Crossover recombination and synapsis are linked by adjacent regions within the N terminus of the Zip1 synaptonemal complex protein

Accurate chromosome segregation during meiosis relies on the prior establishment of at least one crossover recombination event between homologous chromosomes. Most meiotic recombination intermediates that give rise to interhomolog crossovers are embedded within a hallmark chromosomal structure called the synaptonemal complex (SC), but the mechanisms that coordinate the processes of SC assembly (synapsis) and crossover recombination remain poorly understood. Among known structural components of the budding yeast SC, the Zip1 protein is unique for its independent role in promoting crossover recombination; Zip1 is specifically required for the large subset of crossovers that also rely on the meiosis-specific MutSγ complex. Here we report that adjacent regions within Zip1’s N terminus encompass its crossover and synapsis functions. We previously showed that deletion of Zip1 residues 21–163 abolishes tripartite SC assembly and prevents robust SUMOylation of the SC central element component, Ecm11, but allows excess MutSγ crossover recombination. We find the reciprocal phenotype when Zip1 residues 2–9 or 10–14 are deleted; in these mutants SC assembles and Ecm11 is hyperSUMOylated, but MutSγ crossovers are strongly diminished. Interestingly, Zip1 residues 2–9 or 2–14 are required for the normal localization of Zip3, a putative E3 SUMO ligase and pro-MutSγ crossover factor, to Zip1 polycomplex structures and to recombination initiation sites. By contrast, deletion of Zip1 residues 15–20 does not detectably prevent Zip3’s localization at Zip1 polycomplex and supports some MutSγ crossing over but prevents normal SC assembly and Ecm11 SUMOylation. Our results highlight distinct N terminal regions that are differentially critical for Zip1’s roles in crossing over and SC assembly; we speculate that the adjacency of these regions enables Zip1 to serve as a liaison, facilitating crosstalk between the two processes by bringing crossover recombination and synapsis factors within close proximity of one another.


Introduction
A unique feature of the meiotic cell cycle is how chromosomes are parsed at the first division: Homologous chromosomes (homologs) orient and precisely segregate from one another on the meiosis I spindle due to the prior establishment of recombination-based associations between homologs. Interhomolog crossover recombination creates a reciprocal splice between non-sister DNA molecules; in conjunction with sister cohesion, this DNA exchange provides a physical association between replicated homologs that is stable but nevertheless can be released to allow disjunction after the bivalent has acquired a proper orientation on the spindle [1].
Interhomolog crossovers form during meiotic prophase through the homologous recombination-based repair of a large number of programmed DSBs catalyzed by the meiosisspecific, topoisomerase-like protein, Spo11 [2]. For many organisms, the repair pathway that allows a subset of Spo11-mediated DSBs to become interhomolog crossovers involves the formation and processing of Holliday junction intermediates by meiosis-specific proteins, including the MutSg and MutLg heterodimeric complexes which have homology to the bacterial MutS and MutL protein families, respectively [3][4][5][6][7][8][9][10][11][12][13][14]. DSB repair processes in meiotic cells also rely on meiosis-specific proteins and pathways to ensure specific desired outcomes of meiosis: for example that crossovers preferentially involve non-sister chromatids of homologous chromosomes (as opposed to involving the sister chromatids that comprise a single chromosome), and that every chromosome pair, no matter how small, receives at least one crossover [15].
Many of the meiosis-specific factors that function in MutSg crossover recombination also serve in the assembly of a widely-conserved, prominent feature of meiotic prophase chromosomes: the synaptonemal complex (SC) [16,17]. The SC is a proteinaceous macromolecular structure comprised largely of proteins with extensive coiled-coil (called transverse filaments) that form parallel higher order units that accumulate in an ordered fashion to create "rungs" connecting aligned chromosome axes along their entire lengths (chromosome axis structures are referred to as lateral elements in the context of the mature SC structure). As demonstrated in multiple organisms by electron and super-resolution microscopy using epitopespecific antibodies, multiple SC transverse filament protein units span the conserved 100 nm width of the SC and orient with their opposing C termini toward lateral element structures [18,19]. In many organisms including budding yeast and mammals, a separate set of structural proteins, comprising the "central element" substructure, interface with N terminal regions of (head-to-head-oriented) transverse filament units at the midline of the SC.
While the processes of SC assembly (synapsis) and recombination are mechanistically independent and separable, nodes of crosstalk exist between the two processes. One widelyconserved example of such crosstalk is the reliance of meiotic crossover events on SC assembly proteins. Especially noteworthy is the fact that mutants missing building block components of the SC, particularly transverse filament proteins such as the budding yeast Zip1 protein, are typically deficient in MutSg crossover formation [16]. The reliance of crossover recombination on SC proteins is perhaps unsurprising given that meiotic recombination intermediate-associated complexes fated to become chiasmata (cytological manifestations of the crossover links between homologs) embed directly within the central region of the SC [20,21]. However we note that, at least in budding yeast, not all SC structural components play a role in crossing over and the mature SC structure itself is not a prerequisite for even MutSg crossover recombination: Budding yeast mutants deficient in the SC central element components Ecm11 or Gmc2, or expressing an ecm11 allele that prevents Ecm11 SUMOylation, fail to assemble tripartite SC but nevertheless exhibit excess MutSg crossovers [22]. This finding indicates not only that tripartite SC is dispensable for crossing over, but that SC is associated with an activity that antagonizes interhomolog crossover formation; at least one aspect of the observed anti-crossover activity of the budding yeast SC is likely to be a capacity to inhibit Spo11 DSB formation [23,24].
Genetic evidence from multiple systems also suggests that the recombination process directly influences SC assembly. In organisms including budding yeast and mammals, early steps in homologous recombination are a prerequisite for proper synapsis. In spo11 mutants, which fail to initiate meiotic recombination, SC assembly does not occur extensively in mammals, or at all in budding yeast [2,25]. In budding yeast, a subset of proteins that co-localize with MutSg on recombining meiotic chromosomes are required downstream of DSB formation for the formation of stable, crossover-designated recombination intermediates [8,14] and for robust SC assembly.
This group of proteins includes the so-called "Synapsis Initiation Complex" factors (Zip2, Zip3, Zip4 and Spo16 [19,[26][27][28][29][30][31]. In the absence of any one of these proteins, the MutSg heterodimer, or the SC transverse filament protein Zip1, recombination intermediates fail to form stable joint molecule (Holliday junction) structures. In the absence of Zip2, Zip4 or Spo16 (and, by default, Zip1), SC assembly is completely abolished. However, in the absence of MutSg or the putative E3 SUMO ligase, Zip3, SC assembly is not absent but diminished, presumably due to a failure or severe delay in synapsis initiation from non-centromeric (presumed recombination-associated) chromosomal sites [26,32]. SIC proteins have also been found to regulate the post-translational SUMOylation of SC central element component, Ecm11 [29]. Taken together, these data suggest that intermediate events in the budding yeast meiotic recombination process mediate the gradual, stepwise assembly of a recombination intermediate-associated complex that has the capacity to trigger SC elaboration. Interestingly, even in C. elegans where SCs assemble in the absence of recombination initiation, MutSg-associated crossover recombination intermediates locally influence the physical and dynamic properties of the C. elegans SC, through a Polo-like kinase (PLK-2) signaling mechanism [33]. The observed interdependencies between synapsis and meiotic recombination indicate that the two processes are not only spatially correlated but that the mechanisms involved in each process are functionally intertwined, however we currently lack a substantial molecular understanding of the how these hallmark meiotic processes intersect.
In budding yeast it is clear that the SC transverse filament protein, Zip1, serves an early role in promoting crossover recombination independent of (and prior to) its structural role in assembling the SC; in this case, a single protein evolved dual functions to promote these unlinked but coordinated meiotic prophase processes. The budding yeast Zip1 protein thus provides an opportunity to understand how SC transverse filaments can both promote and coordinate interhomolog recombination and SC assembly. Here we present a phenotypic analysis of mutants carrying a series of non-null zip1 alleles that encode small in-frame deletions within Zip1's (putatively unstructured) N terminus; together with our previously-published analysis of the zip1[D21-163] mutant, these zip1 alleles encompass distinct and nearly reciprocal phenotypes with respect to synapsis and crossover recombination. Our data identify critical N-terminal residues that correspond to Zip1's dual function in regulating crossing over and synapsis, and suggest that these residues may encompass adjacent interaction sites for the SIC protein and putative E3-SUMO ligase, Zip3, and the SUMOylated SC central element component, Ecm11.

MutSg crossovers rely on residues within Zip1's extreme N terminus
The primary amino acid sequence of Zip1 suggests that the region encompassing residues ~175-748 of the 875 residue protein has the capacity to assemble an extended coiled-coil structure, while the flanking N-and C-terminal regions are likely unstructured. Consistent with the rod-like nature that is predicted by an extended coiled-coil, two Zip1 units assembled in a head-to-head fashion span the ~100 nm width of the budding yeast SC central region [18,19]; Zip1's C termini orient toward aligned chromosome axes (lateral elements) while its N termini orient toward the central region substructure (comprised of Ecm11 and Gmc2 proteins) at the midline of the budding yeast SC. We previously reported that the non-null zip1[D21-163] mutant phenocopies SC central element-deficient ecm11 and gmc2 null mutants: In each of these mutants tripartite SC assembly fails but MutSg-mediated crossover recombination events occur in excess [22]. In order to identify residues within the Zip1 and Zip1[D21-163] proteins that are critical for Zip1's MutSg-mediated crossover activity, we created and analyzed additional nonnull zip1 alleles. We found that alleles encoding disruptions in Zip1's first twenty residues severely compromise Zip1's capacity to promote MutSg crossovers ( Figure 1A).

Removal of residues 2-9 or 10-14 results in a novel separation-of-function Zip1 protein that is crossover-deficient but synapsis-proficient
Residues 21-163 within Zip1's N terminal unstructured region are dispensable for MutSgmediated crossover recombination but essential for tripartite SC assembly [22]. We investigated whether residues 2-20 are critical for the formation of mature SC by asking whether coincident linear structures of the SC transverse filament protein (Zip1) and the SC central element protein Ecm11 assemble on surface-spread meiotic prophase nuclei from strains carrying wild-type or a mutant zip1 allele and missing the Ndt80 transcription factor. Ndt80 is required for progression beyond a mid-meiotic prophase stage when full-length SCs are normally assembled [36], thus the ndt80 null background allows us to maintain cells in sporulation medium for prolonged periods in order to assess the overall capacity of a strain to assemble SC. We initially examined meiotic nuclei at 24 hours after placement into sporulation medium, when ~85% of ZIP1 ndt80 strains in our BR genetic background exhibit nearly full synapsis [37]. We utilized a polyclonal antibody targeted against Zip1's C terminal 264 residues [38] together with an antibody against the MYC epitope tag that is fused to the C terminus of one copy of the ECM11 gene in these strains.
As expected based on the SC-deficient phenotype of the zip1[D21-163] mutant, meiotic prophase nuclei from the zip1[D2-163] mutant strain fail to exhibit extensive Zip1 or Ecm11-MYC linear structures on meiotic chromosomes, but instead display Zip1 or Ecm11-MYC foci of varying sizes, sometimes accompanied by a large "polycomplex" aggregate of these SC central region proteins ( Figure S1). Interestingly, ndt80 meiotic cells expressing zip1 [D2-20] also fail to exhibit any detectable SC formation, even after 24 hours in sporulation medium.
These data in conjunction with the deficient SC assembly phenotype of zip1[D21-163] [22] indicate that residues within both the 2-20 and 21-163 regions of Zip1 are required for Zip1's capacity to assemble SC.
The tripartite nature of the mature SC in budding yeast is reflected by the orientation of Zip1 transverse filament proteins relative to the central element substructure, whereby Zip1 C termini orient toward homologous axes and away from the central element, which is positioned at the midline of the SC. This tripartite organization can be detected using structured illumination microscopy (SIM) on surface-spread meiotic chromosomes labeled with antibodies targeting the C terminal region of Zip1 and SC central element components [19]. With the increased resolution that SIM affords, our anti-Zip1 antibody localizes as a wide ribbon on linear SC structures; one can often observe parallel tracts of Zip1 C termini flanking the central element protein(s) within subsections of such a Zip1 linear element. We used SIM on surface-spread meiotic chromosomes to ask whether the SC structures in zip1[D2-9] meiotic nuclei have this canonical, tripartite organization. We observed no detectable difference in the organization of Zip1 and the central element proteins within SCs assembled by wild type Zip1 versus Zip1[D2 -9] protein: Antibodies targeting the C terminus of Zip1 were observed to flank the SC central element substructure (labeled with antibodies directed against the Ecm11 and/or Gmc2 proteins; see Methods) within SCs assembled by Zip1[D2-9] ( Figure S2).
These observations indicate that, in reciprocal fashion to residues 21-163, residues 2-15 are required for Zip1's MutSg crossover-promoting activity but are dispensable for its capacity to assemble tripartite SC.
In contrast to the robust synapsis observed in zip1[ D2-9] and zip1 [D10-14] strains, extensive coincident linear assemblies of Zip1 and Ecm11 were not detectable in meiotic prophase nuclei from zip1[D15 -20] or zip1 [15-20àA] mutant strains ( Figure 2). Instead, the vast majority of meiotic prophase nuclei from these strains exhibit foci of Zip1 and Ecm11 of varying sizes, which are sometimes, but not always, coincident.
Our phenotypic assessment of three novel non-null zip1 alleles thus indicates that Zip1's first twenty residues correspond to adjacent regions that function somewhat independently of one another: residues 2-14 are essential for normal MutSg crossovers but dispensable for SC assembly, whereas residues 15-20 (and certainly residues 21-163; [22]) are less critical for MutSg crossovers but crucial for SC assembly.

Crossover-deficient, synapsis-proficient zip1 mutants initiate SC assembly from both centromeric and non-centromeric chromosomal sites
While the earliest SC assembly events that occur during meiosis in budding yeast have been found to preferentially initiate at centromeres [32], SC assembly events are also associated with non-centromeric sites (presumably recombination sites) in wild-type meiotic nuclei at intermediate stages of synapsis. In the zip3 mutant, by contrast, new SC assembly events associate predominantly with centromeres at both early and later meiotic prophase stages [32].
As the SC assembly and meiotic crossover phenotypes in zip1[ D2-9] and zip1[D10 -14] strains resemble the phenotypes found in zip3, we asked whether new SC assembly events in zip1[D2 -9] and zip1[D10 -14] meiotic nuclei associate with centromeres more often than wild-type nuclei at intermediate stages of synapsis. Surface-spread meiotic chromosomes from 15, and 18 hour time points were co-labeled with antibodies that target Zip1, and that target the MYC epitope that is fused to the Ctf19 centromere protein in these strains. In order to enrich for new, singular SC assembly events in our analysis, we identified the total number of Zip1 linear stretches measuring between 0.7-1.0 micron in length, and asked which are directly adjacent to (overlapping) a Ctf19-MYC focus. We found that 83% (29 out of 35) of such short SC structures were associated with a Ctf19-MYC focus in zip3 meiotic nuclei, consistent with previouslypublished data. By contrast, 48%, 52% and 64% (35/73, 27/52, and 23/36) of short Zip1 linear structures were associated with Ctf19-MYC zip1[D2 -9], zip1 [D10-14], or ZIP1 ZIP3 meiocytes, respectively ( Figure S4). Thus, in contrast to zip3 mutants, zip1[D2 -9], and zip1[D10 -14] mutants display a robust capacity to initiate SC assembly from non-centromeric sites on meiotic chromosomes.

Adjacent regions within Zip1's first twenty residues have opposing effects on the
SUMOylation of an SC central element protein  demonstrated that SUMOylated Ecm11 is required for SC assembly and that Zip1 along with SIC proteins Zip2, and Zip4 (but not Zip3) are required for a threshold level of Ecm11 SUMOylation during meiosis [29]. This report also revealed that Ecm11 is hyper-SUMOylated in mutants missing the putative SUMO E3 ligase and SIC protein, Zip3. To ask whether the N terminal twenty residues of Zip1 regulate Ecm11 SUMOylation, we evaluated the abundance of Ecm11 forms in meiotic extracts from ndt80 strains homozygous for ZIP1, zip1 [D2-9], zip1 [D10-14], or zip1[D15-20], a zip1 null, or a zip3 null allele.
A Western blot can readily detect three forms of Ecm11-MYC in protein extracts from meiotic cells homozygous for MYC-tagged Ecm11 [19,29,39]. UnSUMOylated Ecm11-MYC migrates near the 75 kD marker on a protein gel, whereas monoSUMOylated and polySUMOylated Ecm11-MYC is positioned near the 100 kD and 150 kD positions, respectively. HyperSUMOylated Ecm11-MYC, which is found in zip3 meiotic extracts, migrates at various positions between the 150 kD and 250 kD markers ( Figure 4A).
We found the proportion of SUMOylated Ecm11-MYC in ZIP1 ZIP3 ndt80 meiotic extracts at the 24 hour time point to be, on average, 14%, wherein 11% of total Ecm11-MYC was in the monoSUMOylated form and 3% was in the polySUMOylated form (three replicates; Figure 4B).
Consistent with prior results [19,29], zip1 null strains exhibited a relatively low level of SUMOylated Ecm11: An average of 4% of total Ecm11-MYC was in the monoSUMOylated form, while polySUMOylated Ecm11-MYC was below levels of detection (less than 1%) 24 hours after placement into sporulation medium ( Figure 4B). Again consistent with prior findings [29], zip3 meiotic extracts exhibited not only an elevated level of polySUMOylated Ecm11-MYC (15% of total Ecm11-MYC, on average, over three replicates; Figure 4B), but also exhibited an abundance of hyperSUMOylated Ecm11 (19% of total Ecm11-MYC, on average, over three replicates; Figure   4B).
We found that residues 15-20 are important for Zip1's capacity to promote Ecm11 SUMOylation. In zip1 [D15-20]  It has been proposed that the hyperSUMOylated forms of Ecm11 that occur in zip3 mutant meiotic cells correspond to Ecm11 linked to poly-SUMO branched chain structures of various sizes and shapes [29,40] The phenotypes of these zip1 point mutants support the idea that while Zip1's first twenty residues encompass both crossover recombination and SC assembly functionalities, adjacent sites within this region maintain different and independent roles in regulating synapsis.

Residues within the 2-14 region influence Zip1's capacity to interface with Zip3 at polycomplex structures
The shared phenotypes of zip1[D2-9], zip1[D10 -14] and zip3 mutants made us wonder whether the N terminus of Zip1 directly or indirectly interacts with the Zip3 protein. Prior evidence for an interaction between Zip1 and Zip3 includes the observation that Zip3 is detected throughout Zip1 polycomplex structures that assemble in contexts where SC assembly fails [26,32]. To explore the possibility that Zip1's N terminus mediates an interaction with the Zip3 protein, we examined the distribution of Zip3 at Zip1 polycomplex structures assembled in spo11 meiotic cells, which fail to initiate recombination and thus also SC assembly [41,42]. These data indicate that, at least in the context of polycomplex structure, Zip1's residues 2-14 mediate a direct or indirect interaction with Zip3.

Residues within Zip1's 2-14 region are critical for Zip3 recruitment to recombination initiation sites
Zip3 and other SIC proteins (such as Zip2, Zip4 and Spo16) form foci that co-localize with MutSg along aligned homologous chromosomes at mid meiotic prophase [26]. Consistent with the notion that such Zip3 foci mark recombination intermediates, Zip3 has been detected at multiple DSB hotspots using Chromatin Immunoprecipitation (ChIP) in conjunction with quantitative PCR (qPCR) [43]. Zip1 was found to be required for the recruitment of Zip3 to the DSB sites examined, thus we asked whether the capacity of Zip1 to recruit Zip3 to DSB sites relies on the N terminal residues of Zip1 that facilitate Zip3's localization to Zip1 polycomplex.
We performed ChIP and qPCR on meiotic cell extracts from ZIP1, zip1 [D2-9], zip1 [D10-14], and zip1 null strains expressing a Zip3 protein with three copies of the FLAG epitope fused to its C terminus. Strains for this experiment were built in the SK1 genetic background, to ensure maximal synchrony over a meiotic time course (SK1 strains enter and/or progress through meiosis more synchronously than the BR strain background). ChIP-qPCR was performed at multiple time points during sporulation, and the time course experiment was performed in triplicate for each strain except for the zip1 null negative control, where the single experiment performed gave results that are consistent with prior published data [43].
We examined Zip3-6xHIS-3xFLAG association with chromosomal sites corresponding to three known DSB hotspots, centromeres, or the chromosome axis [43]. Sequences enriched for Rec8 that are embedded in the proteinaceous chromosome axis are generally anti-correlated with DSB sites, but may associate with DSB repair intermediates, according to a "loop-tether" model for DSB formation in budding yeast [44]. A sequence internal to the large NFT1 open reading frame was previously found to be devoid of Zip3 binding [45,46] and thus served as a normalization control for Zip3 relative enrichment values.
In strains carrying wild-type ZIP1, Zip3 was detected in higher abundance at meiotic centromeres relative to axis or DSB sites within two hours after placement of into sporulation medium, consistent with previously published results. Between two to four hours after placement into sporulation medium, Zip3 became enriched at a chromosome axis site as well as at three DSB hotspots (GAT1, BUD23, and ERG1; Figure 6). Zip3 localization to all sites peaked at the 4 hour time point in ZIP1 meiotic cells, which corresponds to maximal DSB activity at the BUD23 locus in this SK1 strain background [43]. At this four hour time point, GAT1 and BUD23 DSB sites exhibited a more than 20-fold enrichment for Zip3 relative to the NFT1 control ( Figure 6).
Consistent with DSB repair timing in this genetic background, Zip3 enrichment at all sites dramatically diminished between four to six hours and was at pre-meiotic levels by eight hours after placement in sporulation medium. Consistent with prior findings, Zip3 was virtually undetectable at DSB, axis and centromere sites in the zip1 null strain ( Figure 6).
Similar to a zip1 null mutant, little Zip3 was detectable at any of the three DSB sites examined, nor at the axis or centromere sites, in zip1[ D2-9] or zip1[D10-14] strains ( Figure 6).
The phenotype of zip1 [D2-9] in this experiment appeared indistinguishable from the zip1 null, Taken together, our ChIP and cytological studies indicate that residues within Zip1's N terminal twenty amino acids are essential for the proper enrichment of the pro-crossover protein Zip3 to meiotic centromeres, to DSBs, to chromosomal axis sites, and for the accumulation of robust Zip3 and the MutSg foci within the central region of the SC.

Discussion
SC transverse filament proteins from different organisms share no ancestral homology but do share a conserved dual functionality: an activity that facilitates interhomolog crossover recombination events between DNA duplexes and a capacity to assemble the tripartite SC structure on meiotic chromosomes. Despite the absence of primary sequence conservation, SC transverse filament proteins from different phyla typically consist of an extended central "core" predicted to assemble coiled-coil, flanked by predicted unstructured N and C terminal regions.
Sequence alignment of SC transverse filament proteins from related species reveal highly conserved residues throughout the central helical core of the protein, but also small groups of conserved amino acid residues within the N and C predicted unstructured domains (for a mammalian SC transverse filament example, see [47]; Figure 7A illustrates this point for the budding yeast transverse filament, Zip1). The small regions of conservation likely reflect functional residues, perhaps ones that form an interaction interface for a partner protein.
In prior work we reported our discovery that a large in-frame deletion of Zip1's predicted unstructured N terminal region (zip1[D21-163]; originally created in [48]), encodes a separationof-function Zip1 protein that fails to assemble mature SC but is completely capable of executing Zip1's role in MutSg crossing over [22]. Here we describe two novel non-null zip1 alleles,

Two functionally distinct regions within Zip1's N terminal twenty residues: Interaction domains?
The analysis of our novel zip1 alleles moreover leads us to propose the Zip1's first twenty residues correspond to two distinct interaction domains, which directly engage with procrossover and pro-synapsis machinery and/or mechanisms. We propose that these pro-crossover and pro-synapsis domains are spatially positioned adjacent to one another within the N terminal tip of Zip1, as illustrated in the model shown in Figure 7B. One basis for our speculation is that Zip1 protein missing residues 2-9 or 10-14 confers a phenotype that strongly resembles the unique phenotype of cells missing the pro-crossover SIC protein, Zip3: zip1[D2-9], zip1[D10 -14] and zip3 strains [26,28,29]  Additional support for the possibility that Zip1's N terminal residues interact with Zip3 comes from the localization of Zip3 at polycomplex structures, aggregates of Zip1 and other SCassociated proteins that form when SC assembly is compromised. While many if not all SIC proteins have been found to localize to Zip1 polycomplex structures, Zip3's localization shows a greater degree of coincidence with Zip1 throughout the bulk of the polycomplex, relative for example to Zip4 ( Figure 5) or the MutSg component, Msh4 [50], and this localization is completely abolished by the loss of Zip1's residues 2-9 or 10-14 ( Figure 5). Finally, Zip1[D2 -9] and Zip1[D10-14] have lost Zip1's capacity to recruit Zip3 to sites of recombination initiation ( Figure 6).
Based on the absence of SC assembly in meiotic cells expressing zip1[D15 -20], zip1 [15-20àA] and zip1[I18A, I19A], we furthermore conclude that Zip1's first twenty residues correspond to (perhaps in an overlapping manner) at least one interaction domain for an SC assembly factor or complex of factors. Unlike the limited nature of the region within Zip1's N terminal 163 amino acids that is required for MutSg crossovers (twenty residues, based on the fact that zip1[D21-163] is fully capable of MutSg crossing over [22]), groups of residues that are critical for allowing Zip1 to assemble SC may be distributed throughout the entire N terminal region encompassed by residues 15-163, as both zip1[D15 -20] and zip1[D21-163] fail to assemble SC. Nevertheless, we propose that the region overlapping residues 15-20 interfaces with components serving an SC assembly function, based on the fact that alteration of two adjacent residues (I18 and I19) at Zip1's extreme N terminus (a change that is unlikely to alter the overall length or structure of the rod-like protein) completely abolishes Zip1's capacity to assemble SC.
We previously demonstrated that zip1[D21-163] phenocopies the ecm11 and gmc2 null mutant phenotype (a failure in SC assembly but proficiency in MutSg crossing over), suggesting that this N terminal region of Zip1 functionally interacts with the central element in order to assemble SC [22]. Moreover, Leung et. al (2015) demonstrated that Zip1's N terminal 346 residues is sufficient to promote Ecm11 SUMOylation in vegetative (non-meiotic) cells, provided that Gmc2 is also expressed. These data suggest that the N terminal region of Zip1 is able to engage with the Ecm11-Gmc2 proteins, perhaps in a direct manner or perhaps indirectly through a protein expressed in both meiotic as well as mitotic cells [51]. Similar to the uncertainty about whether Zip3 interacts with Zip1 in a direct manner, apart from the genetic interactions found for Zip1 and Ecm11 and the coincidence of Ecm11 and Gmc2 at the midline of SC (where Zip1 N termini also reside [19]), strong evidence of a direct physical interaction between Zip1 and Ecm11 or Gmc2 does not yet exist.

The adjacency between Zip1's pro-crossover and a pro-synapsis regions may serve as a liaison to coordinate SC assembly with intermediate steps in the MutSg crossover pathway.
Finally, we note the tantalizing possibility that the adjacency between the putative pro-crossover and pro-synapsis regions of Zip1's N terminus is functionally important for ensuring that SC assembly occurs in coordination with intermediate steps in the MutSg crossover recombination pathway. Specifically, we speculate that Zip1 may physically connect crossover recombination events to SC assembly through a mechanism that is based, at least in part, on its capacity to stabilize Zip3 at its N terminus. Here, Zip3 (which has SUMO transferase activity [40]) would be expected to be in close proximity to putative pro-synapsis factors stabilized by the adjacent region in Zip1, and could perhaps regulate the extent of SUMOylation of SC central element protein Ecm11. Perhaps negative regulation of Ecm11 SUMOylation by Zip3 is directly involved in ensuring that SC assembly occurs "in the right place at the right time" (i.e. at a MutSg crossover event).

SCs assembled in the absence of MutSg crossing over in budding yeast may be less stable.
During the course of analyzing SC assembly over a time course of meiotic progression in our mutants we obtained an unexpected finding: SCs assembled in zip3, zip1 [D2-9] and zip1 [10][11][12][13][14] strains (corresponding to SCs assembled in the absence of MutSg crossovers) assemble earlier than wild-type SC structures, and appear to be less capable of persisting during an ndt80mediated, meiotic prophase arrest (Figure 3). Pattabiraman (2017) found that a MutSg-associated process affects the dynamic properties of C. elegans SC [33]; our set of preliminary observations raises the intriguing possibility that the MutSg crossover pathway influences the structure and/or dynamics of budding yeast SCs in a similar fashion.

Strains
Yeast strains used in this study are isogenic to BR1919-8B [52] and were created using standard genetic crosses and manipulation procedures. CRISPR-Cas9 methodology was utilized to create  (Table S4).

Secondary antibodies conjugated with Alexa Fluor dyes were purchased from Jackson
ImmunoResearch and used at 1:200 dilution. Microscopy and image processing was carried out using a Deltavision RT imaging system (Applied Precision) adapted to an Olympus (IX71) microscope. Structured illumination microscopy was carried out using Applied Precision's OMX Blaze Structured Illumination Microscope system at The Rockefeller University's Bio-Imaging Resource Center.

Calculations and Statistical Analysis
Genetic crossover data was compiled and processed using an Excel Linkage Macro program, created by Jonathan Greene (Rhona Borts, pers. comm.) and donated by Eva Hoffmann (University of Copenhagen, Denmark). Crossover values (and their standard errors) were obtained using the Stahl lab online tools (http://molbio.uoregon.edu/~fstahl/), with the method of Perkins [54]. Non-mendelian segregation is reported in Table S3. Recombinant spore values were calculated according to the following: 100(r/t), where r= the number of colonies carrying a chromosome which is recombinant in the interval and t; the total number of colonies assessed.

Western Blot
Western blotting was performed as described previously [19] with the following modifications: Amersham Protran 0.2μm NC was used as the transfer membrane following the manufacturer's recommendation; after secondary antibody incubation the membrane was processed with a final wash in 100mM Tris-Cl pH 9.5, 100mM NaCl, 5mM MgCl2 to boost the HRP-mediated chemiluminescence using Amersham ECL Prime Western Blotting Detection Reagent.
2 µg of the mouse monoclonal anti-FLAG antibody M2 (Sigma) and 30 µL Protein G magnetic beads (New England Biolabs) were used. Quantitative PCR was performed from the immunoprecipitated DNA or the whole-cell extract using a 7900HT Fast Real-Time PCR System (Applied Biosystems, Thermo Scientific) and SYBR Green PCR master mix (Applied Biosystems) as described [56]. Results were expressed as % of DNA in the total input present in the immunoprecipitated sample and/or normalized to the negative control site in the middle of NFT1, a 3.5 kb long gene. Primers for GAT1, BUD23, ERG1, Axis and NFT1 loci have been described [45,46,57].   are stained with DAPI to label DNA (white), anti-Zip1 to label Zip1 (green), and anti-MYC to label Ecm11 (magenta). The merge between Zip1 and Ecm11 channels is shown in the final column. Quantitation of the number and cumulative length of SC linear assemblies in wild type as well as each internal deletion zip1 mutant strain is given in Fig. 3. Arrows point to polycomplex structures. Arrowhead indicates a large focus or pair of foci that measures at 0.7 µm and thus may have been included in the assessment of "linear" SC structures (see Fig. 3).
Scale bar, 1 µm.  of Ecm11-MYC in meiotic extracts prepared from ZIP1 ZIP3, zip3, or various zip1 mutant strains (protein alterations caused by each zip1 allele is indicated on the x axis). All strains carry an ndt80 null allele, which causes a meiotic arrest that ensures maximal enrichment of midmeiotic prophase stage cells at 24 hours after placement into sporulation medium [37]. Meiotic extracts were prepared as previously described at the 24 hour time point [19,29].    amino acids 21 and 163 are essential for Zip1's SC assembly function but completely dispensable for MutSg crossing over [22], in contrast to the reciprocal phenotype exhibited by zip1 [D2-9] or zip1[D10 -14]. We note that the region directly adjacent to residues 2-14 (corresponding to residues 15-20) is functionally important for both Zip1's crossing over and synapsis activities, although this region has a bigger impact on Zip1's SC assembly and Ecm11 SUMOylation activity (as demonstrated by the zip1 null phenocopy displayed by zip1 [D15-20] and zip1[I18A, F19A] strains) and plays a lesser role in Zip1's pro-MutSg crossover function relative to the 2-14 region. We speculate that the adjacency between these functionally distinct regions of Zip1's N terminus may mechanistically underlie the coordination between MutSg crossing over and synapsis, via direct molecular communication (blue arrow) between crossover factors and synapsis proteins.

Table 1. MutSg-and Zip1-dependent crossovers rely on Zip1's N terminal residues.
Map distances and interference values were calculated using tetrad analysis or random spore analysis and coefficient of coincidence measurements as described previously [19,39]. 4-spore viable tetrads with no more than 2 gene conversion (non-2:2) events were included in calculations; See Table S3 for gene conversion frequencies. Table indicates  Data from this table is graphed in Fig. 1C.

Table 2. MutSg crossovers rely on twenty residues within Zip1's N terminus.
Map distances and interference values were calculated using tetrad analysis as described previously [39]. 4-spore viable tetrads with no more than 2 gene conversion (non-2:2) events were included in calculations; See Table S3 for gene conversion frequencies. Table indicates map distances and their corresponding percentages of wild-type values for individual intervals, and the map distances and the corresponding percentage of wild type for the entire chromosome (by summing the intervals on III or VIII). For the intervals marked with (n.d.), interference measurements are not obtainable using the coefficient of coincidence method due to an absence of NPD tetrads. Data for strains marked with an asterisk were published previously [22]. Data from this table is graphed in Fig. 1D. While the vast majority of meiotic nuclei from zip1[D15 -20] strains exhibited only abundant and varying-sized foci of Zip1 and Ecm11, occasionally frail-looking linear assemblies of diffuse Zip1 (green) would accompany similar types of Ecm11 linear assemblies (magenta) on surfacespread meiotic chromosomes (labeled with DAPI, white). These assemblies (as well as instances of adjacent large focal deposits of Zip1 and Ecm11) were included in our linear assembly measurements (Fig. 3). The abnormal-looking linear assemblies appear wavy, and often taper at their ends. The top row presents the one rare nucleus out of the (more than 100) nuclei examined, in which these frail linear assemblies were most abundant. We note that unlike the robust linear assemblies of coincident Ecm11 and Zip1 observed in zip1[ D2-9]  where light shading represents the percentage of total ≤1 µm Zip1 assemblies in each strain that are unassociated with a centromere signal. In wild-type strains, 23 out of 36 ≤1 µm Zip1 assemblies (in 10 surface-spread nuclei) were associated with a centromere; In zip3 strains, 29 out of 35 ≤1 µm Zip1 assemblies (in 11 nuclei) were associated with a centromere; In zip1[D2 -9] strains, 35 out of 73 ≤1 µm Zip1 assemblies (in 21 nuclei) were associated with a centromere; In zip1[D10 -14] strains, 27 out of 52 ≤1 µm Zip1 assemblies (in 11 nuclei) were associated with a centromere. A Fishers Exact Test found no significant difference between the proportion of centromere-associated Zip1 stretches in wild-type versus zip3 in this data set (two-tailed P value = 0.107), but did find a significant difference between zip1[D2 -9], and zip3 (two-tailed P value = 0.0007), and between zip1[D10 -14] and zip3 (two-tailed P value = 0.0033).
Sporulation efficiency reflects the fraction of cells that are 2, 3 or 4-spore asci after 5 days on sporulation plates. The frequency of tetrads containing four, three, two, one, or zero viable spores is shown along with the total spore viability (under "% Spore viability"); n.d.= not determined. Full strain genotypes are listed in Table S4. An asterisk indicates data that was previously published [22,39]. Table. Map distances in zip3 and additional zip1 allele strains.

S2
Data display and calculations are as in Table 2. Table. Non-Mendelian segregation (gene conversion events) per locus, measured in 4spore viable tetrads. Shown are the percentages of non-Mendelian segregation events (3:1/1:3 segregation, top; 4:0/0:4 segregation, below) out of the total tetrads analyzed (second column) in each of the indicated strains. Data is derived from 4-spore viable tetrads with no more than 2 gene conversion (non-2:2) events, although cases where adjacent loci segregate non-2:2 were considered a single conversion event. The sum total percentage of observed non-Mendelian events, and the fold increase relative to wild type, is presented, far right. Strains marked with a single asterisk have a different set of genetic markers on chromosome VIII, relative to the wildtype strain used in this analysis. Strains marked with a double asterisk were previously published [22].