TOPII and chromosome movement help remove interlocks between entangled chromosomes during meiosis

During meiosis, unrelated chromosomes frequently become interlocked, and these structures must be removed for complete synapsis and normal chromosome segregation. Martinez-Garcia et al. show that the active removal of interlocks requires topoisomerase II and chromosome movement.


Introduction
In sexual reproduction, gamete formation is dependent on meiosis, a specialized cell division pathway involving a single round of DNA replication followed by two rounds of division. At the onset of meiotic prophase I, pairs of newly replicated sister chromatids of each homologue become organized into linear looped arrays conjoined at the loop bases by a protein axis. Homologue pairs then align and undergo synapsis through the formation of the synaptonemal complex (SC), a tripartite protein structure comprising the homologous chromosome axes brought into close apposition by polymerization of transverse filaments. This chromosome remodeling is closely coordinated with homologous recombination, which leads to the formation of genetic crossovers (COs) in late prophase I. At the end of prophase I, the SC is disassembled, allowing the CO sites to be visible cytologically as chiasmata as the homologue pairs (bivalents) condense before the first meiotic division.
During installation of the SC in the zygotene stage of prophase I, unrelated chromosomes can become interlocked. Interlocks arise when a bivalent or chromosome becomes entrapped between a different pair of homologues undergoing synapsis at each end ( Fig. 1 A). It is essential that interlocks are removed in order to complete pairing and avoid possible restrictions to chromosome segregation. Interlocks were first observed in the flat-worm Dendrocoelum lacteum (Gelei, 1921). Although interlocks appear rare in species with small chromosomes (Holm et al., 1981;Storlazzi et al., 2010), ultrastructural analysis of meiocytes at zygotene and pachytene in species with larger chromosomes such as some insects, mammals, and plants have shown that they occur at high frequency and can involve multiple chromosomes (von Wettstein et al., 1984;Zickler and Kleckner, 1999;Wang et al., 2009). Nevertheless, their presence in meiocytes at pachytene is rare. For instance, in the silkworm Bombyx mori, ∼2-3% of cells at pachytene were found to have interlocks, whereas at zygotene, the figure was 87% (von Wettstein et al., 1984). Interlocks have been classified into two categories: the first, in which a bivalent is entrapped in a loop between two stretches of SC of another bivalent, and the second, where an unsynapsed chromosome is trapped (Gelei, 1921). Complex interlocks involving more than two bivalents also occur (Sax and Anderson, 1934;Wang et al., 2009). If a chromosome or bivalent interlocked between two COs remains unresolved, the link can persist until metaphase I without affecting segregation (Heslop-Harrison and Bennett, 1985), but complex structures can impede correct orientation of centromeres (Yacobi et al., 1982). Bivalents interlocked at metaphase I appear as chains of rings or as a rod bivalent in between a bivalent with two chiasmata.
Suggestions have been made to account for the dramatic reduction in the frequency of interlocks observed at pachytene and metaphase I compared with zygotene. Although it is possible that this is simply due to the failure of zygotene cells with interlocks to progress to pachytene or that progression is delayed, leading to an overestimation of the relative abundance of these cells, substantial cytological evidence supports the active removal of interlocks (Zickler and Kleckner, 1999). Ultrastructural studies in human spermatocytes, lily, and B. mori have revealed breaks in the chromosome axes and SC at zygotene that are clearly associated with entanglements. By pachytene, these breaks are no longer observed. Hence, it is proposed that formation of a transient chromosomal break allows resolution of the interlock by passage of the trapped chromosome or bivalent, after which the break is rejoined. It is suggested that this might be accomplished by a type II topoisomerase (TOP II), although this would require first remodeling the meiotic axis to create an environment in which multiple DNA strand-passage events by TOP II would be possible Zickler and Kleckner, 1999). Consistent with this possibility, immunolocalization studies reveal that TOP II is associated with the chromosome axes during meiotic prophase I in mammals and in budding yeast (Klein et al., 1992;Liang et al., 2015;Guturi et al., 2016). Use of TOP II inhibitors in mouse male and female meiocytes lead to defects in chromosome condensation and segregation at the first meiotic division (Kallio and Lahdetie, 1996;Russell et al., 2004;Gómez et al., 2014). Different topII conditional mutants in budding yeast and mammals undergo checkpoint arrest in mitosis and meiosis (Holm et al., 1985;Rose and Holm, 1993;Mundbjerg et al., 2015). However, a direct link between TOP II and interlock removal has thus far not been demonstrated.
Another proposal to explain interlock removal invokes chromosome movement toward the end of the other bivalent combined with de-and repolymerization of the SC and telomere detachment from the nuclear envelope (NE) in order to liberate the trapped chromosome . In support, these authors reported the observation of lily meiocytes at early zygotene, where 14 of the 48 telomeres were not associated with the NE. A variation of this model suggests that instead of detaching homologous telomeres from the NE, simply unpairing them would create a gap through which the trapped chromosome could escape (Kleckner and Weiner, 1993). Evidence for this has been found in maize (Golubovskaya et al., 2002), in sun1sun2 mutants in Arabidopsis thaliana (Varas et al., 2015), and in disruption of the nuclear lamina in Caenorhabditis elegans (Link et al., 2018).
In this study, we describe the identification of a viable hypomorphic topII mutant of the plant A. thaliana coupled with meiosis-specific knockdown of TOP II activity using RNAi. Analysis of these lines using a combination of epifluorescence microscopy and SIM has enabled us to confirm a key role for the protein in interlock resolution. Moreover, further analysis using a mutant defective in a nucleoporin, NUP136, which compromises chromosome movement, reveals that although TOP II localization is required for the removal of a proportion of interlocks, chromosome movement is also necessary. Thus, our study indicates that at least two mechanisms operate to remove interlocks during zygotene stage in order to correctly form the SC between homologous chromosomes. In the nucleus, numerous entanglements can impede SC polymerization. When two synapsis initiation sites are formed in between homologous chromosomes with another bivalent or chromosome in between, an interlock arises, and synapsis is delayed in that area. dsDNA, double-stranded DNA. (B) AtTOP II gene structure and localization of topII mutations. Boxes in dark blue indicate exons; lines joining the boxes indicate introns; boxes in light blue indicate 5′ and 3′ UTRs. Arrows represent primers used for RT-PCR. (C) Expression of TOP II assessed by RT-PCR of buds of topII-1 and topII-RNAi and their respective WT backgrounds. Exons 19-20 amplify the region before and after the T-DNA insertion in topII-1. Exons 6-10 correspond with the region complementary for the RNAi construct. Exon 20 amplifies the coding region immediately after the T-DNA insertion. GAP DH and Actin2 are constitutive controls. (D) TOP II localization by Western blotting in floral tissue from WT (Ws) and topII-1. Section from ∼150-200 kD. Coomassie staining is shown as a loading control. (E) Representative silique length comparison of Ws, topII-1, and topII-RNAi. (F) Representative pictures of 33-d WT (Ws), topII-1, and topII-RNAi plants. (G) Immunolocalization of TOP II (mouse) and ASY1 in prophase I of WT (Ws), topII-1, and topII-RNAi meiocytes using widefield fluorescence microscopy. Bars, 5 µm.

Results and discussion
From a set of five topII transfer DNA (T-DNA) insertion lines, only one, FLAG_476H07 (herein after topII-1), present in the Wassilewskija (Ws) ecotype, could be recovered as a homozygous mutant ( Fig. 1 B). Previous attempts to identify homozygous mutants from four other lines were also unsuccessful (Makarevitch and Somers, 2005). The T-DNA in topII-1 is inserted at position 5,927 bp within the last exon, generating an early stop codon that leads to a deletion of the last 82 amino acids of the protein. RT-PCR expression analysis confirmed that the amino terminal region is expressed at near-WT levels in topII-1, whereas expression spanning the T-DNA is not detectable (Fig. 1 C). However, a low level of transcription (2%) of the coding sequence downstream of the T-DNA is detectable in buds ( Fig. 1 C; Fig. S1 A shows seedling RT-PCR). RT-PCR analysis using T-DNA left-border primer LB4 and TOP II exon 20 primer indicates that this arises from transcription from within the T-DNA, a phenomenon that is not unusual (Wang, 2008). Western blot analysis of WT floral protein extracts with anti-TOP II antibody detected a product around the size (164 kD) predicted for A. thaliana TOP II (Fig. 1 D;Xie and Lam, 1994). This was absent in topII-1 extracts, where a smeary band of lower molecular weight was detected instead. This suggests that TOP II is truncated and that the mutant protein may be slightly less stable. Studies have suggested that the TOP II C-terminal region is poorly conserved and not essential for the catalytic activity of the protein, although it is implicated in its regulation (Shaiu et al., 1999;Dickey and Osheroff, 2005;Kawano et al., 2016). This probably explains why it was possible to identify viable topII-1 homozygotes. Vegetative development of topII-1 plants appeared delayed ( Fig. 1 F) but was phenotypically normal by maturity except for a 22% reduction in seeds per silique compared with WT (38.5 ± 0.7 and 49.5 ± 0.9; n = 25 per genotype; P = 0.001; Fig. 1 E). Given the apparent lethality of most mutant topII alleles, we sought to specifically deplete expression of the gene in meiocytes using RNAi. A TOP II RNAi cassette based on exons 6-10 was placed under the control of the meiosis-specific DMC1 promoter in ppF408, which has previously been used for meiosis-specific gene-knockdown research (Fig. 1 B;Higgins et al., 2005). The selected line, topII-RNAi, showed a 92% reduction in fertility relative to WT Col-0 (4 ± 0.47; n = 30; P < 0.001) but had sufficient residual fertility to enable it to be maintained ( Fig. 1 E). RT-PCR analysis revealed apparently normal expression of TOP II in buds ( Fig. 1 C), but this was anticipated as meiocytes constitute <1% of anther tissue (Chen et al., 2010). Immunostaining using anti-TOP II showed general staining in the chromatin and some punctate signals in WT and topII-1 but reduced intensity in topII-RNAi meiocytes (Col,433.34 ± 22.26,n = 8;291.89 ± 27.57, n = 14; P = 0.013; Fig. 1 G). The same pattern was confirmed in Brassica oleracea meiocytes (Fig. S1 B) and using an independent antibody (Agrisera) in WT and topII-1 (Fig. S1 C).
topII-1 and topII-RNAi exhibit increased interlocks at metaphase I DAPI-stained chromosome spread preparations from male meiocytes of topII-1 were examined using epifluorescence microscopy. No obvious differences between topII-1 and WT (Ws-4) were observed at leptotene. However, we noted apparent inter-locks in 57% of topII-1 late zygotene/early pachytene cells (n = 14) compared with 18% (n = 17) in WT (Fig. 2, A, E, and e′). It should be noted that it was difficult to reliably distinguish interlocks from chromosome overlaps in a quarter of the cases. Thus, for cells in prophase I, we henceforth use the term entanglement rather than interlock, which we only apply at metaphase I. At diplotene/ diakinesis as the SC disassembled, the topII-1 bivalents were less condensed than WT and often entangled (Fig. 2, B and F). At metaphase I, pairs of interlocked bivalents were present in 20% (n = 79) of topII-1 meiocytes but were absent in WT (Fig. 2, C and G; n = 94). A few chromatin bridges were observed in anaphase I nuclei (Fig. 2 H; 1.4 ± 0.3 bridges per cell; n = 14). As they involved homologous chromosomes (Fig. S1 D), they are unlikely to arise from an interlock related problem, but they could be a contributory factor in the reduced fertility of topII-1.
Examination of topII-RNAi meiocytes revealed that at early prophase I, most cells appeared similar to topII-1 (Fig. 2 I), but in some cases, there was evidence of DNA damage as highly condensed nuclei were observed in slides with zygotene/pachytene stages (12%; n = 126), reminiscent of cells undergoing programmed cell death (Fig. S1 H; Kurusu and Kuchitsu, 2017). At diakinesis, 81% of cells had most bivalents aggregated and a more pronounced condensation defect (Fig. 2 J). At metaphase I, interlocks were observed in 88% of the nuclei (n = 61; Fig. 2 K). Most cells (54%) had more than two interlocked bivalents (Fig. S1 I), and in nine cases, all five bivalents were involved. Anaphase I and II cells showed high levels of DNA fragmentation, most likely leading to the reduced fertility of topII-RNAi plants (Fig. 2 L).
TOP II is associated with the chromosome axis and accumulates in entangled regions Immunofluorescence microscopy of chromosome spread preparations of WT meiocytes at zygotene using antibodies that recognize TOP II and chromosome axis protein ASY1 revealed that TOP II mostly localizes as numerous diffuse foci associated with the axis (Fig. 1 G). We observed that in some regions, the foci seemed to accumulate as patches. A TOP II signal was present in topII-1, indicating that the mutant protein could associate with chromatin.
Further analysis with SIM confirmed the presence of abundant chromatin-associated TOP II foci in WT at zygotene (mean 108.2 ± 10.62; n = 5; Fig. 3 A). We also noted that TOP II twisted around the axis in a braid-like manner in a fashion similar to that reported for TOP II distribution in Muntjac mitotic cells (Fig. 3 B; Liang et al., 2015). SIM also suggested that TOP II was enriched in certain regions of the nucleus. Moreover, it appeared (n = 23; five cells) that these regions coincided with sites of chromosomal entanglements (Fig. 3, A, C, and D; and Video 1). In some cases, we observed breaks in the chromosome axis in the vicinity of the TOP II-enriched region, reminiscent of those reported in the earlier EM research (Fig. 3 C and Video 2;Holm and Rasmussen, 1980). Although we have not previously observed such breaks, we cannot exclude the possibility that they arose during the spreading procedure. In topII-1 (Fig. S2, A and B), enrichment of TOP II with entangled regions was less pronounced than in WT (36 entanglements, cells = 4, two of which had a TOP II discrete focus on top, compared with 33 entanglements in WT; Fig. S2 C). Also, the foci were smaller than those in WT (topII-1, 25.60 ± 1.79 nm, n = 67, vs. 68.90 ± 4.00 nm, n = 541; P < 0.001) and less abundant (16.8 ± 6.7 foci per cell; n = 4), and the patches of foci observed in WT were absent (Fig. S2). This could suggest that in WT, TOP II is actively recruited to the entangled chromosomes but that the mutation in topII-1 compromises or reduces this response; however, confirmation requires further study.
A role for chromosome movement in interlock resolution? Inspection of the interlocked chromosomes in topII-1 revealed that in all 16 cases, they exhibited a "closed" configuration at metaphase I; that is, they involved two or more ring bivalents (those with a chiasmus on either side of the centromere) held together like links in a chain (Fig. 3 E). This configuration likely originated in zygotene, when one of the chromosomes of a bivalent was trapped in the synaptic bubble of another bivalent, and both had CO events at each end (Fig. 4 E). In topII-RNAi, 85% (n = 59) of the interlocks had a "closed" configuration. As rod bivalents are frequent in A. thaliana, interlocks with an "open" configuration would also be expected to arise. In this case, a rod bivalent is entrapped between two flanking chiasmata in another bivalent (Figs. 3 E and 4 D). During zygotene/pachytene, this structure would be maintained by the SC. However, at diplotene/ diakinesis, as the SC breaks down and the chromosomes begin to condense, prophase I chromosome movement can potentially release the entrapped rod without a requirement for chromosome breakage, unlike "closed" interlocks. We therefore investigated whether the absence of open interlocks in topII-1 and topII-RNAi might reflect their removal via a separate route dependent on chromosome movement. We examined nup136-2, a mutant defective in NUP136 (Tamura et al., 2010), which is required for normal chromosome movement (Varas García, 2014). Although prophase I appeared normal in chromosome spread preparations of nup136-2 meiocytes (Fig. S3, A-D), examination at metaphase I revealed interlocks in 24.5% (n = 49) of the cells. Also, the mutant had a reduced chiasma frequency of 6.0 ± 0.2 (n = 40) compared with 8.7 ± 0.2 (n = 40) in WT Col-0 (Varas García, 2014), which could presumably reduce the number of interlocks observed at metaphase I. In contrast with topII-1 and topII-RNAi, almost all interlocks involved an open configuration, with only two having a closed configuration (Varas García, 2014). Together with the analysis of topII-1 and topII-RNAi, these observations support a model whereby interlock removal involves both TOP II activity and chromosome movement.

topII-1 and nup136-2 have an additive impact on interlock removal
To gain further support for two mechanisms for interlock removal, we attempted to construct a topII-1/nup136-2 double mutant. This proved difficult due to the proximity of the loci on chromosome 3 and the impact on viability whereby nearly all progeny died (n = 283). Eventually, one surviving double mutant (topII-1 hom nup136-2 hom ) was identified (Fig. S3, E-H). Although the plant was dwarfed, a few buds were recovered, which allowed the identification of 13 metaphase I cells in chromosome spreads ( Fig. S3 C). Examination revealed a chiasma frequency of 6.8 ± 0.4, with an overall mean of 1.5 ± 0.4 interlocks per cell (Table 1). This limited sample suggested that the mutations were additive.
FISH using 45S and 5S ribosomal DNA (rDNA) probes enabled the determination of interlock involvement of each chromosome (Fig. 4, C and D). Unsurprisingly, the two longest chromosomes, 1 and 3, were most frequently involved (in 24% of cases; n = 50). However, interlocks between the two short nuclear organizer region-associated chromosomes 2 and 4 were also frequent (14%), probably related to the proximity of these chromosomes in the premeiotic nuclei (Sanchez Moran et al., 2001;Fransz et al., 2002).
We next analyzed the contribution of each type of interlock observed in metaphase I cells of topII-1 hom nup136-2 het . Unfortunately, interlocks in cells in which most of the bivalents remained  entangled proved difficult to reliably classify. Hence, it is possible that as in other mutants with high-interlock frequency factors (Martinez et al., 2001), such as condensation problems, ectopic synapsis, or recombination, may also contribute to formation of these structures. We found one such example in which bivalents 2 and 4 were connected in metaphase I by their 45S region (Fig. 4 D,  arrow), probably reflecting ectopic recombination. However, we did not observe terminal connections between chromosomes, indicating nonhomologous end joining activity was not increased. Moreover, mitotic anaphase bridges were not significantly different in topII-1 (14%; n = 194) and topII-1 hom nup136-2 het (21%; n = 207; P = 0.22), suggesting no additional DNA repair defect in this line. As it is likely that closed interlocks are more prone to result in highly entangled chromosomes, their proportion may have been underestimated in the analysis. Nevertheless, of the 50 interlocks that could be classified, 27 had a closed configuration, and 23 were open (Fig. 3 E). Thus, both classes of interlocks remain unresolved when TOP II activity and chromosome movement are both compromised (Fig. 4 E).
In summary, we have provided evidence that chromosome entanglements leading to interlocks during meiosis, first described almost a century ago (Gelei, 1921), are actively removed during meiotic prophase I by at least two distinct pathways working in unison: chromosome movement and a TOP IIregulated pathway. Recent work relates how attachment of chromosomes to the NE avoids entanglements in interphase and meiosis, in line with our findings (Kinney et al., 2018;Link et al., 2018). Questions still remain, particularly in relation to how TOP II activity is regulated to coordinate the different steps in the putative breakage and religation of the chromatids. Moreover, it remains possible that other factors may contribute to interlock removal. For example, directionality toward solving the interlock could be imposed by cohesins (Sen et al., 2016) and distinction between homologous chromosomes by mismatch repair proteins like MLH1. Indeed, a potential role for MLH1 in interlock resolution has previously been proposed based on research in the filamentous fungus Sordaria macrospora (Storlazzi et al., 2010).
Our working model proposes that interlock resolution is governed by whether it is a bivalent or a single chromosome that becomes trapped and the relative position of designated COs. For instance, during zygotene/early pachytene, when the telomeres are attached to the nuclear envelop, if a single chromosome of a bivalent with designated COs at each distal end is entrapped between a synapsing bivalent with similarly placed designated COs, then release will require coordinated chromosome and axis breakage followed by break resealing. Failure to do so would result in the presence of closed interlocks at diakinesis/metaphase I. In the case where COs are not designated in all four distal regions, the entrapped chromosome may be released by a combination of chromosome movement and SC depolymerization. In this instance, failure to release the entrapped chromosome would lead to the presence of an open interlock structure. Finally, where a synapsed bivalent is encircled by another synapsing pair of chromosomes, this may be resolved by chromosome movement combined with release (or separation) of the telomeres from the NE (Fig. 4 E).

Cytogenetic techniques
A. thaliana inflorescences were conserved in 3:1 fixative and stored at 4°C for chromosome preparations by spreading technique (Fransz et al., 1998; modified as in Martinez-Garcia and Pradillo, 2017). After cell wall digestion, a single bud was fixed in 60% acetic acid at 45°C and stirred for 1 min. Refixation at 3:1 was followed by drying at RT. Slides were stained with DAPI and mounted in Vectashield. The same procedure was used for FISH, followed by DNA denaturation and hybridization protocol (Sanchez Moran et al., 2001) with the probes 45S rDNA (from plasmid pTa71; Gerlach and Bedbrook, 1979) and 5S rDNA (from plasmid pCT4.2).

Image acquisition and processing
For image acquisition, a fluorescent microscope (Nikon i90) equipped with a Nikon DS-Qi1Mc digital camera at RT and NIS Elements software (Nikon) were used. Flat cells on spreading slides were acquired in a Plan Apochromat VC 100× 1.40 NA oil ∞/0.17 differential interference contrast N2 objective by Multichannel acquisition, whereas 3D cells were captured with the Z series acquisition tool using 0.1-µm steps per Z stack. Overexposure was avoided, and exposure time was maintained constant between slides of the same batch for intensity and detailed comparisons. Images were not deconvoluted in any way when using the epifluorescent Nikon microscope except for γH2AX images (MexicanHat). Brightness and contrast were adjusted slightly to remove background but respecting the different intensities in the cell. Adjustments were applied uniformly in mutant and control images. 3D images were processed by maximum-intensity projection for figure presentation.
To analyze the ultrastructure of immunosignals and chromatin beyond the classical Abbe/Raleigh limit at a lateral resolution of ∼120 nm (superresolution; achieved with a 488-nm laser), spatial SIM (3D-SIM) was applied using a 63× 1.4 NA oil Plan Apochromat objective of an Elyra PS.1 microscope system at 30°C and using ZEN Black software (ZEI SS). Images were captured separately for each fluorochrome using the 561-, 488-, and 405-nm laser lines for excitation and appropriate emission filters (Weisshart et al., 2016). Maximum-intensity projections of whole meiocytes were calculated with ZEN software. Enlarged sections were presented as single slices to indicate the subnuclear chromatin and protein structures at the superresolution level. 3D rendering based on SIM image stacks was done using the Imaris 8.0 (Bitplane) software.
Pixel intensity (in AU) was measured in total nucleus area in all Z slices per cell with ImageJ. Average intensity per cell was used to compare TOP II signal in WT and topII-RNAi samples. The number of TOP II foci per nucleus and their area were measured by ImageJ processing: threshold was adjusted to obtain a binary image without background, watershed was applied in order to individualize the foci, and the Analyze Particles tool gave the results table and the final image with the particles.

Statistical analysis
Average figures were followed by their standard errors as a deviation measurement. Statistical analysis to compare means was Mann-Whitney U Test performed using SPSS.