was shifted to the centromere-proximal regions in both c(3)GccΔ1 homozygotes and heterozygotes. On the 3rd chromosome, we observed a slight decrease in recombination in the distal interval [roughoid–hairy (ru–h)] of both homozygotes and heterozygotes (Fig 4E; Table 3), while on the 2nd chromosome, only c(3)GccΔ1 homozygotes displayed a decrease in the distal interval net-decapentaplegic (net–dpp) (Fig 4C; Table 2).
However, the most striking effects of the c(3)GccΔ1 mutants on autosomal crossover distribution can be seen by considering the intervals that span their centromeres (Fig 4C,E; Tables 2, 3 and S2). On the 2nd chromosome, exchange in the pericentromeric purple to cinnabar (pr–cn) interval increased more than three fold in both heterozygotes and homozygotes when compared to WT. Moreover, on the 3rd chromosome, both c(3)GccΔ1 heterozygotes and homozygotes also showed a greater than 300% increase in recombination frequency for the centromere spanning interval scarlet to curled (st–cu). Indeed, for the 3rd chromosome more than one third of the crossover events occurred in the pericentromeric st–cu interval (as compared to 12% in wild type). Thus, both autosomes displayed a substantial increase in pericentromeric recombination frequency compared to WT, suggesting that the c(3)GccΔ1 mutation may alleviate constraints that mediate recombination in such intervals, such as centromere effect, which normally functions to restrict recombination in these intervals.
Since the X chromosome displayed both a centromere-proximal shift in recombination and a premature loss in homolog pairing, we wondered if the discontinuous SC in c(3)GccΔ1 was affecting homolog pairing on the autosomes in a similar manner. In c(3)GccΔ1 homozygotes, these decreases in recombination were paralleled by progressive decreases in euchromatic pairing for the distal probe on each autosome (Fig 4C–F; Tables 2, 3 and S3). Both the proximal and medial probes remained relatively well paired on the autosomes throughout zygotene to mid pachytene (region 2A to region 3) in c(3)GccΔ1 homozygotes.
c(3)GccΔ1 heterozygotes were able to maintain chromosome pairing along the entire length of both the 2nd and 3rd chromosomes throughout zygotene to mid pachytene (region 2A to region 3) (Fig 4D,F; Table S3). Because both c(3)GccΔ1 homozygotes and heterozygotes displayed the same altered crossover distribution, but only c(3)GccΔ1 homozygotes lost pairing in the distal chromosomal regions, the loss of homolog pairing in the distal regions of the autosomes is likely not the cause of the centromere-proximal shift in crossing over. Rather, our observations in c(3)GccΔ1 heterozygotes suggest that discontinuity of the SC may cause the shift in crossover distribution. It is possible that because centromeric SC appears to be unaffected by the c(3)GccΔ1 mutation, the SC near the centromeric heterochromatin is more stable, thus making it more suitable for crossovers.
c(3)GccΔ1 homozygotes and heterozygotes exhibit greatly reduced crossover interference
It is known that intact SC is required for crossover interference, which acts to regulate the number of crossovers formed on each chromosome arm by inhibiting crossovers from forming within several megabases of each other [reviewed in 4, 31]. In other organisms, mutants that display fragmented SC and/or chromosome axes have defects in crossover interference that result in an increase in the number of crossovers [reviewed in 4]. On the X chromosome neither c(3)GccΔ1 homozygotes nor heterozygotes displayed an increase in crossing over, in fact c(3)GccΔ1 homozygotes caused a severe reduction in crossing over (Fig 4A). Thus, it is not surprising that crossover interference is unaffected on the X chromosome in either c(3)GccΔ1 homozygotes or heterozygotes; but rather the observed loss of recombination is the result of a loss of homolog pairing and/or synapsis (Table 1).
However, unlike the X chromosome, the autosomes displayed an increase in the frequency of crossing over in c(3)GccΔ1 homozygotes and heterozygotes suggesting that crossover interference may be affected (Fig 4C,E). Indeed, crossover interference is reduced in c(3)GccΔ1 mutants on chromosome 2. c(3)GccΔ1 heterozygotes have a complete or near complete loss of crossover interference across the entire chromosome (Table 2). This decrease in interference allows for an increase in the occurrence of double and triple crossovers (Table 2) and results in an increase in map length [WT = 52.2 cM; c(3)GccΔ1 homozygotes = 67.6 cM; c(3)GccΔ1 heterozygotes = 72.8 cM]. It is possible that the decreased interference on the 2nd chromosome results from the fragmented SC observed in the c(3)GccΔ1 mutants.
On chromosome 3, c(3)GccΔ1 homozygotes displayed levels of crossover interference similar to WT in the distal regions of the chromosomes (Table 3). However, in the centromere-proximal regions [thread/scarlet/curled (th/st/cu)], the effect of interference seemed to get significantly weaker (WT = –4.5; c(3)GccΔ1 homozygotes = –12.6). While we did not expect to observe positive crossover interference across the centromere due to the centromere effect and other potential factors functioning to suppress crossing over in this region, the large increase in negative crossover interference was unexpected. Furthermore, this increase in negative crossover interference may explain the observed increase in map length for c(3)GccΔ1 homozygotes on chromosome 3 (WT = 50.9 cM; c(3)GccΔ1 homozygotes = 59.6 cM). Curiously, c(3)GccΔ1 heterozygotes displayed near-WT levels of crossover interference across all the intervals assayed on chromosome 3 (Table 3). Thus, the observed increase in map length in c(3)GccΔ1 heterozygotes cannot be explained by a loss in crossover interference (WT = 50.9 cM; c(3)GccΔ1 heterozygotes = 63.8 cM).
Chromosome segregation is unaffected by the c(3)GccΔ1 mutation
It is well known that a decrease in recombination on the X chromosome should result in an increase in frequency of X chromosome nondisjunction, such that the frequency of X chromosome nondisjunction is equal to the frequency of E0 cubed [39]. This reflects the requirement for a pair of nonexchange autosomes to mediate missegregation of nonexchange X chromosomes [40, 41]. Curiously, we did not observe this expected increase in the frequency of X nondisjunction in c(3)GccΔ1 homozygotes (Table S4).
Moreover, mutants that decrease recombination on the X chromosome and increase the frequency of X chromosome nondisjunction typically also decrease recombination on the autosomes [37, 38]. However, c(3)GccΔ1 homozygotes showed robust recombination on the autosomes (Fig 4). Previous studies have shown that the distributive system, which normally functions to segregate the obligately achiasmate 4th chromosomes, can accurately segregate two achiasmate chromosomes, such as the X and 4th chromosomes [41]. Therefore, in c(3)GccΔ1 homozygotes, it is likely that the lack of nondisjunction is due to the distributive system’s ability to accurately segregate the two achiasmate X and 4th chromosomes. The mechanism of how the distributive system accomplishes accurate segregation without a crossover is unclear, thus future studies are needed to elucidate this mechanism.
DISCUSSION
Although our understanding of SC structure has been greatly facilitated by advancements in superresolution microscopy, which have revealed a more precise three-dimensional model of SC ultrastructure, there is still much to learn about SC function. The multiple roles the SC plays during meiotic prophase illustrate the functional importance of this protein complex in ensuring the successful transmission of genetic information from one generation to the next. However, because the SC is integral for the proper execution of so many meiotic processes, it has been difficult to elucidate which SC proteins or protein domains are required for which functions. Here, we show that a deletion in the large coiled-coil domain of C(3)G reveals two different requirements for the SC: first, the deleted coiled-coil region in c(3)GccΔ1 is required to maintain both SC structure along the chromosome arms and homolog pairing during meiotic prophase; and second, the SC’s ability to facilitate recombination depends upon its full-length assembly. In addition, each chromosome appears to respond differently, on both local and global levels, to structural changes in the SC.
Centromeric SC versus euchromatic SC
The difference between the kinetics of SC assembly and disassembly along the chromosome arms versus at the centromeres suggests that the SC at the centromeres might be structurally different than the SC along the chromosome arms. Indeed, it is known that the SC at the centromeres requires certain proteins that euchromatic SC does not. For example, the cohesion complex proteins ORD, SUNN, and SOLO are required for assembly of centromeric SC but not the assembly of euchromatic SC [42-45], while euchromatic—but not centromeric—SC requires the LE protein C(2)M to completely synapse homologs [12]. We now show in c(3)GccΔ1 mutants that a large region of the major coiled-coil domain in C(3)G is also dispensable for both centromeric SC assembly and centromere clustering.
This difference between the SC at the euchromatin and at the centromere is not unique to Drosophila. In yeast, the transverse filament protein Zip1 also persists long after the SC along the chromosome arms has disassembled, and it is necessary for accurate achiasmate chromosome segregation [46, 47]. Similarly, mice spermatocytes retain CR components of the SC at the centromeres long after the SC has disassembled from the chromosome arms [48]. In both mice and Drosophila, it is unclear why the SC persists at the centromeres, but it is possible that centromeric SC in these organisms may function in achiasmate chromosome segregation, as it does in yeast. Although errors in Drosophila female meiosis and mice spermatocytes are rare, they do occur. Thus, persistence of centromeric SC may act as a backup segregation mechanism to ensure proper chromosome segregation when a rare error does occur.
Chromosome-specific regulation of recombination
All previously characterized mutants in Drosophila that affect recombination on the X chromosome show a similar, if not identical, effect on the autosomes. Prior to c(3)GccΔ1, all known SC mutants abolished or severely reduced recombination on all three crossover chromosomes [22, 24, 25, 27]. c(3)GccΔ1 is unique when compared to other SC mutants because it differentially affects recombination on each chromosome, which suggests that each chromosome has different rules for positioning crossovers along each chromosome arm [9].
While subtler than the differences observed between the X chromosome and the autosomes, the differences between the two autosomes in c(3)GccΔ1 was indeed surprising. Similar differences have been observed before. Miller et al. found that chromosome 2 displayed a stronger effect of crossover interference than chromosome 3 [9]. They noticed that the 2nd chromosome rarely had double crossovers (out of 100 flies, only 14 double crossovers were detected) and the double crossovers were positioned on average 11.5 Mb apart [9]. Meanwhile, the 3rd chromosome exhibited nearly twice as many double crossovers as the 2nd chromosome (out of 100 flies, 27 double crossovers were detected), and the double crossovers were positioned on average 11.0 Mb apart [9, 49].
In c(3)GccΔ1 homozygotes and heterozygotes, both of the autosomes display defects in crossover interference that result in an increase in the number of crossovers, but the 2nd chromosome seems to display a stronger effect. Perhaps this result is best interpreted in the context of Miller et al.’s characterization of double crossovers on the two autosomes [9]. Because the 2nd chromosome relies heavily on crossover interference to regulate crossing over, a loss of crossover interference, like that observed in c(3)GccΔ1 homozygotes and heterozygotes, would cause a large increase in the number of crossovers. Perhaps, then, the 3rd chromosome places more emphasis on an alternative mechanism, such as the centromere effect, to regulate crossing over instead of crossover interference. This may explain why the autosomes display subtle differences in recombination when crossover interference is diminished to a similar degree in c(3)GccΔ1 homozygotes and heterozygotes.
Additionally, c(3)GccΔ1 demonstrates the importance of assaying multiple chromosomes when mutants are defective in recombination. Although numerous previous characterizations of recombination-defective mutants in Drosophila have only measured recombination on one chromosome [16, 22, 24], our data from the c(3)GccΔ1 homozygotes illustrate how misleading it can be to assay only one chromosome. In this case, the X chromosome is more sensitive to the c(3)GccΔ1 mutation than are the autosomes. Differences between the sex chromosomes and the autosomes are not specific to Drosophila. For example, studies in Caenorhabditis elegans found that the X chromosome initiated meiotic DNA replication later than the autosomes [50, 51]. While the X chromosome and the autosomes seemed to homologously pair during the same small window of time in C. elegans, complete synapsis of the X chromosome was delayed and/or slower than the autosomes [50]. Currently, it is unknown in Drosophila if X chromosome synapsis is delayed, but our data support a view that initiating a synapsis delay of the X chromosome may result in a later initiation of recombination on the X chromosome.
Multiple factors work in concert to regulate recombination, and the SC appears to play a vital role in this regulation. The SC is required for crossovers in all organisms that assemble the structure, and c(3)GccΔ1 has demonstrated an additional role of the Drosophila SC in regulating the positions of crossovers. Future studies are needed to determine the mechanism of how C(3)G and the SC regulate recombination. Additionally, c(3)GccΔ1 is a large deletion mutant, thus future experiments dissecting this deletion may reveal insights into this mechanism.
MATERIALS AND METHODS
Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, R. Scott Hawley (RSH{at}stowers.org).
Experimental model and subject details
The Drosophila stocks used in the foregoing assays were kept at a humidity-controlled 25°C and on standard food. All mention of “wild type” (WT) refers to the genotype: y w; +/+; +/+; svspa-pol, unless stated otherwise. The key resource table contains a list of all the fly stocks used in this manuscript.
CRISPR/Cas9 generation of the c(3)GccΔ1 deletion flies
To aid in screening for c(3)GccΔ1 mutant flies, we incorporated a piggy BAC transposon carrying a 3xP3-DsRed that expresses in the fly eye into the intron directly downstream of the c(3)GccΔ1 deletion [intron 5 of c(3)G] in the homologous repair template plasmid. The piggy BAC transposon plasmid was constructed to have flanking Aarl and Sapl restriction sites (Addgene 51434). We used PCR to obtain two fragments of c(3)G from the Drosophila genome that flanked the position where the piggy BAC would be inserted and added in either an Aarl or Sapl restriction site. The ∼2600-bp fragment upstream of the piggy BAC insertion site contained Aarl sites and was obtained using these primers: Forward, tataCACCTGCattaCCGAcgctagtggctcctagagttcag; Reverse, gcagCACCTGCgcggTTAAtgaaaaagaatttataagtcttaccattaggttatc. The ∼1000-bp fragment downstream of the piggy BAC insertion site contained Sapl sites and was obtained using these primers: Forward, gccgGCTCTTCNTAAccttttttctacaaaatgatttatt; Reverse, gtatGCTCTTCNCGGtcatcaaaacatagtttagtatcg.
To insert these fragments into the piggy BAC plasmid, the plasmid and the downstream Sapl-containing PCR was digested with Sapl (also called Lgul from ThermoFisher ER1931), phosphatase treated (Antarctic phosphatase, NEB M0289S), and ligated together using T4 ligase (NEB M0202S). The upstream Aarl-containing PCR fragment was TOPO cloned using the Zero Blunt TOPO kit (ThermoFisher 451245). Then, TOPO Aarl plasmid was cut using the restriction enzyme Hindlll (NEB R0104S), which removes a 708-bp fragment from c(3)G. The cut TOPO Aarl plasmid was then phosphatase treated and ligated back together to create the c(3)GccΔ1 deletion. Then, this plasmid was digested with Aarl (ThermoFisher ER1581) to generate a ∼1900-bp fragment containing the c(3)GccΔ1 deletion, which was cloned into the piggy BAC plasmid containing the downstream Sapl c(3)G fragment.
A CRISPR target sequence was selected from the flyCRISPR Optimal Target Finder (http://tools.flycrispr.molbio.wisc.edu/targetFinder/). Only a single site upstream of the c(3)GccΔ1 deletion was selected (AAAGCTTTGTTGGCCTGTATTGG) and constructed into the pU6-BbsI-chiRNA guide RNA (gRNA) plasmid (Addgene 45946). Sense (CTTCGAAAGCTTTGTTGGCCTCTAT) and antisense (AAACATAGAGGCCAACAAAGCTTTC) oligonucleotides were ordered from IDT and cloned into the gRNA plasmid as described by the flyCRISPR subcloning pU6-gRNA protocol (http://flycrispr.molbio.wisc.edu/protocols/gRNA). Three single nucleotide polymorphisms (SNPs) were made in the CRISPR target sequence (the mutated bases are shown in bold: ccaatagaagcgaataaagcttt) in the c(3)GccΔ1 homologous repair template plasmid to prevent Cas9 from cutting this plasmid. These SNPs were made using the Quik Change II XL Site-Directed Mutagenesis Kit (Agilent Technology, 200521) and the reaction was performed as described in the kit protocol. The gRNA and c(3)GccΔ1 homologous repair template plasmid were sent to Genetivision (Houston, Texas) for injection into y m[VASA-Cas9-3xGFP]ZH-2A-3xRFP w1118/FM7c flies (BLM 51323). Genetivision injected the gRNA plasmid at 250 ng/µl and the c(3)GccΔ1 homologous repair template at 500 ng/µl.
c(3)GccΔ1 was isolated by crossing the G0 injected flies to y w; Pr/TM3; svspa-pol, then the F1 progeny were screened for expression of dsRed in the fly eyes. Unfortunately, the genomic locus marking the VASA-Cas9 transgene is marked with RFP, so only F1 males were screened for dsRed expression. PCR and Sanger sequencing were used to confirm the c(3)GccΔ1 deletion and a transposase (BLM 32073) was crossed in to excise the transposon from the gene. Those flies that successfully excised the transposon were PCR and Sanger sequenced to confirm the absence of the transposon and presence of the c(3)GccΔ1 deletion. Then, the c(3)GccΔ1 chromosome was allowed to freely recombine with a multiply marked third chromosome to remove any potential off-target mutations caused by removal of the piggy Bac and/or Cas9. Two c(3)GccΔ1 fly stocks were established and used for all the assays in this manuscript: (1) carrying all the markers from 3L [roughoid (ru), hairy (h), thread (th), scarlet (st)] and two markers from 3R [curled (cu), claret (ca)] in addition to the c(3)GccΔ1 deletion, and (2) only carrying ca with the c(3)GccΔ1 deletion.
Method details
Immunohistochemistry for DeltaVision and STORM microscopy
Ovary fixation and immunofluorescence for DeltaVision imaging were performed as detailed in Lake et al. 2015 [29]. Briefly, ovaries were dissected in PBS with 10% Tween (PBST) for 15 min and fixed in a 2% formaldehyde solution with heptane [165 µl PBS, 10 µl nonidet-P40 (Sigma-Aldrich, 11332473001), 600 µl heptane and 25 µl of a 16% formaldehyde (Electron Microscopy Sciences, 15710)] for 20 min. Then, ovaries were washed 3 times for 10 min each with PBST. For experiments when only the early stages of oogenesis (stages 1–6) were analyzed, the ovaries were clipped to remove the later stages of oogenesis. However, for the experiments analyzing stages 7–9, the ovaries were rapidly pipetted to separate out the stages. Next, ovaries were blocked in PBST+0.1% bovine serum albumin (BSA) for at least 1 hr, the block was removed, and fresh PBST was added with the primary antibodies and left rocking overnight at 4°C. Primary antibodies used in this manuscript include mouse anti-C(3)G C-terminus (1A8– 1G2; used at 1:500; [23]), affinity-purified rabbit anti-Corolla (used a 1:2000; [24]), rat anti-CID (used at 1:3000; from Claudio Sunkel), and mouse anti-γH2AV (used at 1:500; Iowa Hybridoma Bank, UNC93-5.2.1; [30]). The next day, the primary antibodies were removed and the ovaries were washed 3 times for 15 min each with PBST. Then, secondary antibodies were added in PBST and incubated for 2 hr, rocking at room temperature. All secondary antibodies were used at 1:500, and the secondary antibodies used were Alexa Fluor 488 goat anti-mouse (ThermoFisher, A11001), Alexa Fluor 555 goat anti-mouse (ThermoFisher, A21422), Alexa Fluor 647 goat anti-mouse (ThermoFisher, A21235), Alexa Fluor 488 goat anti-rabbit (ThermoFisher, A11008), Alexa Fluor 555 goat anti-rabbit (ThermoFisher, A21428), Alexa Fluor 647 goat anti-rat (ThermoFisher, A21434), and Alexa Fluor 555 goat anti-rat (ThermoFisher, A21247). After 1 hr and 45 min, 5 µl of 100X 4’6-diamididino-2-phenylindole (DAPI) was added and left to incubate with the secondaries for 15 min. Ovaries were then washed with PBST 3 times for 15 min and mounted in Prolong Gold (Life Technologies, P36930) using an NA 1.5 glass coverslip and frosted glass slides.
For STORM of C(3)G, ovary fixation and immunofluorescence was performed as described above with minor changes. The primary antibody was a mouse monoclonal anti-C(3)G C-terminal (1A8–1G2, used at 1:500, [23]) and the secondary antibody was Alexa Fluor 647 mouse (used at 1:500; ThermoFisher, A21235). Since samples had to be imaged the same day they were mounted, secondary antibody incubation was left rocking on the nutator overnight at 4°C. Then, the sample was washed 2 times in PBST for 15 min each. No DAPI was added to these samples. Following the washes, the sample was optically cleared using 2,2-thiodiethanol (TDE; VWR, 700008-210). The sample was rocked on the nutator at room temperature for 10 min each in 10% TDE in PBS, followed by 20% TDE in PBS, and finally in 50% TDE in PBS. For mounting, a mixture of TDE and VECTASHIELD (VWR, 101098-042) was used since it had been shown that low concentrations of VECTASHIELD allowed for blinking of the Alexa Fluor 647 dye [52]. We found that for WT samples, a mixture of 91%TDE/9% VECTASHIELD resulted in robust blinking of the dye and higher VECTASHIELD concentrations stabilized the dye, preventing it from blinking. In c(3)GccΔ1 homozygotes and heterozygotes, we had to increase the concentration of VECTASHIELD to obtain the same robust blinking as WT, thus these samples were mounted in 88%TDE/12% VECTASHIELD. All samples were mounted with NA1.5 coverslips on glass slides and sealed using clear nail polish for STORM imaging that same day.
Fluorescent BAC probes for fluorescent in situ hybridization (FISH)
FISH probes were designed from bacterial artificial chromosomes (BACs) obtained from the Children’s Hospital Oakland Research Institute (CHORI; http://bacpacresources.org/library.php?id=30). The following BACs were used: for 2L RP98-28O9 (polytene band 22A2-22A4), RP98-43K24 (polytene band 32E2-32F2), RP98-7D17 (polytene band 38E4-38F4); for 3L RP98-2N23 (polytene band 61D-61E), RP98-26C20 (polytene band 69B1-69C2), RP98-3J2 (polytene band 77F5-78B1); for the X RP98-3D13 (polytene band 3C3-3C7), RP98-9H1 (polytene band 15C1-15D6).
Bacteria containing each BAC were grown on LB+chloramphenicol and the BACs were isolated using a modification of the QIAGEN Plasmid Midi Kit protocol (QIAGEN, 12143). A single bacteria colony was grown in 5 mL LB+chloramphenicol for 6–8 hours, then that culture was moved into 100 mL culture of LB+chloramphenicol and grown overnight. Next, cells were spun down in two 50-mL bottles at 4500 g for 20 min at 4°C and the supernatant was removed (Note: Can stop here and freeze cells at 80°C.) The pellets were resuspended in 10 mL Buffer P1 followed by adding 10 mL Buffer P2 to each bottle and gently mixed by inverting 4–6 times. Then, the tubes were incubated at room temp for 5 min. Chilled Buffer P3, 10 ml, was added to each tube and immediately mixed by gently inverting 4–6 times. The tubes were incubated on ice for 15 min. The lysed cells were centrifuged at ≥ 20,000 g for 30 min at 4°C (JA-20 rotor) and the supernatant was moved to a new tube followed by centrifugation at ≥ 20,000g for 15 min at 4°C (JA-20 rotor). The QIAGEN-tip 100 were equilibrated by applying 4 mL Buffer QBT and allowed to flow through the column by gravity. The two supernatants were pooled together and added to the QIAGEN-tip column allowing gravity flow to run the liquid through the column. The QIAGEN-tip column was washed twice with 10 mL Buffer QC and the BACs were eluted off the column with 5 mL Buffer QF pre-warmed to 65°C. Following elution, the DNA was precipitated by adding 3.5 mL isopropanol, mixed well, and centrifuged at ≥ 15,000g for 30 min at 4°C (JA-20 rotor). The DNA pellet was washed with 2 mL 70% ethanol and centrifuged at ≥ 15,000g for 10 min, then allowed to air dry for 5–10 min. The DNA was dissolved in 50 μl of water, and additional water may need to be added to solubilize all the DNA.
To make the FISH probes, the BACs were PCR amplified using the Illustra GenomiPhi V2 DNA Amplification Kit (GE 25-6600-30). The concentration of the BAC DNA was determined using the Quibit and 10 ng of BAC DNA was used for one amplification reaction. The amplification reaction was performed as described in the kit protocol. To confirm that the amplification reaction worked, each sample was run on a DNA agarose gel to display a bright smear ranging from ≥10 kb–3 kb DNA sized fragments. Also, the Qubit dsDNA HS assay kit protocol (ThermoFisher, Q32851) was used to confirm that each amplification reaction had ∼4– 7 ng of DNA.
Next, the amplified BAC was restriction enzyme digested using AluI (NEB R137S), HaeII (NEB R107S), MseI (NEB R0525S), RsaI (NEB R0167S), MboI (NEB R0147S) and MspI (NEB R0106S). A 100-µl digestion reaction was set up adding in all 20 µl of the amplified BAC DNA, 64.5 µl of the New England BioLabs smart cut buffer, and 1 µl of each restriction enzyme (for Msp1, only 0.5 µl was added). The digestion reaction was left overnight at 37°C in a thermocycler. Complete digestion of the BAC DNA was checked on a DNA agarose gel where most of the DNA fragments should be below the 500-bp DNA size marker. Following the digestion, the DNA was precipitated using 2.5 volumes 100% ethanol, 1/10 volume 3M sodium acetate, and 1 µl of 20 mg/mL glycogen (ThermoFisher, 10814010). The precipitated DNA was washed using 70% ethanol and resuspended in 50–60 µl of the labeling buffer from the ULYSIS Nucleic Acid Labeling Kits (ThermoFisher – AF647 kit, U21660; AF546 kit, U21652).
To label the DNA with AF647 or AF546, the ULYSIS Nucleic Acid Labeling Kits were used. For labeling, we used 10 µl of the digested BAC DNA and followed the protocol provided in the ULYSIS kits. We removed the unreacted dyes from the labeling reaction using Centri-Sep Columns (Princeton Separation, CS-900). The protocol provided with the columns was used to remove the unreacted dyes, thereby isolating the fluorescently labeled BAC probes. The resulting labeled BAC DNA was stored at –20°C.
Fluorescent in situ hybridization of ovaries
Ovaries were dissected in PBST and fixed in a 3.7% formaldehyde and sodium cacodylate solution. A 2X fix buffer (200 mM sodium cacodylate pH7.2, 200 mM sucrose, 80 mM sodium acetate, 20 mM EGTA) was made in advance and stored in 500-µl aliquots at –20°C. The 1X fix solution (500 µl 2X Fix buffer, 232 µl 16% formaldehyde, 268 µl sterile water) was made immediately prior to dissecting the ovaries and each ovary sample was fixed in 500 µl of the 1X fix solution. Ovaries were fixed for 4 min rocking on the nutator at room temperature. Then, the ovaries were rinsed three times in 2X Saline Sodium Citrate buffer with 0.1% Tween-20 (SCCT) for 10 min each. Next, the ovary tips were clipped off using forceps to help isolate the germarium from the later stages of oogenesis. These tips were transferred into new thin-walled 0.5-mL tubes and stepped into 2X SCCT with 50% formamide by sequentially adding each formamide solution for 10 min each while nutating. First, the tips were incubated in 500 µl of a 2X SCCT with 20% formamide, then 2X SCCT with 40% formamide, and finally 2X SCCT with 50% formamide. The previous 2X SCCT with 50% formamide was removed and replaced with fresh 2X SCCT with 50% formamide that was then incubated in a thermocycler at 37°C for 1–2 hr.
Following that incubation, the 2X SCCT with 50% formamide was removed and the hybridization mixture was added to the ovary tips. For all the BAC probes, the hybridization mixture contained 2 µl BAC probe labeled with AF647, 2 µl BAC probe labeled with AF546, and 36 µl 1.1X hybridization solution (1.0 g dextran sulfate, 1.5 mL 20X SSC, 5 mL formamide, water up to 9 ml; made in advance and stored at 4°C). In a thermocycler, the probes and chromosomal DNA were denatured for 2 min at 91°C, then hybridized overnight at 37°C.
The following day 500 µl 2X SCCT with 50% formamide was added to the hybridization mixture and left at 37°C to settle. Then, the solution was removed and the ovaries were washed with 2X SCCT with 50% formamide three times at 37°C for 20 min each. Next, the ovary tips were exchanged back into 2X SCCT without formamide while nutating at room temperature by adding 2X SCCT with 40% formamide for 10 min and then adding 2X SCCT with 20% formamide for 10 min. The ovaries were washed once with 2X SCCT for 10 min, then washed two more times with PBST for 10 min each. Then, block [1% Fraction V BSA in PBST] was added and ovary tips were incubated for 1 hr at room temperature on the nutator. The block was removed and primary antibodies were added in PBST and incubated overnight at 4°C on the nutator. For all FISH samples, a cocktail of three mouse monoclonal anti-C(3)G C-terminal antibodies (1A8–1G2, 5G4–1F1, 1G5–2F7; [23]) was added and each antibody was used at a dilution of 1:500.
Ovary tips were washed three times for 20 min each in PBST and the Alexa Fluor 488 mouse secondary was added at a dilution of 1:500. The secondary antibody was incubated for 2 hr rocking at room temperature. Then, 5 µl of 100X DAPI was added and incubated for an additional 10 min. Next, the ovary tips were washed three times for 20 min each in PBST and mounted onto glass slides with NA1.5 coverslips in Prolong Gold. After 24 hr the slides were sealed using clear nail polish and kept at 4°C prior to imaging on the DeltaVison.
DeltaVision and STORM microscopy
Except for the STORM imaging (see below), all images were acquired on an inverted DeltaVision microscopy system (GE Healthcare) with an Olympus 100x Objective (UPlanSApo 100x NA 1.40) and a high-resolution CCD camera. The images from the DeltaVision were deconvolved using SoftWorRx v. 6.5 software following the GE healthcare protocol. Images were cropped and brightness and contrast was slightly adjusted using ImageJ.
STORM Imaging was performed on an OMX V4 microscopy system (GE healthcare) with an Olympus 60x TIRF Objective (APO N 60x NA 1.49). Alexa Fluor 647 was continuously excited with a 642 nm laser without UV activation, and the emission photons were filtered using a band-pass emission filter (679/41, Semrock) and collected by a PCO-Edge sCMOS camera with an 80-nm pixel size. Laser intensity measured after the objective was 4–6 kW/cm-2. 16,000–20,000 frames were recorded for each STORM data with 12–15 ms exposure time depending on signal intensity.
Meiotic recombination
The multiply-marked X chromosome (y sc cv v f y+) used in all the X recombination assays carries two versions of the yellow gene: (1) a mutant version of the yellow gene at the genomic locus on the tip of the long arm of the X chromosome (represented as y) and (2) a wild type copy of the yellow gene that was integrated onto the short arm of the X chromosome on other side of the centromere (represented as y+). To assay the frequency of recombination on the X chromosome in c(3)GccΔ1 homozygotes, males of the genotype w+/BsY; mm c(3)GccΔ1/TM3; svspa-pol/+ were crossed to FM7w/y sc cv v f y+; D/TM3 virgin females. The y sc cv v f y+/BsY; mm c(3)GccΔ1/+; svspa-pol/+ male progeny from this cross were then crossed to w+/y w; mm c(3)GccΔ1/TM3; svspa-pol/+ virgin females, which generated females carrying y sc cv v f y+/y w; mm c(3)GccΔ1; svspa-pol/+. These females were crossed individually to y sc cv v f y+/BsY males and the female progeny were scored for the presence of the sc, cv, v, f and y+ markers to determine the X recombination frequency. c(3)GccΔ1 heterozygotes were generated by crossing males of the genotype y w/BsY; mm c(3)GccΔ1/TM3; svspa-pol/+ to FM7w/y sc cv v f y+ virgin females. The resulting female progeny of the genotype y sc cv v f y+/y w; mm c(3)GccΔ1/+; svspa-pol/+ were individually crossed to y sc cv v f y+/BsY males. The X recombination frequency was determined by scoring the female progeny for the presence of sc, cv, v, f and y+ markers. To assay X recombination in WT, y w; svspa-pol virgin females were crossed to y sc cv v f y+/BsY males. The virgin female progeny from this cross were individually crossed to y sc cv v f y+/BsY males. The X recombination frequency was determined by scoring the female progeny for the presence of sc, cv, v, f and y+ markers.
To assay recombination frequency along the 2nd chromosome in c(3)GccΔ1 homozygotes, virgin females of the genotype y w; +/+; c(3)GccΔ1 ca; svspa-pol were crossed to w/y; net dpp dpy b pr cn/cyo; mm c(3)GccΔ1/TM3; svspa-pol/+ males. Single virgin female progeny of the genotype yw/w; net dpp dpy b pr cn/+; mm c(3)GccΔ1/ c(3)GccΔ1 ca; svspa-pol/+ were crossed to w+/y; net dpp dpy b pr cn males. The recombination frequency was determined by scoring the female progeny for the presence of net, dpp, dpy, b, pr, and cn markers. To assay recombination frequency along the 2nd chromosome in c(3)GccΔ1 heterozygotes, virgin WT females were crossed to w/y; net dpp dpy b pr cn/cyo; mm c(3)GccΔ1/TM3; svspa-pol/+ males. Single virgin female progeny of the genotype yw/w; net dpp dpy b pr cn/+; mm c(3)GccΔ1/+; svspa-pol/+ were crossed to w+/y; net dpp dpy b pr cn males. The recombination frequency was determined by scoring the female progeny for the presence of net, dpp, dpy, b, pr, and cn markers. Lastly, to assay recombination frequency along the 2nd chromosome in wild type, virgin WT females (y w; svspa-pol) were crossed to w+/y; net dpp dpy b pr cn males. Single virgin female progeny of this cross were then crossed to w+/y; net dpp dpy b pr cn males. The 2nd recombination frequency was determined by scoring the female progeny for the presence of net, dpp, dpy, b, pr, and cn markers.
To assay recombination frequency along the 3rd chromosome in c(3)GccΔ1 homozygotes, virgin females of the genotype y w/w+; mm c(3)GccΔ1/TM3; svspa-pol/+ were crossed to y w; c(3)GccΔ1 ca; svspa-pol males. Single virgin female progeny of this cross that were homozygous for the c(3)GccΔ1 deletion were then crossed to ru h th st cu sr e ca males. The recombination frequency was determined by scoring for the presence of ru, h, th, st, and cu in female progeny. To assay recombination frequency along the 3rd chromosome in c(3)GccΔ1 heterozygotes, virgin wild type females were crossed to y w/w+; mm c(3)GccΔ1; svspa-pol/+ males. Single virgin female progeny of this cross were then crossed to ru h th st cu sr e ca males. The recombination frequency was determined by scoring for the presence of ru, h, th, st, and cu in female progeny. Lastly, to assay recombination frequency along the 3rd chromosome in WT, virgin WT females (y w; svspa-pol) were crossed to ru h th st cu sr e ca males. Single virgin female progeny of this cross were then crossed to ru h th st cu sr e ca males. The recombination frequency was determined by scoring for the presence of ru, h, th, st, and cu in female progeny.
Meiotic nondisjunction
For the X and 4th chromosome nondisjunction assay, we introduced y w marked X chromosomes and svspa-pol marked 4th chromosomes into the c(3)GccΔ1 ca/TM3 stock to create a y w/y+Y; c(3)GccΔ1/TM3; svspa-pol stock. This was done by crossing males from y w/y+Y; Pr/TM3; svspa-pol to the c(3)GccΔ1/TM3 stock. All nondisjunction assays were performed using the y w/y+Y; c(3)GccΔ1/TM3; svspa-pol stock. X and 4th chromosome nondisjunction was assayed by crossing virgin y w; c(3)GccΔ1; svspa-pol and y w; c(3)GccΔ1/+; svspa-pol females to attached-XY, y+ v f B; C(4)RM, ci eyR males, as described in Harris et al. 2003 (the attached-XY symbol represents the chromosome C(1;Y), IN(1)EN, which is arranged as such YsX•YL) [53]. Virgin homozygous females for c(3)GccΔ1 were collected from the y w/y+Y; c(3)GccΔ1/TM3; svspa-pol stock. Virgin heterozygous c(3)GccΔ1 females were created by crossing y w/ y+Y; c(3)GccΔ1/TM3; svspa-pol males to y w; svspa-pol virgin females and collecting the virgin female progeny that displayed the correct genotype (y w; c(3)GccΔ1/+; svspa-pol).
In this assay, X chromosome nondisjunction at meiosis I will result in yellow white (y w) females, which are created by having the two maternal X chromosomes segregate to the same pole (also known as a diplo-X ova), or yellow-plus vermillion forked Bar (y+ v f Bar) males, which are created by the segregation of none of the maternal X chromosomes into the egg, resulting in an attached-XY/0 male (also known as a nullo-X ova). Meiosis II nondisjunction is extremely rare and using this assay as described above, the meiosis II nondisjunctional progeny will phenocopy the meiosis I nondisjunctional progeny. Since c(3)GccΔ1 mutants display no nondisjunction, we did not assay for meiosis II nondisjunction.
4th chromosome nondisjunction at meiosis I will result in sparkling poliert (svspa-pol) progeny, which are created by having the two maternal 4th chromosomes segregate to the same pole (also known as a diplo-4 ova), or cubitus interruptus eyeless Russian (ci ey) progeny, which are created by the segregation of none of the maternal 4th chromosomes into the egg, resulting in a C(4)RM/0 progeny (also known as a nullo-4 ova). While rare, flies with four 4th chromosomes are viable and indistinguishable from regular progeny.
Quantification and statistical analysis
γH2AV, CID, and SC intensity quantifications
Oocyte staging for the γH2AV and CID quantifications was done according to Matthies et al. [54]. Briefly, the germarium was identified by DAPI staining and each region was defined by the pattern of SC staining. Region 2A contains at least 4 cells of the 16-cell cyst with either partially or fully assembled SC, and in region 2B only 2 cells contain full-length SC. Region 3 (stage 1) is at the base of the germarium where only 1 cell has full-length SC and the other cell has either fully disassembled or is in the process of disassembling the SC. The rest of the stages (stage 2–9) were defined by measuring the size of the oocyte. Stage 2 is 25×25 µm, stage 3 is 35×35 µm, stage 4 is 40×50 µm, stage 5 is 55×75 µm, stage 6 is 60×85 µm, and stage 7 is 70×115 µm. Each oocyte was measured using the SoftWoRx software, then classified as the stage to which it was the closest in size. For stages 8 and 9 there is no quantitative way to measure the egg to determine these stages. Instead, stage 8 is defined by being bigger than a stage 7 with the oocyte cytoplasm occupying less than one-third of the egg. Stage 9 is defined by the oocyte cytoplasm occupying one-third of the egg and the follicle cells beginning to migrate over the oocyte.
Prior to the quantification of γH2AV and CID foci, individual nuclei from each region and stage were 3D-cropped in ImageJ (NIH open source software). The number of foci present for both γH2AV and CID foci was determined by analyzing each cropped nucleus in 3D and counting the number of foci. N value for both γH2AV and CID data sets represents the number of nuclei scored in each stage. In the germarium, more nuclei can be scored in region 2A than in region 2B and region 3, thus the number of germaria analyzed varies accordingly, such that the total nuclei scored in each region was 30. The analysis of region 2A came from 4–7 individual germaria. The analysis of region 2B came from 13–19 individual germarium. The analysis of region 3 came from 30 individual germaria. A Wilcoxon rank sum test was used for the P-value calculation of the γH2AV data.
To estimate protein density at different stages for the mutants acquired here, we performed quantitative imaging of the SC antibody Corolla labeled with Alexa Fluor 488. For this analysis, we assumed that the antibody penetration and epitope availability was the same between each of the regions/stages and for each mutant analyzed. SC intensity measurements were obtained by performing sum projections over SC regions and then measuring the integrated intensity of the sum projected region after subtracting the average intensity from a manually selected region near each individual SC. In some cases, the exposure time and laser power were changed in order to obtain reasonable images of low-intensity SCs. In those cases, the intensity values were corrected to account for these changes.
Euchromatic FISH pairing quantification
The same staging of the germarium as described above was done for the quantification of the euchromatic FISH pairing. To measure the distance between the FISH probe foci, a custom ImageJ plug-in was used. First, the two FISH probe foci were selected as points using the points tool and the coordinates of each point were recorded in the ROI manager. Then, the plugin measure “3D jru v1” was run to calculate the 3D distance (in µm) between ROI point one and ROI point two (available at http://research.stowers.org/imagejplugins). A slice spacing of 0.20 and pixel spacing of 0.06370 were used in the plugin to calculate the 3D distance between the foci. A locus was considered paired if the distance between the FISH probe foci was <0.75 µm and unpaired if the distanced between the FISH probe foci was ≥0.75 µm. The n value for this data is the number of nuclei scored. Similar to the γH2AV and CID quantification, the number of germaria scored for region 2A, region 2B, and region 3 vary because the number of quantifiable nuclei in each region varies. For all the probes assayed, on average ∼4 germaria were scored for region 2A, ∼8 germaria were scored for region 2B, and ∼13 germaria were scored for region 3.
STORM quantification
STORM data was analyzed through an open source ImageJ plug-in [55]. Briefly, the acquired image was smoothed by a Gaussian filter, and then each blinking molecule was fitted to a Gaussian function by maximum likelihood fitting. Individual molecule’s position, photon counts and fitting standard deviation were stored for future processing. To correct and optimize the results of the analysis, only molecules having 1–16 nm localization precision and 80–160 nm fitting standard deviation were chosen for rendering and quantification. Cross-correlation between subsets (∼2000 frames) of localized molecules was calculated to correct lateral drift.
Measurements were performed using ImageJ and the following custom plugins: fit_multi_gaussian_jru_v4; set_multi_plot_offsets_jru_v1; resample_plot_jru_v1; normalize_trajectories_jru_v1; average_trajectories_jru_v1 (available at http://research.stowers.org/imagejplugins). Measurements were performed on raw molecule position accumulation images. Intensity (i.e. molecule position accumulation) profiles were then averaged over a five-pixel-wide stripe perpendicular to the paired C(3)G C-terminal tracks. Profile fits to double Gaussian functions were used to determine the centers of each profile. Profiles were then realigned to these centers, normalized, and averaged to obtain the final averaged histograms as described in [21]. This averaged profile was then fit to a double Gaussian function, by nonlinear least squares as described in [56], using the open source tools described above. Error analysis of the fit parameters was performed using a Monte Carlo approach with 100 random simulations (also described in [56]). For wild type, n=26 SC measurements; for c(3)GccΔ1 heterozygotes, n = 34 SC measurements; and for c(3)GccΔ1 homozygotes, n = 32 SC measurements. Standard t-test was used for statistical comparisons between the c(3)GccΔ1 mutants and WT.
Meiotic recombination quantification
N values for the recombination assay refer to the number of progeny scored in each genotype. For all the chromosomes assayed, map length was determined by dividing the number of progeny that had a crossover between a given set of markers by the total number of progeny scored for one genotype on a specific chromosome. For example, the map length from sc-cv was determined by dividing the number of progeny that had a crossover between sc-cv markers by the total number of X chromosome recombinant progeny scored for a specific genotype.
Interference was calculated by:
The number of expected double crossovers occurring in adjacent intervals was determined from the WT recombination data by multiplying the map length of one interval by the map length of the adjacent interval and then multiplying by the total number of recombinants scored. For example, on the X chromosome the expected number of double crossovers occurring between the sc-cv and cv-v intervals was calculated by:
The number of observed double crossovers between adjacent intervals was empirically determined from the raw data. The total number of noncrossovers (NCOs), single crossovers (SCOs), double crossovers (DCOs), triple crossovers (TCOs) and quad crossovers (QCOs) were empirically determined from the raw data. Exchange rank was determined according to the method in Weinstein [36, 57].
Meiotic nondisjunction quantification
For both the X and 4th chromosomes, half of the progeny from the nondisjunction assay are inviable. Diplo-X ova are only viable when fertilized with sperm not carrying the attached-XY. If these ova are fertilized with sperm carrying the attached-XY, the resulting progeny will be XX/attached-XY, which is inviable. Nullo-X ova are only viable when fertilized with sperm carrying the attached-XY. If these ova are fertilized with sperm not carrying the attached-XY, the resulting progeny will not have any sex chromosomes, which is inviable. Since only half of the nondisjunctional progeny are recovered, we adjusted the total by doubling the number of X chromosome nondisjuctional progeny.
Nullo-4 flies, from nullo-4 ova fertilized by nullo-4 sperm, are not able to survive. Additionally, haplo-4 minute flies (or a single 4/0) will occur as regular progeny from the cross, but these flies are very poor and recorded but not scored in the nondisjunction tables. If both the X and 4th chromosomes nondisjoin together (X–4) only two of the four possible progeny are viable. Thus, we adjusted the total by doubling the number of the X–4 nondisjunctional progeny. For both X and 4, the frequency of nondisjunction was calculated as described in Hawley et al. 1992 [40].
Data and software availability
Primary data files for the figures in this paper are publicly accessible at www.stowers.org/research/publications/odr. For data analysis, the custom ImageJ plugins used are available at research.stowers.org/imagejplugins/zipped_plugins.html.
Key resources table
See document.
Acknowledgments
We thank Claudio Sunkel for antibodies; past and present members of the Hawley lab for helpful discussion and comments on this manuscript especially Satomi Takeo, Cathy Lake, Rachel Nielsen, and Elisabeth Bauerly; Diana Libuda for comments on this manuscript; and Angela Miller for editorial and figure preparation assistance. R.S.H. is an American Cancer Society Research Professor.