Excess crossovers impede faithful meiotic chromosome segregation in C. elegans

During meiosis, diploid organisms reduce their chromosome number by half to generate haploid gametes. This process depends on the repair of double strand DNA breaks as crossover recombination events between homologous chromosomes, which hold homologs together to ensure their proper segregation to opposite spindle poles during the first meiotic division. Although most organisms are limited in the number of crossovers between homologs by a phenomenon called crossover interference, the consequences of excess interfering crossovers on meiotic chromosome segregation are not well known. Here we show that extra interfering crossovers lead to a range of meiotic defects and we uncover mechanisms that counteract these errors. Using chromosomes that exhibit a high frequency of supernumerary crossovers in Caenorhabditis elegans, we find that essential chromosomal structures are mispatterned in the presence of multiple crossovers, subjecting chromosomes to improper spindle forces and leading to defects in metaphase alignment. Additionally, the chromosomes with extra interfering crossovers often exhibited segregation defects in anaphase I, with a high incidence of chromatin bridges that sometimes created a tether between the chromosome and the first polar body. However, these anaphase I bridges were often able to resolve in a LEM-3 nuclease dependent manner, and chromosome tethers that persisted were frequently resolved during Meiosis II by a second mechanism that preferentially segregates the tethered sister chromatid into the polar body. Altogether these findings demonstrate that excess interfering crossovers can severely impact chromosome patterning and segregation, highlighting the importance of limiting the number of recombination events between homologous chromosomes for the proper execution of meiosis.


Introduction
Meiosis is a specialized, reductional form of cell division necessary for the production of haploid sperm or egg cells. One hallmark of meiosis is the requirement of genetic exchange through recombination. Meiotic recombination is initiated by the formation of double strand DNA breaks (DSBs), which are repaired to form crossover (CO) and noncrossover events [1]. Germ cells in most organisms require a CO between homologous chromosomes to physically link each pair of homologs, enabling their proper segregation during the Meiosis I division. Despite the formation of many programmed DSBs, most organisms are limited in the number of COs formed, and formation of a CO tends to inhibit formation of other COs nearby on the same chromosome pair, a conserved phenomenon known as CO interference [2,3]. COs that are subject to interference are considered interfering COs (or Class I COs). Additionally, some organisms have a subset of COs that are not subject to interference (called "non-interfering", or Class II COs), but these non-interfering COs represent only 5-35% of all meiotic COs in A. thaliana, M. musculus, and S. cerevisiae [1]. While CO interference is a well conserved phenomenon among most eukaryotes, the molecular consequences that result from the occurrence of multiple interfering COs are not clear. In addition, it is also unclear how these consequences may have contributed to the conservation of CO interference.
The model organism Caenorhabditis elegans is a particularly powerful system to study CO regulation because it exhibits remarkably strict CO control; under wild-type conditions, only one DSB per chromosome is repaired as a CO and all COs are interfering COs [4]. Several studies have demonstrated that even in the presence of an extreme excess of DSBs (10-fold greater than wild-type levels), only a single interfering CO, marked by the pro-crossover factor COSA-1, is made per pair of homologous chromosomes [5,6]. Additionally, it has been shown that CO interference can operate over distances longer than the length of a normal chromosome axis [5,7]; in the case of end-to-end fusions of chromosomes (that still accurately segregate in C. elegans), many meioses still only have one CO per fusion chromosome pair [5,7].
COs form physical connections between the homologs, known as a chiasmata. Since in C. elegans each chromosome pair typically has one CO that occurs off-center along the chromosome length, the chromosomes reorganize around this single chiasma to form cruciform bivalents with long and short arms [8,9]. These bivalents then align on the spindle, and in anaphase I, cohesion is lost along the short arm axis, enabling segregation of homologous chromosomes [10,11]. Aurora B kinase (AIR-2) and other members of the conserved chromosomal passenger complex (CPC) are targeted to this short arm region in prophase to protect sister chromatid cohesion in that region until anaphase I is triggered [8]. Moreover, the CPC also directs the formation of a larger meiotic protein complex that forms upon nuclear envelope breakdown [12,13]; at this stage, the CPC reorganizes from a linear distribution along the short arm axis to a ring encircling this region, and targets a number of other conserved proteins to form a structure known as the Ring Complex (RC) [14]. In addition to the CPC, the RC contains other conserved components, such as the kinase BUB-1 [13] and the microtubule de-stabilizing kinesin MCAK   [15,16]. Furthermore, SUMO and SUMO pathway enzymes localize to the RC and are required for RC assembly and stability [17,18]. Chromosomes that lack RCs have congression and segregation errors [19], and depletion of RC components causes a variety of meiotic defects [12,13,15,18,[20][21][22][23], highlighting the importance of this complex.
One function of the RC is to promote chromosome congression [12,18]. In C. elegans oocyte spindles, microtubule bundles run laterally along the sides of bivalents instead of forming canonical end-on kinetochore attachments. A component of the RC, the kinesin-4 family member KLP-19, has been proposed to walk along these laterally-associated bundles towards microtubule plus ends located in the center of the spindle, thus providing chromosomes with plus-end directed forces that mediate metaphase alignment [12]. Then, at anaphase onset, the enzyme separase (SEP-1) is targeted to the midbivalent region to cleave cohesin and allow homologous chromosomes to segregate to opposite spindle poles [19,24]. At this stage, the RCs are removed from chromosomes and remain in the center of the spindle, where they disassemble [13,17,19,25,26]. Oocytes extrude one set of homologs into a polar body, and then Meiosis II (MII) proceeds. The MII chromosomes assemble RCs encircling the sister chromatid interface and repeat the segregation process to form a matured haploid egg.
Although extensive research has focused on CO formation and RC function independently, it is still unclear how early meiotic processes affect chromosome structure and function in meiotic divisions. Moreover, why chromosomes in many organisms are so tightly restricted to 1-2 COs per homolog pair has not been addressed. Here we utilize C. elegans strains containing a chromosome that exhibits increased crossover numbers under specific conditions to assess the effects of supernumary crossovers on meiosis. We find that the RC is mispatterned in the presence of multiple COs, leading to defects in chromosome congression, and that chromosomes with multiple chiasmata exhibit extensive chromatin bridging in anaphase. Thus, excess crossovers can severely impact chromosome patterning and segregation, highlighting the importance of limiting the number of recombination events between homologous chromosomes for the proper execution of meiosis. Further, our studies uncovered multiple mechanisms by which oocytes are able to correct these errors, demonstrating that species have evolved ways to combat the deleterious defects caused by excess crossing over.

Fusion chromosomes generate bivalents with multiple crossovers and chiasmata
To understand the effects of multiple COs on proper chromosome segregation and gamete formation, we sought to exploit the advantages of the C. elegans model system to consistently generate multiple COs along a single chromosome during meiosis. Since normal C. elegans chromosomes typically experience only one CO per meiosis [5,7], we utilized a strain containing the three-chromosome fusion meT7 (end-on-end fusions of chromosomes III, X, and IV; Fig 1A) [7]. While wild-type strains contain six individual chromosomes, strains containing the meT7 fusion chromosome have a total of four individual chromosomes total (meT7 III; X; IV fusion chromosome, and chromosomes I, II, and V; Fig 1A and 1B, and S1 Fig). When grown under standard conditions of 20˚C, wild type and meT7-containing strains have relatively normal progeny viability (100% versus 98.8%; Table 1), indicating that meT7 segregation errors leading to aneuploidy are rare [7]. Since meT7 is three times the length of a normal chromosome, previous studies using genetic assays found the occasional occurrence of multiple COs along this fusion chromosome and that this chromosome is easily identifiable in meiotic nuclei due to its size [7,8].
To determine the exact number of Class I interfering COs occurring along the length of meT7 within a population of nuclei, we assessed the number of COSA-1 foci along meT7 chromosomes in single nuclei of strains grown in standard conditions; COSA-1 is a conserved protein required for CO formation that acts with other pro-crossover factors (MSH-5, ZHP-1, ZHP-2, ZHP-3, and ZHP-4) and serves as a robust cytological marker of Class I interfering COs during late pachytene [6,[27][28][29]. In wild-type strains containing unfused chromosomes, the six individual chromosomes in each nucleus obtain a single CO per chromosome resulting in 6 COSA-1 foci per nucleus, with <0.4% of nuclei obtaining more than 6 foci [5,6] (Fig 1B  and Materials and Methods). In contrast, strains grown at 20˚C with the meT7 fusion chromosome obtain 4-6 COSA-1 foci per late pachytene nucleus (Fig 1B arrowheads), with 77% of meT7 chromosomes exhibiting �2 COSA-1 foci, consistent with previous analysis of meT7 chiasmata [8] (Fig 1C). To further increase the number of COs occurring along meT7, we grew strains containing the meT7 fusion chromosome at 25˚C, a temperature that was previously found to increase the number of COs along the length of the two-chromosome fusion mnT12 [5] (S1 Fig and S2 Fig). Compared to 20˚C, we found that meT7 strains grown at 25˚C had an increase in COSA-1 foci, with >90% of meT7 chromosomes with 2 or more COSA-1 foci ( Fig  1B arrowheads   Left, schematic depicting the chromosomes in a wild-type strain (top) and the meT7 (X;III;IV) fusion chromosome strain (bottom). Right, immunofluorescence images of GFP::COSA-1 in fixed late pachytene nuclei from wild-type and meT7 fusion chromosome strains grown at 20˚C and 25˚C. GFP::COSA-1 is shown in green, synaptonemal complex protein SYP-1 is shown in red, and DNA is shown in blue. Arrowheads indicate nuclei with �2 crossovers along the meT7 fusion chromosome. Scale bars = 5 μm. (C) Top, quantification of percentage of meT7 nuclei with indicated number GFP::COSA-1 foci on meT7 in late pachytene at 20˚C and 25˚C. Number of late pachytene nuclei scored for COSA-1 foci: 20˚C, N = 290; 25˚C N = 287. Bottom, representative immunofluorescence images of fixed single nuclei with meT7 chromosomes with the indicated number of COSA-1 foci. White line indicates traced chromosome axis of each chromosome in a single nucleus. Yellow outline highlights the meT7 fusion chromosome within the nucleus. (D) Three-dimensionally rendered immunofluorescence images of individual wild type and meT7 diakinesis bivalents. Dashed lines (white) indicate traced HTP-3 axes (green), with crossing of axes and arrows indicating chiasmata. Scale bars = 1 μm.  Table 1). This enrichment of COSA-1 foci in meT7 strains indicates the occurrence of multiple COs along the length of the meT7 fusion chromosome.

PLOS GENETICS
To determine whether these increased COSA-1 foci along meT7 fusion chromosomes represent COs that become chiasmata, we assessed bivalent structure and chiasma number at diakinesis along fusion chromosome pairs (Fig 1D and S3 Fig). COs are normally formed off-center of the C. elegans chromosome and some meiotic chromosome proteins reorganize around these CO sites starting at the late pachytene-diplotene transition, resulting in cytologically distinguishable long and short bivalent arms; in this structure, the chiasma is positioned where the chromosomes axes cross [8,9,30] (Fig 1D; chromosome axes marked by HTP-3 immunofluorescence). In accordance with previous analysis performed at 20˚C [8], we found that some meT7 bivalents grown at 25˚C exhibit unique meiotic chromosome structural reorganization, which is consistent with the occurrence of multiple COs. Notably, some meT7 bivalents exhibited multiple chiasmata (Fig 1D, chiasmata indicated with arrows) and multiple short arms (S3 Fig, long and short arms marked with HTP-1/2 and SYP-1, respectively); we also observed similar structures in the two-chromosome fusion mnT12 [8] (S3 Fig). While unfused chromosomes at diakinesis only had one chiasma (21/21 at 20˚C and 21/21 at 25˚C), we found that 13/21 diakinesis nuclei with meT7 bivalents at 20˚C had multiple chiasmata, consistent with 60-80% of the meT7 chromosome pair having more than one COSA-1 focus at the late pachytene stage ( Fig  1C). In meT7 strains grown at 25˚C, we found that 19/21 (90.5%) of meT7 fusion chromosomes had more than one chiasma, consistent with 90% of meT7 chromosome pairs having 2 or more COSA-1 marked crossovers ( Fig 1C) as well as a temperature-dependent increase in crossovers that result in chiasmata (20˚C versus 25˚C; P = 0.02; Mann-Whitney, two-tailed). Together, this corresponding increase in both COSA-1 foci and chiasmata indicates that the temperature-associated extra COSA-1 foci observed at pachytene in meT7 chromosomes represent bona fide cytologically-differentiated meiotic CO events that can result in atypical bivalent structures.

Bivalents with multiple crossovers have mispatterned Ring Complexes
Given that we can increase CO numbers along an easily identifiable chromosome under a specific condition, we set out to investigate how extra chiasmata might affect other aspects of bivalent organization by assessing the Ring Complex (RC), a structure comprised of a set of critical meiotic proteins. AIR-2 and other CPC components localize along the short arms of each cruciform bivalent during diakinesis, and upon nuclear envelope breakdown, the CPC reorganizes into a ring encircling the short arm axis of the bivalent, and forms the RC by recruiting other components [12,13,18]. To determine if this structure properly forms on bivalents with multiple short arm regions (resulting from excess COs) (Fig 1D), we compared the organization of the CPC and other RC components on wild-type and meT7 prometaphase bivalents.
The three wild-type bivalents in the meT7 strain formed single whole rings of AIR-2 (75/75 had whole rings at 15˚C; 74/75 at 25˚C). However, while AIR-2 localized to meT7 fusion bivalents, it was sometimes improperly shaped: 72.0% of meT7 bivalents had single rings at 15˚C, but other meT7 bivalents formed either slightly (18.7%) or severely (9.3%) mispatterned structures (Fig 2A and 2B; S4 Fig; "slight mispatterning" denotes that ring components are localized to a single plane but are not confined to a single ring around the outside of the bivalent, while "severe mispatterning" indicates that components form more complex structures). Additionally, we also observed similar types of ring mispatterning along the two chromosome fusion mnT12 (S5A Fig). BIR-1 Survivin , another CPC component, colocalized with AIR-2 on all ring structure types (S6 Fig), suggesting that the pattern of AIR-2 localization reflects assembly of the entire CPC. Similar to the increase in COs and chiasmata at 25˚C along meT7 (77% meT7 with >2 COs at 20˚C versus 90% meT7 with >2 COs at 25˚C; Fig 1D), these CPC patterning defects on the meT7 bivalent increased in number (21/75 total mispatterned rings at 15˚C versus 48/75 total mispatterned rings at 25˚C, P = 0.0001; Fisher's exact, two-tailed) and severity (observed severely mispatterned ring frequency was 9.3% at 15˚C versus 33.3% at 25˚C, P = 0.0006; Fisher's exact, two-tailed) with increased temperatures (Fig 2A), suggesting that increased CO numbers increase the likelihood of CPC mispatterning.
We next assessed the localization of RC components that are dependent on the CPC for targeting, and found that both SUMO ( Fig 2B) and KLP-19 ( Fig 2C) localize on meT7 bivalents. Therefore, although the CPC is mispatterned, other RC components can still target to the bivalents. Linescans across SUMO-stained bivalents showed that, in contrast to the single tight ring peak and bilobed bivalent structure characteristic of wild-type bivalents, meT7 bivalents have broad variation in both RC and chromosome shape ( Fig 2D). Importantly, although RCs also form in Meiosis II around the interface between sister-chromatids, the RC patterning defects were more prevalent on Meiosis I meT7 bivalents (45.3% of all MI meT7 bivalents, compared to 12.0% of MII chromosomes, p<0.0001, Fisher's exact test, two-tailed; Fig 2E). This result is consistent with the idea that the RC defects are primarily caused by excess COs between homologous chromosomes, rather than general problems with RC formation in the meT7 strain.
Next we assessed kinetochore organization on meT7 bivalents. In C. elegans, kinetochore proteins coat the holocentric chromosomes in meiosis, forming a cup-like structure around each end of the bivalent [31,32]. Although microtubules run along the sides of chromosomes and do not form end-on kinetochore attachments [12], kinetochores help orient bivalents within the spindle [13], such that the long axis of the bivalent is parallel to the long axis of the spindle; this arrangement ensures that the two homologs are pointed towards opposite spindle poles, with the shortarm axis, where cohesion will be released, in the middle. We found that two kinetochore components, BUB-1 and SEP-1 Separase , still localized to the meT7 fusion chromosome bivalent (S7A Fig  and S7B Fig), suggesting that the targeting of kinetochore proteins to meT7 was not affected. However, the fusion chromosome bivalents are often misshapen and many do not have a clear long axis (S7C Fig and example images throughout the paper). This altered bivalent shape would therefore prevent meT7 from being able to properly assemble the two distinct cup-like kinetochores, causing orientation problems for the meT7 bivalent within the spindle.

Fusion chromosome bivalents can form multiple Ring Complexes that act independently of one another
Next, we wanted to characterize the mispatterned meT7 RC structures. In some of our images there appeared to be multiple distinct RCs forming on a single meT7 bivalent (e.g. Fig 2A, row  3; Fig 2B, row 3), which could reflect the ability of each chiasma to organize its own RC. However, since many of the mispatterned RCs appeared to be complex structures, it was difficult in many cases to determine if a particular RC was comprised of multiple rings close together, or (A) Immunofluorescence of AIR-2 localization in fixed wild type (N2) or meT7 oocytes. Single-chromosome zooms show that AIR-2 (red) localizes in a whole ring shape encircling normal bivalents in both strains, while its localization is either whole ring-like or mispatterned on meT7 fusion bivalents (yellow arrows). Note that the zoomed images throughout this figure are partial projections, chosen to highlight individual chromosomes. Quantification in graph to the right of the images (N = Whole ring/Slight/Severe: N2 at 15˚C N = 31/0/0, N2 at 25˚C N = 44/0/0, meT7 at 15˚C N = 54/14/7, meT7 at 25˚C N = 27/23/25). This analysis shows that meT7 mispatterning increases with increased temperatures. For overall mispatterning (slight + severe): 15˚C (21/75) versus 25˚C (48/75), P = 0.0001; Fisher's exact, two-tailed. For severe mispatterning: 15˚C (7/75) versus 25˚C (25/75), P = 0.0006; Fisher's exact, two-tailed. (B) SUMO (green) co-localizes with AIR-2 (red) in all ring structure types on both fixed wild type (N2) and meT7 fusion bivalents; RCs on meT7 bivalents denoted with yellow arrows. (C) KLP-19 (kinesin 4) (red), localizes to both "whole ring" and "mispatterned" meT7 RCs in fixed oocytes. (D) Linescans across bivalents of fixed oocytes show that both the bivalent and RC have wider spread and higher variance in meT7 fusion bivalents as compared to wild-type bivalents. (N = 25, all categories). (E) Quantification of ring mispatterning in metaphase I bivalents (34/75, 45.3%) and metaphase II sister chromatid pairs (9/75, 12.0%). Metaphase I bivalents have much higher rates of mispatterning as compared to metaphase II sister chromatid pairs (P<0.0001, Fisher's exact test, two-tailed). All scale bars = 2.5μm. https://doi.org/10.1371/journal.pgen.1009001.g002 instead represented a single intertwined structure. Therefore, to distinguish between these possibilities, we performed an "RC stretching assay." This assay exploits our previous finding that under extended metaphase arrest, RCs begin to stretch away from the chromosomes towards microtubule plus ends, which we interpret to mean that they contain a plus-end-directed activity that in normal conditions would provide chromosomes with plus-end-directed forces [19]. For the purposes of the current study, we thought that this behavior might spatially separate distinct rings from one another, enabling us to distinguish and quantify the total number of rings formed on each meT7 fusion bivalent. Additionally, we performed this "RC stretching assay" on monopolar spindles, in which the microtubule minus ends are organized at a central pole and the plus ends radiate outward forming an aster. We reasoned that this feature might enable us to determine if the meT7 ring structures were functional (i.e. whether they were able to stretch towards the outside of the aster, thus exerting plus-end forces). Consistent with previous work [19], we found that in this RC stretching assay, rings on normal bivalents typically stretch off as one entity in a single direction towards microtubule plus ends (31/38), with 7/38 of the observed bivalents having two stretches (Fig 3, rows 1 and 2). In contrast, on many meT7 bivalents, it was clear that there was more than one RC, often with two (27/38) or rarely three (2/38) separate entities stretching off of the bivalent (Fig 3, rows 5 and 6). Notably, the frequency of more than one stretching RC on the meT7 fusion chromosome (76% at 15˚C) is In this assay, monopolar spindles were generated by depleting the force-generating motor KLP-18 [12,68] and metaphase I arrest was achieved by depleting anaphase promoting complex component EMB-30. Zooms are partial projections, chosen to highlight individual chromosomes. RCs on monopolar spindles under prolonged metaphase arrest tend to stretch in a plus end-directed manner, towards the outside of the microtubule aster. RCs on normal bivalents tend to stretch as mostly one unit in a single direction, while meT7 bivalents can have multiple RCs that stretch in independent directions. Yellow arrows denote stretching RCs. Scale bars = 2.5μm.
https://doi.org/10.1371/journal.pgen.1009001.g003 close to the frequency of more than one CO along meT7 (77% at 20˚C; Fig 1C). This result suggests that multiple distinct RC structures can form at different chiasmata on the same bivalent.
Interestingly, when multiple RCs are present, they stretch towards microtubule plus ends but end up oriented along different microtubule bundles, suggesting that these RCs are acting independently of one another; in the context of a bipolar spindle where microtubule bundles are not all in the same orientation, this could result in a single chromosome being pulled in opposite directions. Furthermore, unlike on normal bivalents, 3/38 rings on meT7 bivalents were not stretching in any direction (the "zero stretches" category in Fig 3, row 3), despite normal bivalents in the same spindle having stretching rings, suggesting that mispatterning may sometimes affect RC function. Together, these results suggest that the mispatterned RCs on meT7 may not be able to provide chromosomes with normal plus-end-directed forces.

Fusion chromosome bivalents show defects in metaphase alignment
To investigate the possibility that meT7 bivalents are experiencing abnormal forces, we asked whether mispatterning of the RCs in the meT7 bivalents had functional consequences on the alignment of those bivalents on bipolar spindles. We therefore evaluated spindles where the three normal bivalents had aligned at the metaphase plate and scored the position of the meT7 fusion bivalent. We found that while meT7 bivalents with single whole rings aligned with the other bivalents in 92% (46/50) of these spindles, only 62% (31/50) with mispatterned rings successfully aligned ( Fig 4A); this correlation suggests that the alignment defects are not caused by the large size of meT7 but by defects in RC organization, including the acquisition of multiple RCs on the meT7 bivalent (Fig 3). Since this is fixed imaging, we do not know if this phenotype represents a complete failure of meT7 to align, or just a delay compared to the other bivalents. However, either way our data suggest that the acquisition of multiple crossovers on the meT7 bivalent affects the efficiency of chromosome congression.
Building on this finding, we assessed the position of chromosomes on monopolar spindles as a read-out of plus-end directed forces ( Fig 4B); in this context, bivalents that are able to generate normal plus-end forces migrate away from the center of the aster [12]. To define this center, we used ASPM-1 as a pole marker, and measured the distance from the center of each bivalent to the center of the ASPM-1 staining (Fig 4B and 4C). Unlike the RC stretching assay with monopolar spindles (Fig 3), this experiment does not involve an extended metaphase arrest so the RCs retain their original morphology (Fig 4). This analysis revealed that meT7 bivalents with mispatterned rings do not tend to migrate as far away from the pole as the three wild-type bivalents (P = 0.002; Mann-Whitney U test, two tailed, Fig 4C). Moreover, in two extreme cases (out of 12), the fusion bivalents remained stuck at the center of the monopole (Fig 4B, bottom row). Importantly, this defect appeared to be caused by the chromosome patterning defects and not by the large size of the fusion bivalent, since meT7 bivalents with single rings exhibited normal chromosome positioning (Fig 4B and 4C). Together, these findings suggest that the mispatterned RCs on meT7 bivalents are not able to provide these bivalents with normal plus-end forces, potentially impacting their ability to achieve proper metaphase alignment. Since meT7 bivalents often have multiple RCs that can function independently of each other (Fig 3), we postulate that these RCs could provide bivalents with forces in opposing directions thereby reducing the efficiency of chromosome movement.

Fusion chromosome bivalents show defects in chromosome segregation that carry over into Meiosis II
Given the defects in the structure and alignment of meT7 bivalents, we next wanted to determine if having excess COs impacts chromosome segregation during Meiosis I and II. First, we imaged anaphase I to assess the ability of homologous chromosomes to segregate. Notably, we found that in early anaphase I, the meT7 fusion bivalent was often present in the middle of the spindle in cases where the three wild-type bivalents were segregating, suggesting that its segregation was delayed relative to the other chromosomes. However, the frequency of this phenotype did not appear to be affected by temperature, as meT7 bivalents showed comparable segregation delays in early anaphase I spindles at both 15˚C (44/75, 58.7%) and 25˚C (47/75, 62.7%) (P = 0.738, Fisher's exact test, two-tailed, Fig 5A). In contrast, in late anaphase I, meT7 bivalents had chromatin bridges that increased at higher temperatures where meT7 experiences elevated COs (17/75, 22.7% at 15˚C vs. 34/75, 45.3% at 25˚C; P = 0.0056, Fisher's exact test, two-tailed), while wild-type bivalents segregated normally at all temperatures ( Fig 5A). Similarly, the mnT12 two chromosome fusion bivalents also experience late anaphase I chromatin bridges (S5B Fig). Collectively, these results suggest that the early anaphase I defects are not due to increases in CO numbers, and may instead be due to the large size of the fusion chromosome. Conversely, extra COs between homologous chromosomes likely cause chromatin bridges in late anaphase I. Consistent with this hypothesis, we rarely observed chromatin bridging in anaphase II when sister chromatids rather than homologous chromosomes were segregating (2/65, 3.1%; Fig 5C).
To determine if the anaphase I defects had lasting consequences, we examined oocytes that progressed to Meiosis II. Notably, we observed a range of severe defects at 15˚C, from DNA connecting the polar body to chromatin in the MII spindle ("tethered polar body" category; Fig 5B, arrows) to either gain or loss of the entire meT7 bivalent. Similar to the chromosome bridging observed in MI, these defects increased at the higher temperature of 25˚C (20%, 18/ 90 at 15˚C versus 36.6%, 33/90 at 25˚C; P = 0.02, Fisher's exact test, two-tailed; Fig 5B), suggesting that these defects were the result of excess COs between meT7 chromosomes. We also observed Meiosis II defects in the mnT12 strain (S5C Fig), demonstrating that these observations are not specific to meT7. The frequency of the severe defects seen for meT7 at Meiosis II (33/90, 36.6% at 25˚C) is close to the frequency of embryonic lethality seen in the meT7 strain at 25˚C (27.6%; Table 1), suggesting the majority of the embryonic lethality is due to a combination of aneuploidy and/or polar body extrusion defects.

Persistent chromatin bridges can be recognized and resolved by the oocyte
Despite the fact that we observed meiotic chromosome segregation defects (Fig 5), the embryonic lethality of the meT7 strain is surprisingly low ( Table 1). The meT7 strain exhibits 27.6% and SUMO to mark the RCs (red). Alignment of the fusion chromosome was assessed on spindles where the three normal bivalents were aligned; note that alignment was assessed by rotating the images in 3D, since this is harder to discern in projection images, like those shown. A diagram showing our quantification scheme and our quantification is below the images. 92% of meT7 bivalents with a whole ring aligned with the three normal bivalents, compared to 62% of meT7 bivalents with mispatterned rings (N = 50, all conditions). Scale bars = 2.5μm. (B,C) Examples and quantification of fused chromosome bivalent position in fixed oocytes; meT7 bivalents indicated by arrows. Monopolar spindles were generated by depleting KLP-18, and the center of the aster was determined by staining with spindle pole marker ASPM-1 and determining the center of the ASPM-1 signal. The distance between each of the bivalents and the center of the ASPM-1 signal was measured using Imaris, and the distance of the fused chromosome bivalent was compared to the average of the three normal bivalents' distance; normal bivalents were also assessed with this same measure (comparing the position of each individual bivalent with the average of the three normal bivalents on the same spindle). On the graph, blue dots denote wild-type bivalents, and green dots denote meT7 bivalents; these measurements were derived from monopolar spindles where the meT7 bivalent was present in a whole ring (left side of graph) or where the RC was mispatterned (right side of graph). meT7 bivalents with whole rings (N = 16) tended to be positioned similarly to normal bivalents on the monopole (P = 0.90; Mann-Whitney, two-tailed), while bivalents with mispatterned rings (N = 12) showed much more variation in their distance from the monopole as compared to the normal bivalents (P = 0.002; Mann-Whitney, two tailed). Scale bars = 5μm.
https://doi.org/10.1371/journal.pgen.1009001.g004 embryonic lethality at 25˚C ( Table 1), indicating that most oocytes are able to generate viable progeny even in the presence of multiple COs (at 25˚C, >90% of meT7 chromosomes have �2 COSA-1 foci; Fig 1C) and chromosome segregation defects (at 25˚C,~50% of meT7 anaphase I nuclei have chromatin bridges; Fig 5). Similarly, Meiosis II oocytes have a lower frequency of errors than was observed during Meiosis I. Therefore, it is possible that the anaphase I bridges could in some cases be resolved (i.e. not persist into Meiosis II) in a manner that enables viability. To investigate this possibility, we performed live imaging of the meiotic divisions using a meT7 strain expressing GFP-tubulin and mCherry-histone to mark microtubules and chromosomes, respectively. Confirming our fixed imaging results, we found that a majority of anaphase I spindles displayed extended chromatin bridges (8/14, 57%) (Fig 6). In some cases (5/8) these bridges persisted and either 1) the first polar body was tethered to the developing Meiosis II spindle (S1 Movie), or 2) meT7 was not able to segregate in anaphase I, and the fully retained bivalent segregated with a chromatin bridge in anaphase II (S2 Movie). This supports the conclusion that oocytes face severe chromosome segregation defects in the presence of supernumerary crossovers. However, our live imaging also revealed instances where anaphase I DNA bridges were resolved as the oocyte progressed to Meiosis II (3/8; Fig 6; S3 Movie). Thus, oocytes appear to have mechanisms to correct and resolve chromosome segregation defects.
One protein that we hypothesized might contribute to error correction is LEM-3, a late-acting nuclease proposed to resolve chromatin connections in mitotic and meiotic anaphase but that is not thought to affect either crossover numbers or chromosome segregation in the absence of persistent recombination intermediates [33,34]. Therefore, we depleted lem-3 with RNAi in the meT7 strain and assessed chromosome segregation. In support of our hypothesis, we found that lem-3-depleted meT7 oocytes had higher incidences of bridges and segregation defects. Specifically, there was a modestly higher proportion of anaphase I bridges at 25˚C (25/49, 51.0%, in control vs. 27/37, 72.97%, in lem-3 RNAi; P = 0.047, Fisher's exact test, two-tailed; Fig 7A). Importantly, these defects persisted into Meiosis II in higher proportions compared to oocytes without lem-3 depletion, with MII spindles having significantly increased instances of chromosome segregation issues (categorized as having either tethered polar bodies or aneuploidy) at 25˚C, the temperature that causes increased CO events along meT7 (15/41, 36.6% in control vs. 20/28, 71.4% in lem-3 RNAi; P = 0.007, Fisher's exact test, two-tailed; Fig 7B). Live imaging of lem-3 depleted meT7 oocytes corroborated these findings, showing an increase in chromosome segregation defects, including two cases in which chromatin bridges appeared to fragment (2/ 13, 15%), and one failure to extrude a Meiosis I polar body (1/13, 7%) (Fig 7D; S1 Table). In five cases, the meT7 chromosome was unable to segregate in anaphase I and was subsequently extruded into the first polar body (5/13, 38%) ( Fig 7D). Further, we saw a 6.3% increase in embryonic lethality in meT7 treated with lem-3 RNAi (27.6% meT7 untreated vs. 33.9% meT7 Immunofluorescence of anaphase I chromosome segregation in fixed wild-type (N2) and meT7 oocytes; shown are DNA (blue), microtubules (green), and AIR-2 (red), with quantification to the right of the images. All chromosomes (N = 75 for early and mid-to-late anaphase) segregated without errors in wild type (N2) oocytes (rows 1 and 2), but meT7 bivalents were often found in the center of the spindle in early anaphase when the normal bivalents were segregating (row 3, indicated with arrow, "early anaphase quantification" graph), suggesting a segregation delay. This phenotype was not different in meT7 oocytes between 15˚C and 25˚C (P = 0.738, Fisher's exact test, two-tailed). In mid-to-late anaphase, 17/75 (22.7%) of meT7 bivalents showed extended chromatin bridging, indicated by arrows, at 15˚C, which increased to 34/75 (45.3%) at 25˚C (rows 4 and 5, "late anaphase quantification" graph; P = 0.0056, Fisher's exact test, two-tailed). (B) Metaphase II spindles in meT7 oocytes. While all N2 metaphase II spindles were euploid (N = 90), at 15˚C 18/90 (20.0%) of meT7 spindles were either aneuploid (rows 1 and 2) or had DNA tethered to the first polar body (indicated with arrows, rows 3 and 4). At 25˚C this frequency was 33/90 (36.6%), suggesting that these defects increase at higher temperature (P = 0.02, Fisher's exact test, two-tailed). (C) Quantification of meT7 chromatin bridging in anaphase I vs. anaphase II. 22 PLOS GENETICS + lem-3 RNAi; P = 0.0008; Fisher's exact test, two-tailed; Table 1) that is a larger increase than the 0.9% that is observed in N2 treated with lem-3 RNAi (0.5% N2 untreated vs. 1.4% N2 + lem-3 RNAi; P = 0.02; Fisher's exact test, two-tailed; Table 1). Together, these results suggest that LEM-3 plays a role in correcting meT7 chromosome segregation defects during oocyte meiosis.
Furthermore, our analysis suggested another potential mechanism that could contribute to error correction in cases of chromatin bridges caused by excess COs. Specifically, we noticed that when oocytes had a chromosome tethered to the polar body that persisted into anaphase II, the segregating sister chromatid connected to the polar body via a chromosome tether was located closer to the cortical side (23/26 cases; 3 examples shown in Fig 7C), suggesting that the sister chromatid associated with the chromosome tether may be expelled with the second polar body, leaving the oocyte with the untethered sister chromatid and therefore euploid (i.e. with the correct number of chromosomes). This fixed imaging result is supported by our live imaging, where we found that in 4/5 movies containing persistent chromosome tethers in anaphase II (Fig 6A v-vi and 6B v-vi), the sister chromatid associated with the chromosome tether was extruded with the second polar body during Meiosis II (Fig 6A v and 6B v). Currently, it is unclear whether the oocyte actively recognizes the sister chromatid with the chromosome tether or whether the chromosome tether biases the direction of spindle rotation towards the cortex, thereby facilitating elimination of tethered sister chromatids by their deposition into the polar body. In either case, this mechanism could serve to eliminate improperly segregated chromosomes from the resulting gamete.

Discussion
In summary, our work provides molecular insights into how crossover limitation promotes proper chromosome congression and segregation in the oocyte. We found that increased COs cause defects in bivalent organization that likely impact the ability of chromosomes to achieve metaphase alignment (Fig 7E). In normal meiosis, bivalents organize around a single CO into a structure with long and short arms, and then a ring complex (RC) composed of essential meiotic proteins forms around the short arm interface. In contrast, an increased number of COs leads to the designation of additional short arms and subsequent formation of extra RCs. Interestingly, when multiple RCs form along a single chromosome, they all appear to be functional. The inability to inactivate extra RCs in C. elegans meiosis appears to have consequences, as there are chromosome congression errors that may arise from differential forces exerted on bivalents. This result highlights the importance of proper RC patterning, and demonstrates that the structure of this protein complex affects its function. Moreover, our studies revealed that increasing CO number has consequences for homolog segregation, as bivalents with excess crossovers have frequent chromatin bridging during the first meiotic division (Fig 7E).

Crossover limitation promote accurate chromosome segregation
Strong CO interference, where there is only~1-3 COs per chromosome pair per meiosis, is observed in several model systems, including fruit flies, worms, and mice [1]. Even in human and anaphase Meiosis II (MII) in live wild-type and meT7 oocytes expressing GFP::TBB-2 and mCherry:HIS-11 to mark microtubules and histones, respectively. Examples of meT7 chromosome segregation phenotypes shown in i-vi. White arrowheads point to meT7 fused chromosomes as they fail to segregate in anaphase I (iii), yellow arrowheads indicate anaphase I meT7 chromatin bridges between segregating chromosomes (iv, v, vi), and blue arrowheads point to chromatin tethers between the MI polar body (MPB) and the anaphase II spindle (v, vi). White arrows indicate lagging chromosome(s) in anaphase II (ii). Time zero is the time at which chromosome segregation appears to begin. Scale bars = 5μm. (B) meT7 chromosome segregation phenotype descriptions and observation frequency. Of the 8 meT7 oocytes that had chromatin bridges at anaphase I, 3/8 bridges were able to resolve prior to the end of anaphase I. For the remaining 5 meT7 oocytes that were not able to resolve their anaphase I bridges, 4/8 extruded the tether (with the associated chromatid) into the polar body at anaphase II and 1/8 exhibited persistent tethers through anaphase II.
https://doi.org/10.1371/journal.pgen.1009001.g006 oocytes, CO interference is strong, with 1-2 COs per homolog [35]. Although CO interference was initially observed over a century ago in Drosophila [2,3] and was subsequently found to be widely conserved among many model systems [1], the consequences of extra interfering COs along a pair of homologous chromosomes has not been directly determined. Our study is the first, to our knowledge, in any organism to assess these consequences. Our findings using fusion chromosomes in C. elegans suggest that CO interference helps promote proper meiotic chromosome alignment by limiting homolog pairs to one crossover per bivalent under normal circumstances (thereby preventing the difficulties that ensue when bivalents assemble multiple RCs). Moreover, we also observed severe chromosome segregation defects in the presence of extra COs. These segregation errors could have multiple causes. First, the inability to properly congress chromosomes could be one contributing factor; if anaphase is initiated when the meT7 bivalent is not aligned, this could cause errors. However, the defective bivalent structure that arises in the presence of extra COs likely also causes other problems, beyond the RC defects. Second, there could be defects resolving recombination intermediates when extra COs are present, thereby resulting in the thick anaphase I bridges that we observe in our study. Another possibility is that limiting COs could promote the creation of a chromosome structure that facilitates the loss of cohesion in the correct domain. Since AIR-2 is thought to phosphorylate the cohesin subunit REC-8 to enable cleavage by separase [11,36], the AIR-2 mispatterning that occurs in the presence of extra crossovers could lead to defects in cohesion release, which may cause chromosome bridging. Finally, it is also possible that when a bivalent has multiple crossovers, this may involve 3-4 chromatids (instead of the usual two chromatids in a single crossover), which may result in chromosome bridging and segregation issues.
Overall, the ability to limit crossovers is important for the faithful execution of meiosis, which may represent a driving force that has contributed to the conservation of CO interference in many organisms. In the future, it will be interesting to investigate how supernumary COs affect meiotic chromosome segregation in other organisms. This is important since C. elegans oocytes have a number of features not shared by all organisms, including holocentric kinetochores [37], reliance on the RCs for chromosome congression [12,19], and a unique chromosome segregation mechanism [13]. Given the conservation of CO interference, we predict that excess COs may also affect the fidelity of chromosome segregation in other organisms, but this remains to be experimentally determined. Moreover, it will also be important to investigate this question using normal length chromosomes, as it is possible that some the defects we documented for meT7 and mnT12 may not be exclusively caused by excess COs, and that other properties of the fused chromsomes may contribute.

Crossover interference and holocentric chromosomes
In comparison to humans and other model organisms, C. elegans exhibits extremely strong CO interference known as complete CO interference, where one (and only one) CO is formed per chromosome pair [4]. One potential reason why CO interference may be so robust in C. elegans is that they have holocentric chromosomes, which assemble centromere and kinetochore proteins along their entire length. In comparison to monocentric organisms, most holocentric organisms experience complete CO interference, suggesting that there is a strong selection bias for this highly stringent form of CO interference in these organisms [38,39].
In contrast to monocentric chromosomes, holocentric chromosomes are faced with the unique challenge that microtubules could theoretically attach to any part of the bivalent surface, rather than associating in a manner that promotes the biorientation of homologous chromosomes [40]. One way that C. elegans counteract this problem is the asymmetric positioning of a single crossover along the length of a chromosome; this generates a single point of organization along the length of each chromosome, which patterns the chromosome in a manner that facilitates chromosome biorientation, alignment, release of cohesion, and segregation [40,41]. Moreover, a recent C. elegans study has found that the asymmentric placement of the crossover along the chromosome is important for proper chromosome remodeling and segregation [42]. Perhaps the acquisition of holocentric chromosomes requires extreme crossover interference in addition to asymmetric placement of a crossover along the chromosome length. Future experiments or analysis of the evolution of holocentric organisms may parse out the correlation between holocentricity and strong crossover interference.

Mechanisms exist to correct errors caused by supernumary crossovers
Our studies also uncovered two mechanisms that counteract defects caused by excess COs and likely account for some of the relatively low embryonic inviability in fusion chromosome strains. We found that depletion of the conserved nuclease LEM-3/Ankle1 increases the frequency and persistence of anaphase bridging, consistent with its previously proposed role in processing erroneous recombination intermediates and correcting meiotic errors [34]. Given the role of LEM-3 in processing anaphase bridges from persistent recombination intermediates, it is possible that some of the chromatin bridges we are observing from excess COs along meT7 are unresolved crossover recombination intermediates. In the future, it will be interesting to determine whether additional proteins assist in processing DNA tethers. Given the relatively low embryonic lethality of the meT7 fusion chromosome strain in contrast to the frequency of DNA tethers and other segregation errors, it is likely that other correction mechanisms exist. In addition to its role in meiotic recombination, the BLM/HIM-6 helicase (which localizes to interfering CO sites [43]) has been shown in multiple systems to localize to and to help resolve DNA bridges during both mitotic and meiotic chromosome segregation [44][45][46][47][48][49]. Similarly, the nucleases MUS-81, SLX-1 and SLX-4 have genetic interactions with both LEM-3 [34] and HIM-6 [50] and may act combinatorially to resolve DNA bridging due to excess COs. Further, topoisomerase II has been shown to remove DNA tethers from heterochromatin regions that connect achiasmate homologs in Drosophila [51]. Given the conservation of the nuclease LEM-3/Ankle1, these mechanisms that counteract the effects of multiple COs may be augmented in organisms that occasionally experience more than one CO along a chromosome.
Although nucleases can serve as a generalized solution to resolve erroneous DNA connections, we also discovered a second mechanism specific to oocytes. Since each CO involves only two of the four sister chromatids, there could be cases where one sister chromatid of meT7 is tethered to the polar body due to the presence of an unresolved chiasma, but the other sister chromatid is unaffected since it was not involved in the CO that caused the problem. Thus, if the tethered sister chromatid is expelled from the oocyte in anaphase II and the normal chromatid is retained, the egg would be euploid. We found that in oocytes where a chromosome tether was not resolved in anaphase I, the tethered sister chromatid was preferentially extruded from the oocyte in anaphase II, revealing another mode of error correction. This mechanism is made possible by the asymmetric nature of the oocyte divisions, in which half of the genetic material is discarded into the polar body during each division. This asymmetry has been shown to allow the preferential elimination of univalent chromosomes in C. elegans [52], and here may provide a second opportunity for correction of errors associated with excess COs. It is interesting to speculate on whether the oocyte spindle actively recognizes the chromosome tether, or whether the chromosome tether passively biases the rotation of the spindle, positioning the tethered sister chromatid adjacent to the cortex, where it will be extruded into the second polar body. Currently, little is known about how spindle rotation is influenced in C. elegans; however, previous work has shown that the minus-end binding protein ASPM-1 is brighter on the spindle pole that rotates towards the cortex, suggesting that spindle asymmetry could play a role [53] similar to spindle asymmetry and selfish centromeres in mouse oocytes [54]. Going forward, it would be interesting to ask whether unresolved DNA connections influence spindle factors to allow for preferential spindle rotation as a form of meiotic drive.

Licensing of Ring Complex formation by interfering crossovers
Our results further establish how the coordination and interplay between recombination and chromosome segregation are critical for the maintenance of genomic integrity through generations. We observed that the occurrence of multiple Class I interfering crossovers leads to the formation of multiple RCs. Moreover, we found a strong correlation between interfering CO number and eventual RC number along a single chromosome. This result leads to the compelling hypothesis that an interfering CO site may be a licensing event for RC formation. Through either its specific DNA conformation or the recombination machinery associated with it, a CO may designate a location along the chromosome for RC components to assemble. Previous experiments have already found that Aurora B kinase (AIR-2) localizes to sites of COs [8]. Further, the HIM-6/BLM helicase has been found to localize to interfering CO sites and help maintain chiasmata until metaphase I [47,49]. Future experiments exploring whether there is direct interaction between recombination machinery and RC components may establish this connection between CO sites and RC formation.
Our study finds that excess Class I interfering COs affect chromosome segregation in C. elegans. In contrast, the occurrence of excess Class II non-interfering crossovers does not affect meiotic chromosome segregation in Arabidopsis thaliana, Schizosaccharomyces pombe, and Aspergilis nidulans [55][56][57][58]. C. elegans normally do not form Class II non-interfering COs, but in a mutant for the helicase RTEL-1, non-interfering COs are now formed alongside interfering COs [59]. Notably, in rtel-1 mutants, this 16-fold increase in non-interfering COs results in only a very modest effect on embryonic lethality (3%; [60]), suggesting that an excess of non-interfering COs does not affect chromosome segregation. The differences between the effects of excess Class I versus Class II COs on chromosome segregation suggests the possibility that interfering COs are differentiated from non-interfering COs at the level of licensing a location for RC formation.
Overall, our study provides new molecular insights into how excess interfering COs affect chromosome congression and segregation. Further, our results indicate the existence of mechanisms to assist with correcting errors associated with the formation of excess COs, thereby lending important insight into how some organisms, such as Saccharomyces cerevisiae, can tolerate more than two interfering COs per chromosome. Future studies investigating why certain organisms can experience higher levels of interfering COs will lend further insight into these critical mechanisms for maintaining genomic integrity through generations.

C. elegans strains, genetics, and culture conditions
All strains are from the Bristol N2 background and were maintained at 15˚C or 20˚C and crossed at 20˚C under standard conditions. Temperatures used for specific experiments are indicated below. For all experiments with meiotic mutants, homozygous mutant worms were derived from balanced heterozygous parents by selecting progeny lacking a dominant marker (Unc and/or GFP) associated with the balancer. For experiments marked 25˚C, L4 worms were shifted to 25˚C 24-48 hours prior to dissection. For lower temperature controls, either 15˚C or 20˚C was used; these temperatures are both well below the identified threshold for destabilization of the synaptonemal complex (�25˚C; [61]), which has a role in crossover number and formation [5,62].
The following strains were used in this study:

RNAi treatments
RNAi by feeding was performed as previously described [19]. Worms were synchronized at the L1 phase by bleaching adults and allowing resultant eggs to hatch on unseeded NGM plates at 20˚C for 20-24 hrs. Synchronized L1s were then washed off of the unseeded NGM plates with M9 and placed on NGM+1mM IPTG+100μg ampicillin plates that were poured within 30 days of use and freshly seeded one day before use with clones picked from the Ahringer Lab RNAi feeding library [64] or the empty L4440 vector (referred to as "control RNAi" in figures and text). The RNAi plates with L1s were then placed at 15˚C and grown to adulthood. For RNAi experiments performed at 25˚C, L4 worms were shifted to 25˚C for 40-48 hours prior to dissection.

Immunofluorescence for late meiotic prophase I
Immunofluorescence was performed as in [5]. Gonads from adult worms at 18-24 hours post-L4 stage were dissected in 1x egg buffer with 0.1% Tween on VWR Superfrost Plus slides, fixed for 5 min in 1% paraformaldehyde, flash frozen with liquid nitrogen, and then fixed for 1 minute in 100% methanol at -20˚C. Slides were washed 3 x 5 min in 1x PBST and blocked for one hour in 0.7% BSA in 1x PBST. Primary antibody dilutions were made in 1x PBST and added to slides. Slides were covered with a parafilm coverslip and incubated in a humid chamber overnight (14-18 hrs). Slides were washed 3 x 10 min in 1x PBST. Secondary antibody dilutions were made at 1:200 in 1x PBST using Invitrogen goat or donkey AlexaFluor labeled antibodies and added to slides. Slides were covered with a parafilm coverslip and placed in a humid chamber in the dark for 2 hrs. Slides were washed 3 x 10 min in 1x PBST in the dark. All washes and incubations were performed at room temperature, unless otherwise noted. 2 μg/ml DAPI was added to slides and slides were subsequently incubated in the dark with a parafilm coverslip in a humid chamber. Slides were washed once for 5 min in 1x PBST prior to mounting with Vectashield and a 20 x 40 mm coverslip with a 170 ± 5 μm thickness. Slides were sealed with nail polish immediately following mounting and then stored at 4˚C prior to imaging. All slides were imaged (as described below) within two weeks of preparation. The following primary antibody dilutions were used: rabbit anti-GFP (1:1000) [6]; chicken anti-GFP (1:1000) (Abcam 13970); guinea pig anti-SYP-1 (1:200) [62]; rabbit anti-HIM-3 (1:200) [65]; chicken anti-HTP-3 (1:500) [66]; rabbit HTP-1/2 (1,500) [8].

Immunofluorescence for prometa-, meta-and anaphase I and II
Immunofluorescence was performed as in [67]. Briefly, adult worms were picked into a drop of M9 media on poly-l-lysine coated slides (Fisher Scientific), cut in half to release oocytes, covered with a coverslip and frozen for 7 minutes in liquid nitrogen. The coverslip was quickly cracked off, and slides were fixed in -20˚C methanol for 35 minutes. Samples were then rehydrated in PBS and blocked in AbDil (PBS with 4% BSA, 0.1% Triton X-100, 0.02% Sodium Azide) for one hour at room temperature, and then washed with PBST (PBS + 0.1% Triton X-100). Primary antibodies were diluted in AbDil, applied to samples and incubated overnight at 4˚C. Samples were washed with PBST, and secondary antibodies were diluted in AbDil, applied to samples, and incubated for 2 hours at room temperature. Slides were washed and incubated with Hoechst 33342 (Invitrogen) at 1:1000 in PBST for 10 minutes at room temperature. Slides were washed a final time and mounted in 0.5% p-phenylenediamine in 90% glycerol, 20mM Tris, pH 8.8, sealed with nail polish, and stored at 4˚C prior to imaging.

Fixed imaging
For Fig 1, S1 Fig, S2 Fig and S3 Fig, IF slides were imaged at 512 x 512 pixel dimensions on an Applied Precision DeltaVision microscope using a 60x objective (NA = 1.42) with 1.5x optivar. Images were acquired as Z-stacks at 0.2 μm intervals and deconvolved with Applied Precision softWoRx deconvolution software. For quantification of GFP::COSA-1 foci, nuclei that were in the last 4-5 rows of late pachytene and were completely contained within the image stack were analyzed. Foci were quantified manually from deconvolved three-dimensional stacks. meT7 chromosomes in pachytene nuclei were identified based on size. Regardless of temperature (20˚C or 25˚C), all three wild-type/non-fused chromosomes in the nuclei of the meT7 strains contained only one GFP::COSA-1 focus per chromosome (n = 30), which is consistent with all chromosomes in wild-type strains and the wild-type/unfused chromosomes in mnT12 (X;IV) fusion chromosome strain (this study, [5,6]). Given the consistency of COSA-1 counts for the 3 wild-type/non-fused chromosomes in the meT7 strains, the number of COSA-1 foci per meT7 was calculated by subtracting 3 from the total number of COSA-1 foci per nucleus. For visualization and quantitation of chiasmata (Fig 1D), individual meT7 bivalents from diakinesis nuclei in -2, -3, or -4 oocytes were identified based on size, cropped, and rotated in three-dimensions using Imaris (Bitplane/Oxford Instruments) three-dimensional rendering software. Scoring of chiasmata was based primarily on HTP-3 (chromosome axis) or HIM-3 (chromosome axis) and DAPI staining, as GFP::COSA-1 dissociates from chromosomes during progression through the diakinesis stage. For quantification of DAPI body counts during diakinesis, counts were performed as described in [6]. For Fig 1B and 1C, images shown are projections through three-dimensional data stacks encompassing whole nuclei, generated with a maximum-intensity algorithm with the softWoRx (Applied Precision) software. For Fig 1D  and S3 Fig, the images of meT7 and mnT12 bivalents shown are snapshots of an Imaris threedimensional rendering of individual diakinesis bivalents with maximum intensity rendering for HTP-3 or HTP-1/2 with SYP-1. The images of wild type unfused autosome bivalents in Fig  1D are projections through three-dimensional data stacks encompassing the whole bivalent, generated with a maximum intensity algorithm with the softWoRx (Applied Precision) software. For

Ring stretching assay
Ring stretching on monopolar spindles in Fig 3 was performed as in [19]. Worms were grown on RNAi plates with a 1:1 mixture of emb-30 and klp-18 RNAi feeding clones from the Ahringer library to induce an arrest in metaphase on a monopolar spindle. RC components tend to stretch away from bivalents in extended metaphase arrest; spindles in extended arrest were determined by eye based on significant ring stretching of the three normal bivalents. The number of stretching rings on either meT7 or a control normal bivalent was counted using Imaris.

Image quantification
Ring structure quantification. To assess ring structure, RCs on meT7 or mnT12 bivalents were rotated in Imaris; prometaphase and metaphase oocytes with bipolar spindles were analyzed. Rings were scored by eye and considered mispatterned if they had more than one plane, consisted of more than one individual unit, or had more than one major loop. For Fig 2A and  S5A Fig, rings were considered "slightly mispatterned" if they consisted of two distinct units on a single plane or appeared as two connected loops on a single plane; given the resolution of our images, it was not possible to discern in most cases if this represents one ring with an aberrant shape, or two distinct rings in the same plane. Rings with more than two units, multiple loops at different angles or planes on the bivalent, or rings with many fragments were considered "severely mispatterned." For Fig 2E, quantification of ring structure was performed in MII, and the meT7 fusion chromosome was identified by its larger size. Because our analysis indicates that MI chromosome segregation is sometimes aberrant (Figs 5 and 6), it is possible that some of the analyzed MII meT7 chromosomes had not been fully segregated in MI; this could potentially account for some of the MII ring structure defects observed.
Linescans. For Fig 2, linescans were performed in ImageJ. Fluorescence intensity linescans of normal or fused chromosomes were performed along the pole-pole axis, as determined by tubulin intensity, at 40 x 30 pixels (L x W) (n = 25). Only clearly bipolar spindles were used for analysis, and all images had the same exposure conditions. Both chromosome length and fluorescence intensity were normalized to a maximum of 1, and the average (solid line) and standard error of the mean (SEM, shaded), of both DNA and SUMO were plotted using the ggplot2 package in Jupyter Notebook.
Metaphase alignment quantification. For Fig 4A, meT7 worms were arrested in metaphase using emb-30 RNAi. The metaphase plate was determined by rotating the image in 3D using Imaris until a single plane could be determined by eye based on SUMO-stained RCs on chromosomes I, II and V. The average total width of SUMO intensity of chromosomes I, II and V was measured using Imaris per spindle at the determined metaphase plate. The meT7 fusion chromosome was considered "unaligned" if greater than 50% of its SUMO intensity fell outside of 2 standard deviations of the average range of SUMO on chromosomes I, II and V from the determined metaphase plate.
Chromosome distance measurements. For Fig 4B, chromosome-to-pole distances were measured using Imaris. The center of the monopole was determined by using the "Surfaces" tool to determine the volume of the ASPM-1 region and to assign the center of this volume. Then, the distance between this point and the center of each chromosome was measured. Per spindle, the average of the distances to chromosomes I, II and V was subtracted from each individual distance to chromosomes I, II, or V (green points on the diagram/graph in Fig 4C) or to the fused chromosome (blue points).
Bivalent roundness quantification. For S7C Fig, bivalent roundness was determined by dividing the length of the shortest axis of an individual bivalent by the length of the longest axis. Sum projections of 3-slice fixed IF images were created in ImageJ, and the DNA channel was thresholded to define the bivalents. An oval shape was fit to these thresholded images, and the "circularity" of each bivalent was calculated using the "Analyze Particles" tool. The data were plotted in Jupyter Notebook.
Chromatin bridging quantification. For Figs 5, 6 and 7, and S5 Fig, DNA was denoted as bridged if in anaphase the width of the bridge was less than the width at either end of the segregating chromosomes. Otherwise, a single meT7 or mnT12 chromosome mass in anaphase was marked as showing delayed segregation, and two distinct segregating meT7 or mnT12 masses with no connecting DNA were considered wild type.
Meiosis II quatification. For Figs 5 and 7 and S5 Fig, ploidy was determined by eye based on chromosome counts on MII spindles (i.e. oocytes were considered aneuploid if the number of Hoechst-stained bodies was not equal to 4). A polar body tether was defined by a clear continual strand of DNA connecting a polar body to a chromatid pair within a forming or formed MII spindle. Polar bodies were determined to be DNA masses in close proximity to, yet remaining outside of, forming MII spindles.

Scoring embryonic viability
Viability counts (percent hatching) were determined by singling out 5-15 L4s onto individual plates and growing them at the temperatures indicated in the figures/tables until broods were produced. Mothers were transferred to new plates each day and allowed to produce broods until they no longer laid fertilized embryos. After mothers were moved, embryos and larvae were counted, and then returned to 25˚C and allowed to develop for 18-24 hours (wild type) or 40-48 hours (meT7) before counting unhatched embryos. meT7 embryos were given more time to develop, as the strain develops more slowly than wild type.

Live imaging
Live imaging of oocyte meiotic chromosome segregation in wild type and meT7 fluorescent fusion lines was accomplished by cutting open adult worms with a single row or less of embryos in 4μl of egg buffer (118mM NaCl, 48mM KCl, 2mM CaCl 2 , 2mM MgCl 2 , and 0.025 mM HEPES, filter sterilized before HEPES addition) on a coverslip and gently mounting onto a 2% agarose pad on a microscope slide. Worms were synchronized by hypochlorite hatch-off, and grown at 20˚C; worms were then upshifted to 25˚C overnight (16-20 hours) before imaging. Oocytes were imaged using a spinning disk confocal unit, CSU-W with Borealis (Andor), and dual iXon Ultra 897 (Andor) cameras mounted on an inverted Leica DMi8 microscope, with a 100x HCX PL APO 1.4NA oil objective lens (Leica). The imaging system was controlled via Metamorph (Molecular Devices) software. Oocytes were imaged every 5 seconds with 1μm Z-spacing (16μm total Z-stack) and the 488nm and 561nm channels were imaged simultaneously. After recording, movies were maximum projected, cropped, and color channels were adjusted independently for brightness and contrast in ImageJ (National Institutes of Health).

Statistical analysis
All reported P-values and statistical tests are specified in the text and figure legends. A P-value less than 0.05 was considered statistically significant.  Fig. HTP-1/2 and SYP-1 patterning on diakinesis chromosomes. Representative immunofluorescence images of fixed diakinesis chromosomes stained with HTP-1/2 (magenta) and SYP-1 (yellow) from an unfused autosome, the meT7 fusion chromosome, and the mnT12 fusion chromosome. The unfused autosome is from an mnT12 nucleus and displays normal patterning of HTP-1/2 on the long arm and SYP-1 on the short arm. In contrast, both fusion chromosomes display defects in establishing the long arm and short arm patterning of HTP-1/ 2 and SYP-1. Dashed lines indicate traced SYP-1 (yellow) and HTP-1/2 (magenta) along the chromosome axes. Scale bars = 1μm. (TIF) S4 Fig. Models for chromosome organization in meT7 bivalents. Cartoon images show potential models for how short arm interfaces are distributed on meT7 prometaphase bivalents, and how they may relate to different ring complex (RC) patterns from fixed oocyte zooms in Fig 2A. One whole ring (second row) may result from a single short arm interface, recruiting RC components in a manner similar to a wild-type-size bivalent (top row) on prometaphase bivalents. meT7 bivalents with a slight mispatterning RC (third row) may have two short arm interfaces, one at each end of the paired homologs, leading to a structure with ring complex components that are distinct, yet visible on the same plane. For bivalents with more complex RC mispatterning (bottom row), it is likely that the short arm interface is significantly impaired, potentially due to incorrect crossover number, and the targeted RC is not restricted to a single plane or a coherent, ordered structure in prometaphase. (TIF) S5 Fig. mnT12 oocytes show similar phenotypes to meT7 oocytes. (A) Immunofluorescence and quantification of AIR-2 localization in fixed mnT12 oocytes. Single-bivalent zooms of mnT12, indicated by arrows, show that we see similar categories of AIR-2 localization as meT7: whole ring (top row), slight mispatterning (middle row), or severe mispatterning. 0% (0/40 at 15˚C and 0/35 at 25˚C) of normal bivalents showed mispatterned AIR-2 localization in mnT12 oocytes. At 15˚C, mnT12 bivalents were slightly mispatterned in 8/36 oocytes and severely mispatterned in 1/36 oocytes, while at 25˚C mnT12 bivalents were slightly mispatterned in 9/27 oocytes and severely mispatterned in 3/27 oocytes. (B) Chromatin bridges, indicated by arrows, are present in fixed mnT12 anaphase oocytes. In N2 mid-to-late anaphase oocytes, 0/75 spindles contained anaphase bridging. However, 3/18 mnT12 spindles showed chromatin bridges at 15˚C, and 3/11 mnT12 anaphase spindles showed chromatin bridges at 25˚C. (C) Fixed mnT12 oocytes show persisting consequences of anaphase bridging in Meiosis II. No chromatin-tethered polar bodies or anueploid Meiosis II spindles were observed in N2 Meiosis II oocytes. At 15˚C, 3/25 Meiosis II mnT12 oocytes contained DNA tethered to the polar body, and 1/25 Meiosis II mnT12 oocytes was aneuploid. At 25˚C, 6/27 Meiosis II mnT12 oocytes had tethered polar bodies, and 3/27 of Meiosis II mnT12 oocytes were aneuploid. All scale bars = 2.5μm. (TIF) Fig 2). Immunofluorescence showing that BIR-1 (green) and AIR-2 (red) colocalize on all ring structure types on meT7 in fixed oocytes. Scale bars = 2.5μm. (TIF) S7 Fig. Kinetochore proteins localize to misshapen meT7 bivalents. (A) Immunofluorescence of BUB-1 in prometaphase meT7 oocytes. BUB-1 targets to the entirety of meiotic bivalents in meT7, shown in a full projection (top row), and specifically cups holocentric normal bivalents (bottom row, left) and meT7 fused bivalents (bottom row, right), as shown in single slices. Scale bars = 2.5μm. (B) Immunofluorescence of SEP-1 in fixed prometaphase meT7 oocytes. SEP-1 targets to the entirety of meiotic bivalents in meT7, shown in a full projection (top row), and specifically cups holocentric normal bivalents (bottom row, left) and meT7 fused bivalents (bottom row, right), as shown in single slices. Scale bars = 2.5μm. (C) meT7 bivalents are shaped differently than normal-sized bivalents. Bivalent roundness, or the ratio of the length of the shortest axis on the bivalent to the length of the longest axis, was calculated using ImageJ. This ratio tended to be higher (P = 0.0027, two-tailed Mann-Whitney U test) for meT7 bivalents (N = 51) than normal-sized bivalents (N = 64), suggesting meT7 bivalents have a less apparent bilobed architecture, and instead have one closer to that of a circle. (TIF) S1 Movie. Example movie showing meT7 segregation occasionally (5 of 14 movies) results in the first polar body being tethered to the Meiosis II spindle by a chromatin bridge (related to Fig 6 and S1 Table). Microtubules (GFP::TBB-2) shown in magenta and chromosomes (mCherry::HIS-11), shown in green, during both meiotic divisions in live cells. Playback framerate is 25 frames per second.  Table). Microtubules (GFP::TBB-2) shown in magenta and chromosomes (mCherry::HIS-11) shown in green during both meiotic divisions in live cells. Playback framerate is 25 frames per second. (AVI) S1