Local inversion heterozygosity alters recombination throughout the genome

SUMMARY

Crossovers (COs) are formed during meiosis by the repair of programmed DNA double-strand breaks (DSBs) and are required for the proper segregation of chromosomes. More DSBs are made than COs and the remaining DSBs are repaired as noncrossovers (NCOs). The distribution of recombination events along a chromosome occurs in a stereotyped pattern that is shaped by CO-promoting and CO-suppressing forces, collectively referred to as crossover patterning mechanisms. Chromosome inversions are structural aberrations that, when heterozygous, disrupt the recombination landscape by suppressing crossing over. In Drosophila species, the local suppression of COs by heterozygous inversions triggers an increase in crossing over on freely recombining chromosomes termed the interchromosomal (IC) effect [1, 2]. The molecular mechanism(s) by which heterozygous inversions suppress COs, whether noncrossover gene conversions (NCOGCs) are similarly affected, and what mediates the increase in COs in the rest of the genome, remain open questions. By sequencing whole genomes of individual offspring from mothers containing heterozygous inversions, we show that, while COs are suppressed by inversions, NCOGCs occur throughout inversions at higher than wild-type frequencies. We confirm that CO frequency increases on the freely recombining chromosomes, yet CO interference remains intact. Intriguingly, NCOGCs do not increase in frequency on the freely recombining chromosomes and the total number of DSBs is approximately the same per genome. Together our data show that heterozygous inversions change the recombination landscape by altering the relative proportions of COs and NCOGCs and suggests that DSB fate may be plastic until a CO assurance checkpoint has been satisfied.

eTOC

Heterozygous inversions suppress crossing over during meiosis while increasing crossover frequency elsewhere in the genome. Crown et al. show that noncrossover frequency decreases on chromosomes able to undergo exchange, inversion breakpoints do not suppress noncrossovers, and that inversions alter genome-wide double-strand break repair outcomes.

RESULTS AND DISCUSSION

To understand the molecular mechanism of the IC effect, we used whole genome sequencing to identify CO and noncrossover gene conversion (NCOGC) events in the presence of heterozygous, multiply inverted chromosomes in individual offspring of D. melanogaster females. This approach allowed us to examine rates of both COs and NCOGCs and to map these events within 100–1000 nucleotides of their precise breakpoints. Specifically, we generated females heterozygous for the X and 3^rd chromosome balancers FM7w and TM6B, or the 2^nd and 3^rd chromosome balancers CyO and TM6B and sequenced their individual progeny (Figure 1A, Table S2). These balancer chromosomes are multiply inverted across their entire lengths and, with the exception of rare double COs, completely suppress COs [3–6]. We sequenced a total of 102 individuals to average depth of 58x (min: 36x, max: 147x), with an average SNP density of 1 SNP every 358 bp for chrX, 1 SNP every 239 bp for chr2L, and 1 SNP every 294 bp for chr2R. This allowed us to identify a total of 146 individual CO events and 100 NCOGC events (data for individual events is in Figures S1 and S2 and Data S1).

Figure 1. Frequency and distribution of recombination events on freely recombining normal sequence chromosomes in the presence of heterozygous inversions.

A) Cross scheme showing the genotypes used in this study. Multiply inverted balancer chromosomes are depicted with hashes and structurally linear (normal sequence) chromosomes are solid. The multiply inverted balancer chromosomes suppress COs between that pair of chromosomes and cause the interchromosomal effect during which the normal sequence chromosomes experience an increase in CO frequency. Recombination on the X chromosome was analyzed in a different genetic background than chromosome 2 because the SNP density between w¹¹¹⁸ and Oregon-R was too low to analyze NCOGCs. B) Per-arm CO frequency and distribution. Frequency is reported as cM/Mb using a 500-kb sliding window with a best-fit line. Wild-type data (blue) are from (5), data from this study are shown in green. Some, but not all centromere-proximal heterochromatin is included in the D. melanogaster release 6 assembly and is shaded in grey. Centromeres are represented as black circles on the x-axes. C) Models of CO interference. Vertical lines represent observed distances between DCOs in wild type (blue) and during the IC effect (green). D) Per-arm NCOGC frequency and distribution. Frequency is reported as number of events per 1 Mb using a 500-kb sliding window with a best-fit line. Wild-type data from (5) is in blue, data from this study is in green. Also see Figure S1 and Data S1.

We first examined changes in the recombination landscape on the un-balanced, freely recombining chromosomes. As expected, by suppressing COs on heterozygous inversions, we observed a large increase in CO frequency on the freely recombining chromosomes as compared to when balancer chromosomes are not present (i.e. a wildtype genome). The map length of the X chromosome increases from 56 cM in wild type to 140 cM (p < 0.001; Chi-square test) and the 2^nd chromosome increases from 104 cM in wild type to 140 cM (p = 0.02; Chi square test) (Figure 1B, Tables S1 and Data S1). Although both chromosomes experience statistically significant changes in CO distribution (Kolmogorov-Smirnov test, chrX p = 0.05, chr2 p < 0.05), the X chromosome has a larger, non-uniform increase in CO frequency than chromosome 2, with disproportionately larger increases in the centromere-proximal and subtelomeric regions (Figure 1B). It is tempting to speculate that structural differences between the chromosomes cause different strengths in their response. Centromeric suppression of COs is much weaker on the X chromosome due to a large block of heterochromatin [7, 8], potentially making it easier to increase the CO rate in that region during the IC effect. Similarly different degrees of responses are also reported in the historical literature on the IC effect [1] and have been observed in natural populations [9].

Crossovers in D. melanogaster are typically limited to an average of one CO per chromosome arm by crossover patterning mechanisms. One of these mechanisms is crossover interference, which prevents COs from occurring close to each other and ultimately results in a non-random distribution of COs along a chromosome [10]. On the X chromosome, the increase in CO frequency results in the recovery of significantly more double CO (DCO) chromosomes than expected (p < 0.001, chi square test), as well as several triple and quadruple crossovers (Table 1, Figure S2), which in wild type are recovered at rates fewer than 1 in 1,000 or 1 in 10,000, respectively [11]. An increase in frequency of multi-CO chromosomes could indicate a loss or weakening of crossover interference. We modeled CO interference by calculating the average physical pairwise distance between single COs in wild type and compared that to the observed distances between DCOs (Figure 1C and Methods). For all three chromosome arms, DCO distances are significantly greater than expected by chance (p < 0.001), showing that interference is still intact. For the X chromosome, DCO distances are not significantly different than distances observed in wild type; however, for chromosome 2, the observed DCO distances are shorter than in wild type (Figure 1C), suggesting that interference may be weaker, but not absent, during the IC effect on this chromosome. Together, these data show that interference is largely intact in the presence of a heterozygous inversion and that chromosome arms can accommodate more than one CO while still maintaining interference.

TABLE 1.

EXPECTED AND OBSERVED FREQUENCIES OF RECOMBINATION EVENTS ON FREELY RECOMBINING CHROMOSOMES IN THE PRESENCE OF HETEROZYGOUS INVERSIONS

	ChrX		Chr2L		Chr2R
	Expected	Observed	Expected	Observed	Expected	Observed
NCOGC	25.2	7	26.2	7	26.2	8
SINGLE CO	22.1	20	23	27	23.0	23
DOUBLE CO	2.7	15	2.8	4	2.8	6
TRIPLE CO	0.1	4	0.1	1	0.1	0
QUADRUPLE CO	0	2	0	0	0	0

Total observed CO and NCOGC frequencies are significantly different than expected based on wildtype frequencies (p < 0.05, Fisher's exact test, both chromosomes).

One possible mechanism for increasing CO frequency is to make more DSBs. Alternatively, the number of DSBs could remain the same, but a higher proportion of those DSBs could be repaired as COs at the expense of NCOGCs. We tested both of these hypotheses by examining the number of NCOGCs that occurred on the freely recombining chromosomes (Figure 1D, Data S1). We found that for both chromosomes, we recovered significantly fewer NCOGCs than expected (Table 1). Additionally, using maximum likelihood analysis, we estimate a rate of 1.4 × 10^-8 NCOGCs per base pair per generation for the freely recombining 2^nd chromosome and 1.9 × 10⁻⁸ for the freely recombining X chromosome, rates that are approximately 65% and 89% of the wild type genome-wide estimate of 2.1 × 10^-8 using the same method [7]. The pattern of NCOGCs for chromosomes 2L and 2R show a uniform decrease in rates across the chromosome, yet the pattern of NCOGCs for the X chromosome shows large decreases in rate for some regions and increases for others, potentially explaining why the per base pair rate of NCOGC is closer to wild type for the X chromosome. Thus, the frequency of COs on freely recombining chromosomes increases in response to heterozygous inversions, yet there appears to be no corresponding increase in the rate of NCOGCs.

We next analyzed the relationship between COs and NCOGCs on the inverted chromosomes. We identified 78 NCOGC events between the 2nd and 3rd chromosome balancers and their structurally linear homologs (Figure 2, Data S1). Using maximum likelihood analysis, we estimate a NCOGC rate between these multiply inverted chromosomes and their homologs of 3.0 × 10^-8 per base pair per generation, a rate approximately 40% higher than wild type. These data suggest that DSBs on inversion chromosomes are preferentially repaired as NCOGCs. We have, however, also considered the possibility that the NCOGC rate is higher on inversion chromosomes due to preferential recovery of non-exchange chromosomes. Sturtevant and Beadle demonstrated that COs within paracentric inversions do not affect viability and are most likely segregated into polar bodies, supporting the possibility that transmission distortion is responsible for the increase in NCOGCs in the present study [12]. However, cytological studies of the X chromosome balancer FM7 (which contains only paracentric inversions) have demonstrated that these chromosomes are rarely linked by chiasmata at metaphase, arguing that DSBs on these chromosomes are rarely repaired as COs and that exchange is truly suppressed [13, 14]. COs within pericentric inversions (as are found on 2^nd and 3^rd chromosomal balancers) result in lethal duplications and deletions but are predicted to segregate normally. However, there are genetic data showing that pericentric inversions surprisingly do not decrease viability, but rather that the frequency of COs within pericentric inversions is reduced to 25% of wildtype [15, 16]. Thus, balancers comprised of paracentric and those comprised of pericentric inversions prevent the recovery of crossovers primarily by reducing exchange, rather than reducing the recovery of exchanges. Therefore, while transmission distortion may be contributing to the increase in NCOGCs in the present study, we believe the effect is largely due to preferential repair of DSBs into NCOGCs.

Figure 2. The frequency and distribution of NCOGCs on multiply inverted balancer chromosomes.

A) Location of NCOGCs onto the multiply inverted 2^nd chromosome balancer CyO. B) Per-arm NCOGC frequency and distribution for the multiply inverted 2^nd chromosome balancer CyO. Frequency is reported as number of events per 1 Mb using a 500 kb sliding window with a best-fit line. Some centromere-proximal heterochromatin is included in the release 6 assembly and is shaded in grey. C) Location of NCOGCs onto the multiply inverted 3^rd chromosome balancer TM6B. Heterochromatin is shaded in grey. D) Per-arm NCOGC frequency and distribution for the multiply inverted 3^rd chromosome balancer TM6B. Frequency is reported as number of events per 1 Mb using a 500 kb sliding window with a best-fit line. Also see Figure S2 and Data S1.

There is mounting evidence that, in Drosophila species, inversion breakpoints suppress COs 1–3 Mb outside the breakpoint [2, 4–6, 17]; this type of suppression is mechanistically distinct from the transmission distortion discussed above as it extends outside of the inversion. Indeed, in this study we recovered one single CO and one DCO that occurred between a balancer chromosome and its structurally linear homolog, with both events occurring no closer than 1.7 Mb from the nearest inversion breakpoint (Figure S2, Data S1). It has remained unclear whether NCOGCs are similarly suppressed near inversion breakpoints. We analyzed the distances between inversion breakpoints and NCOGCs and found that 21 NCOGCs occurred within 1 Mb of an inversion breakpoint. Of those 21 NCOGCS, only 3 were within 500 kb of the breakpoints (Figure S3). While these data suggest that NCOGCs may be suppressed very locally around inversion breakpoints, overall they are clearly less sensitive to the mechanisms that suppress CO formation 1–3 Mb around inversion breakpoints.

The most parsimonious explanation for the genome-wide changes in CO and NCOGC frequencies during the IC effect is that the total DSB number is approximately the same as in wild type. Cytological studies in D. melanogaster have demonstrated that 20–25 DSBs are made during female meiosis and that this number does not increase in the presence of multiple balancer chromosomes [18, 19]. However, the average number of DSBs can also be estimated from our data. Using the NCOGC rate in inversions and the CO and NCOGC rates on freely recombining chromosomes, we calculate that 16.6 DSBs were made per genome per meiosis (Table 2; this calculation corrects for the fact that only one quarter of the NCOGCs and one half of COs are recovered in the progeny.). We used the same analysis on wild-type data [7] and, strikingly, calculated that 17.5 DSBs are made per genome per meiosis (Table 2). The calculated number of DSBs in each experiment is 10–20 percent smaller than cytological estimates, potentially because of genetically invisible events such as noncrossover gene restorations or intersister recombination. Indeed, the ability to adjust the ratio of COs to NCOGCs is reminiscent of CO homeostasis in budding yeast and mice, where if the number of DSBs is decreased, the number of COs is generally maintained at the expense of NCOGCs [20, 21].

TABLE 2.

CALCULATION OF NUMBER DSBS PER MEIOSIS. RELATED TO FIGURES 1 AND 2.

WILD-TYPE DSB CALCULATION
WILD-TYPE STOCKS SEQUENCED		196
NCOGCS (EXPECTED)
NCOGC RATE1 (CALCULATED)		2.10E-08
NCOGCS PER HAPLOID MEIOSIS		2.8
DSBS THIS ACCOUNTS FOR PER MEIOSIS		11.1
COS (OBSERVED)
TOTAL COS OBSERVED		541
DOUBLE CO NUMBER AS ONLY 1/2 ARE OBSERVED		1082
BREAKS PER STOCK ON THE X		5.5
TOTAL BREAKS		16.6
DSB CALCULATION USING DATA FROM THE X AND NCOGC RATE FROM BALANCERS
CYO; TM6B STOCKS SEQUENCED	50
CYO AND TM6B EVENTS (OBSERVED AND EXPECTED)
NCOGC RATE1 (CALCULATED)	2.95E-08
2ND AND 3RD CHROMOSOME NCOGCS PER HAPLOID MEIOSIS	3.2
DSBS THIS ACCOUNTS FOR PER MEIOSIS	12.9
CHROMOSOME X EVENTS (OBSERVED AND EXPECTED)
TOTAL COS OBSERVED	70
DOUBLE CO NUMBER AS ONLY 1/2 ARE OBSERVED	140
NCOGC RATE (CALCULATED)	1.86E-08
X CHROMOSOME NCOGCS PER HAPLOID MEIOSIS	0.44
BREAKS NCOGCS ACCOUNT FOR IN ALL STOCKS	88
TOTAL BREAKS ON THE X	228
BREAKS PER STOCK ON THE X	4.6
TOTAL BREAKS	17.5

¹NCOGC rates are maximum likelihood estimates. See Methods for description.

Our data lead to a model where the recombination landscape in D. melanogaster responds to heterozygous inversions by modifying the fate of DSBs, not by increasing the number of DSBs. Furthermore, most of this shift occurs in sub-telomeric and centromere proximal regions where CO control mechanisms normally favor NCOGC formation [7]. Recent evidence has shown that the number of DSBs formed during meiosis is controlled using multiple feedback mechanisms. In several organisms, including D. melanogaster, there is a DSB-limiting mechanism that shuts down DSB formation [22, 23]. In budding yeast, there is also a feedback mechanism that increases the number of DSBs made in response to problems with “homolog engagement” [24]. While the IC effect has been compared to problems with homolog engagement by others [24], cytological examination of phosphorylated H2AV, a marker of DSBs, and our maximum likelihood estimates presented here suggest that DSBs never accumulate to levels higher than wild type [18]. Thus, a comprehensive assessment of the data suggests that the increase in COs during the IC effect is not due to an increase in DSB number.

It is intriguing to consider how the IC effect might cause this change in DSB fate. One current model is that chromosome axis discontinuities between the inversion and its homolog trigger a checkpoint that extends the time during which DSBs can be repaired as COs [1, 18]. An alternative model is that the absence of a CO on the inverted chromosome triggers a crossover assurance checkpoint. Until this checkpoint is satisfied, DSBs continue to be repaired as COs on the freely recombining chromosome. Since interference is still intact, COs in the medial part of the chromosome arm still prevent neighboring DSBs from being repaired as COs, but DSBs in the subtelomeric and centromere proximal regions are free to be repaired as COs. The latter model implies that NCOGC formation in subtelomeric and centromere-proximal regions is delayed until after CO formation is complete. Existing molecular evidence supports the idea that the final decision to repair a DSB as a CO or NCOGC is made late in the DNA repair process [25]; this late decision may provide flexibility in response to a CO assurance checkpoint. Interestingly, in many organisms, DSB repair is required for full chromosome pairing and formation of the synaptonemal complex (the tripartite structure formed between homologous chromosomes during meiosis) [26]. In D. melanogaster, chromosomes pair and synaptonemal complex is built before DSBs are made [27]; therefore, DSBs likely provide no advantage during homology searching. Our data lead us to believe that in D. melanogaster, where DSB repair is not required for homolog pairing, NCOGCs can function as a reserve of potential COs sites when the desired number of COs have not been formed.

Recombination, and particularly crossing over, is essential for proper chromosome segregation during meiosis, but it is also important for allowing genetic exchange among individuals, which ultimately homogenizes nucleotide diversity within or between populations. Depending on their frequency in a population, inversions can act as barriers to gene flow by suppressing COs, and thus play an important role in sequestering haplotypes away from genetic exchange. This sequestration is essential for maintaining linkage and may be particularly important in cases when inversions harbor blocks of co-adapted alleles, such as supergenes [28] and meiotic drive systems [29], and is also a critical mode of speciation [30, 31]. Suppressing COs is effective at slowing gene flow at the megabase scale, but the rates of NCOGCs we have presented here would allow higher than wild-type rates of gene flow on a small scale (the average NCOGC size is 400 bp in D. melanogaster) [32]. We have shown that NCOGCs clearly happen throughout inversions, are insensitive to the mechanisms that suppress CO formation, and are transmitted at a rate approximately 40% higher than wild type. While NCOGCs are unlikely to completely break down linkage disequilibrium in an inversion, they will allow new genetic information in and out of the sequestered region. Therefore, while inversions may be maintained because of their ability to keep haplotypes together by suppressing COs, NCOGCs may be equally important for their persistence by allowing frequent but small amounts of gene flow. Our baseline estimate of the transmission rate of NCOGC frequency within inversions suggests that NCOGCs may cause faster homogenization of nucleotide diversity than is accounted for in current models and at least necessitates a reexamination of those models [33].

Together, our data lead to a comprehensive model of how inversions act both in cis and trans to change the global landscape of recombination in D. melanogaster. Locally, they prevent DSBs from being repaired as COs but have no negative effect on NCOGCs, while globally, they cause an increase in COs and a decrease in NCOGCs on freely recombining chromosomes. Thus, the net effect of heterozygous inversions in D. melanogaster is not just local suppression of crossovers, but a genome-wide shift in the distribution of large-scale genetic exchange. Such dramatic changes to the pattern of gene flow across genomes will have an impact on the dynamics of population genetics and is especially important to consider when modeling the contributions of inversions to speciation or ecological adaptation.

STAR METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Scott Hawley (rsh@stowers.org).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Crosses used to create the double-balanced females used in this study are shown in Figure 1A. The laboratory w¹¹¹⁸, Oregon-R, ISO-1, FM7w, TM6B and CyO stocks used in this study have been maintained in the labs of RSH and JS separately from the Bloomington Drosophila Stock Center, and are available upon request. The wildtype data used in this study were previously published in (5) and were from a w¹¹¹⁸/Canton-S background. Crosses for the wildtype data and this data were maintained at 25 degrees on standard medium.

METHOD DETAILS

DNA isolation, sequencing, genome alignment, and SNP calling

DNA was isolated from either 10 females from each of the parental lines or single males or females for individual offspring (Table S1). Flies were frozen at −80 C for at least 1 hr before DNA extraction using the Qiagen Blood and Tissue kit following the manufacturer instructions. A Covaris S220 sonicator was used to shear genomic DNA and a Perkin Elmer Sciclone G3 NGS Workstation was used to construct libraries using the KAPA HTP Library Prep kit (KAPA Biosystems, Cat. No. KK8234) using NEXTflex DNA barcodes (Bioo Scientific, Cat No. NOVA-514104). A Pippin Prep (Sage Science) was used for library size selection post-amplification and libraries were pooled after quantification with an Invitrogen Qubit 2.0 Fluorometer and an Agilent 2100 Bioanalyzer. Sequencing was performed on an Illumina NextSeq 500 instrument as 150 bp on a high-output, paired-end flow cell. Illumina NextSeq Real Time Analysis version 2.4.11 and bcl2fastq2 v2.18 were run to demultiplex reads and generate FASTQ files. Bwa version 0.7.15-r1140 [34] was used to align reads to release 6 (dm6) of the Drosophila melanogaster genome and SNPs were called using SAMtools mpileup [34]. For Figures 1 and 2, heterochromatin is defined computationally as highly repetitive regions near the centromere that are assembled in release 6 of the Drosophila genome [35].

Sanger sequencing of individual NCOGCs

Twenty random NCOGC events were chosen for confirmation by Sanger sequencing. Primers were designed using Primer3 to amplify 500–1600 bases around the NCOGC event. Polymerase chain reaction (PCR) was carried out using DNA from the same DNA isolation used in whole genome sequencing described above. All PCR reactions contained: 0.5 ul Primestar GXL Polymerase (Takara Clontech #R050A), 5 ul Primerstar GXL Reaction Buffer, 0.5 ul each primer (primers resuspended at 50 uM), 0.5 ul 10 mM dNTPs, 2–10 ng DNA, water to 25 ul total volume. Reactions were run on a 1% agarose gel and gel purified using 600 ul Buffer QG (Qiagen #19063) and a GeneJet PCR Purification Kit (ThermoFisher Scientific #0701). PCR products were sent to EtonBio for standard Sanger sequencing. All 20 events were confirmed as real.

QUANTIFICATION AND STATISTICAL ANALYSIS

Identification of CO and NCOGC events

Crossover (CO) and noncrossover gene conversion (NCOGC) events were identified as in [5] using custom scripts available at the link below. Briefly, to identify CO and NCOGC events onto chromosomes able to undergo exchange, the SNPs in each parental genotype were called (either Oregon-R and ISO-1 for the X chromosome and Oregon-R or w¹¹¹⁸ for the 2^nd chromosome). For the X, only homozygous parental Oregon-R SNPs were used while only SNPs heterozygous between Oregon-R and w¹¹¹⁸ were used for the 2^nd. Repetitive regions were masked as defined by repeatmasker [36]. For both classes only parental SNPs with quality scores >150 that were unique between the parents were used in order to determine where, if anywhere, in the offspring the parental genotype changed. Only SNPs in the offspring with quality scores >220 were considered and each change in genotype flagged by our script was confirmed by visual analysis.

To identify COs and NCOGCs onto the CyO or TM6B balancer chromosomes, we identified either SNPs or loss of a SNP for each balancer that were unique among all the balancer chromosomes sequenced. For example, 50 CyO chromosomes were sequenced as CyO/ISO-1, so only unique SNPs, or loss of a SNP, present in 1 of 49 individuals was considered a candidate NCOGC event. Each potential NCOGC and CO was visually validated using IGV [37] by confirming that the SNP or loss of SNP was consistent with the sequence of the non-inverted homolog.

Calculation of CO and NCOGC rates

For all analyses, 2L and 2R were treated as independent chromosomes because CO control mechanisms act independently on individual arms. CO rates were calculated as the number of COs occurring per 1 Mb using a 500 kb sliding window. Changing the size of the sliding window to either 250 kb or 1 Mb did not change the statistical significance of the data. NCOGC rate was estimated using maximum likelihood analysis as in [7]. Briefly, for each calculation observed, NCOGCs were fit to the actual SNP distributions from respective chromosome arms. SNP distributions for each chromosome arm were collected by counting the number of intervals in which a single bp separated two SNPs, then the intervals in which 2 bp separated two SNPs, until all intervals up to 10,000 bp between SNPs had been counted. For each trial, tract lengths of 100–500 bp were tested in increments of 5 bp using five different DSB rates from 1.0 × 10⁻⁷ to 4.0 × 10^-8. A local maximum was identified, and then tract length and rate were reset and rerun in smaller increments until the true local maximum was identified. Mathematica scripts of each model are available at the link below.

Modeling of interference

Interference for the X and 2^nd chromosomes was modeled independently by performing 100,000 trials of randomly selecting two single crossover (SCO) events from wild type reported in [7] and measuring the distance between them. The wild-type pool of SCOs selected from was not significantly different from the SCOs recovered in this study (two-sample Kolmogorov-Smirnov p-value 0.28 for the X chromosome, 0.72 for chromosome 2L, and 0.84 for chromosome 2R). We then compared the average distance between double crossovers (DCOs) for the X, 2L, and 2R independently to the SCOs from the respective chromosomes. A p-value was calculated by comparing the observed DCO size to the average DCO size obtained by bootstrapping (X chromosome expected 5,735,929 bp, observed 8,478,101; chromosome 2L expected 5,469,470 bp, observed 9,811,734; chromosome 2R expected 5,018,928 bp, observed 8,448,263). The strength of interference was determined by comparing wild-type data to the DCO data obtained by bootstrapping SCO events. The distance between two randomly selected SCO events was used to construct a histogram using a bin size of 100 kb from which a best-fit curve was derived. The average value between observed DCOs in this study was then plotted and compared to the average value between observed DCOs in [7]. For all observations, any triple crossover (TCO) was treated as two DCOs and any quadruple crossover (QCO) was treated as three DCOs.

Supplementary Material

1

Data S1. Sequence coordinates of all CO and NCOGC events in this study. Related to Figures 1 and 2 and Table 1.

Highlights.

• Heterozygous inversions shift the crossover landscape (the interchromosomal effect).

• Crossovers on freely recombining chromosomes increase while noncrossovers decrease.

• Unlike crossovers, noncrossovers are not suppressed by inversion breakpoints.

• The interchromosomal effect does not increase the number of double-strand breaks.