Kre28–Spc105 interaction is essential for Spc105 loading at the kinetochore

Kinetochore (KTs) are macromolecular protein assemblies that attach sister chromatids to spindle microtubules (MTs) and mediate accurate chromosome segregation during mitosis. The outer KT consists of the KMN network, a protein super-complex comprising Knl1 (yeast Spc105), Mis12 (yeast Mtw1), and Ndc80 (yeast Ndc80), which harbours sites for MT binding. Within the KMN network, Spc105 acts as an interaction hub of components involved in spindle assembly checkpoint (SAC) signalling. It is known that Spc105 forms a complex with KT component Kre28. However, where Kre28 physically localizes in the budding yeast KT is not clear. The exact function of Kre28 at the KT is also unknown. Here, we investigate how Spc105 and Kre28 interact and how they are organized within bioriented yeast KTs using genetics and cell biological experiments. Our microscopy data show that Spc105 and Kre28 localize at the KT with a 1 : 1 stoichiometry. We also show that the Kre28-Spc105 interaction is important for Spc105 protein turn-over and essential for their mutual recruitment at the KTs. We created several truncation mutants of kre28 that affect Spc105 loading at the KTs. When over-expressed, these mutants sustain the cell viability, but SAC signalling and KT biorientation are impaired. Therefore, we conclude that Kre28 contributes to chromosome biorientation and high-fidelity segregation at least indirectly by regulating Spc105 localization at the KTs.


Introduction
During eukaryotic cell division, kinetochores (KTs) facilitate faithful segregation of genetic material from mother to daughter cells. Each KT is a large protein machine that assembles on a specialized chromatin domain called the centromere and establishes end-on attachments between the sister chromatids and spindle MTs emanating from opposite spindle poles. The budding yeast Saccharomyces cerevisiae has one of the 'simplest' KTs known to date, yet it harbours approximately 70 protein subunits. Components of the yeast KT can be divided into two main categories. The first category contains the centromeric DNA-binding components and their associated network known as CCAN (constitutive centromere-associated network). The second comprises the MT-binding protein network: the KMN super-complex, the fungi-specific Dam1 complex and the MT plus-end-binding protein Stu2 [1][2][3][4][5]. The budding yeast KT incorporates each of these proteins positioned at well-defined average locations along with the KT-MT attachment (figure 1a; electronic supplementary material, figure S1B) [7,8]. For molecular and cell biologists, the budding yeast KT serves as an excellent model to determine the mechanisms underlying KT functions.
Spc105 is an essential KT protein that gets co-purified with the COMA complex subunit Mcm21 and with the MIND complex (Mtw1-Nsl1-Nnf1-Dsn1) [9,10]. It forms a complex with another essential KT protein, Kre28, also known as Ydr532C. Kre28 is an orthologue of human Zwint1, C. elegans Kbp-5 and S. pombe Sos7 [10][11][12][13]. Previous studies, using in vitro and in vivo   Figure 1. Defining the localization of the C-terminus of Kre28 in the metaphase kinetochore using FRET and high-resolution colocalization. (a) The organization of kinetochore proteins along with the microtubule axis in bioriented yeast kinetochores. Positions of C termini of Spc24, Ndc80 and the N terminus of Ndc80 are indicated. (b) Frequency distribution of the distance between the centroids of Kre28-mCherry (Kre28-C) and GFP-Ndc80 (N-Ndc80). The black curve line is the maximum-likelihood fit. (c) Proximity ratio for FRET between fluorophores fused to either Spc24-C or Ndc80-C and Kre28-C (mean ± s.e.m.) in bioriented kinetochore clusters. At least 29 bioriented kinetochores were analysed for this dataset. The p-values obtained from unpaired t-tests are displayed above the plot. (d) Proximity ratio for FRET between adjacent C termini of Kre28 in bioriented yeast kinetochores. Eighty-two kinetochores were analysed to obtain these data. (e) Top: line diagram of Spc105 molecule. The illustration was duplicated from our previous study [6]. Red bars represent PP1/Glc7 recruitment site (amino acid 75-79), and six MELT repeats. Amino acid locations of GFP fusion are shown at the top on amino acid residues 222, 455, 709 and C (917). Bottom: proximity ratio for FRET between Kre28-C and different amino acid positions of Spc105 molecules in bi-oriented kinetochores. At least 35 kinetochore foci were analysed for this graph. The p-values obtained from one-way ANOVA test performed on the data are mentioned above the plot. ( f ) Localization of C termini of Kre28 in KMN network of metaphase kinetochores of yeast cells. experiments, provide some insights into how Spc105 and Kre28 are assembled at the KTs. Still, the specific function of Kre28 remains unclear. How Kre28 localizes within the KT-MT attachment site is also unknown [14][15][16]. Here we define the localization of Kre28 in KT-MT attachment sites of bioriented KTs and elucidate its functional role.

Localization of Kre28 in the KMN network of bioriented kinetochores
We have previously shown that the precise organization and alignment of Spc105 can influence proper spindle assembly checkpoint (SAC) activation and silencing [17][18][19][20]. Kre28, being an essential component of the KT, may also contribute to the organization of Spc105 into the yeast KT. Zwint1, the human orthologue of Kre28, localizes very close to Cdc20 at the human KTs [21,22]. Therefore, Kre28 may play a role in determining the Spc105 organization.
To determine Kre28's position with respect to Spc105, we first had to define the organization of the entire Spc105 protein within the yeast KT. Previous studies show that the C-terminal RWD domain of Spc105 binds directly to the Mtw1 complex and remains in the proximity of Spc24/ Spc25 C-termini [15,23,24]. On the other hand, the N-terminus of Spc105 (abbreviated as N-Spc105) consists of a long, disordered phosphodomain that lies somewhere between the Dam1 complex and the C-termini of Ndc80 and Nuf2 within the Ndc80 complex [6][7][8]. To map out the overall organization of the Spc105 phosphodomain, we inserted a GFP at locations within Spc105 that demarcate domains predicted to possess secondary structure (see electronic supplementary material, figure S1). Additionally, we tagged three different KT subunits to position mCherry at different locations along with the KT-MT attachment (electronic supplementary material, figure S1). Quantification of FRET between the GFP inserted in Spc105 and one of the three mCherry acceptors shows that despite being discorded, the Spc105 phosphodomain localization is mainly limited to a span between the Dam1 complex and the C-terminus of the Ndc80 subunit (abbreviated at Ndc80-C) of the Ndc80 complex. The disordered nature of the phosphodomain also gave rise to FRET between different sections of adjacent Spc105 molecules (electronic supplementary material, figure  S1D). Having established Spc105, we examined the localization of Kre28 by centroid measurement and FRET assay. Previous literature suggested that the C-terminal structured domains of Spc105 harbour interaction sites for Kre28 [22,25].
To define the localization of Kre28 within the KMN network of bioriented KTs, we performed high-resolution colocalization to measure the mean separation between Kre28-C and the N termini of the Ndc80 subunit (N-Ndc80) in the bioriented KTs of yeast (figure 1a) [8]. We observed that the C-terminus of Kre28 is positioned between 45 and 50 nm from N-Ndc80, which is consistent with previously published work with Zwint1 [22].
To determine Kre28 localization with higher resolution, we quantified FRET between Kre28-C with either Spc24-C or Ndc80-C in metaphase cells. We obtained a low to moderate proximity ratio in both cases, indicating that Kre28-C may localize somewhere between C-termini of Spc24 and Ndc80 (figure 1b). The absence of FRET between adjacent Kre28 C termini (Kre28-GFP/Kre28-mCherry) indicated that the C termini of Kre28 molecules are farther apart than 10 nm in metaphase (figure 1c). We also measured FRET between Kre28-mCherry and GFP inserted at different positions of Spc105 (222nd, 455th, 709th, or the C terminus) to find that the C termini of Kre28 are proximal of the KT-binding RWD domain (RING finger, WD repeat, DEAD-like helicases) of Spc105 (figure 1d,e). A previous study suggested that the stoichiometry of Kre28 and Spc105 is 2 : 1 [12]. However, a comparison of the intensity of Kre28-GFP or Kre28-mCherry signal per KT revealed that there is one molecule of Kre28 per Spc105 molecule in bioriented KTs of yeast (electronic supplementary material, figure S2A-D).

Interaction of Kre28 with coiled coil domain of Spc105
Studies of Zwint1 (orthologue of Kre28 in humans) found that it interacts with a domain within amino acid 1980-2109 of human Spc105 [25]. Protein cross-linking experiments also revealed that coiled-coiled domains (CCs) Kre28 128-169 and Kre28 229-259 interact with Spc105 551-711 , the predicted CC of Spc105 [14]. We wanted to uncover the domains within both Kre28 and Spc105 that are necessary for their mutual interactions.
To study these domains in the ex-vivo condition, we first used the yeast two-hybrid assay. We chose Kre28 fragments (amino acid 1-201 and 202-385, based on predicted secondary structure of Kre28, http://www.compbio.dundee.ac. uk/jpred4/index_up.html) and Spc105 CC (455-708) and the C-terminal RWD domain of Spc105 (amino acid 709-917 [15]; see also electronic supplementary material, figure S3A). Both Kre28 FL and Kre28 1-201 showed interactions with CC as indicated by the growth of colonies co-expressing GBD + spc105 CC , GAD+Kre28 FL , and GBD + spc105 CC , GAD+ kre28  in synthetic dextrose plates lacking histidine. Interestingly, we did not see any interactions between Kre28 FL and spc105 CC + RWD (figure 2a). This may be because of the misfolding of spc105 CC+RWD fusion with the GAL4-binding domain (GBD_C1). It is also possible that the RWD domain interferes with the interaction of CC and Kre28, pointing to a regulatory mechanism. To dissect the interaction between spc105 CC and kre28 1-201 more thoroughly, we used smaller fragments (1-126 and 1-80) for our yeast two-hybrid assay with spc105 CC . We did not notice any interaction using these combinations (figure 2a). Furthermore, we saw a significant contrast in colony growth between the combinations of spc105 CC + Kre28 FL and that of spc105 CC +kre28 1-201, which denotes a change in the strength of interaction with spc105 CC . In conclusion, our yeast two-hybrid assay data indicated that Spc105 CC binding domain lies within Kre28 127-201 .
Next, we mapped the Kre28 interacting domain of Spc105 in vivo. We performed domain mapping experiments where we truncated the mid strand domain of Spc105 (amino acid 313-708) at different residues based on predicted secondary structure (figure 2b). We constructed versions of GFP-labelled Spc105 with different truncations in the mid strand domain (Δ313-455 harbouring only the unstructured region, Δ313-507 containing unstructured region and a small helical domain, Δ313-638 that contains unstructured region and an alpha helix-rich domain of CC and Δ313-709 that royalsocietypublishing.org/journal/rsob Open Biol. 12: 210274 encompasses the entire mid strand domain, figure 2b top right) and transformed them in a heterozygous diploid strain (AJY3278, SPC105/Δ::NAT). We examined the localization of these mutants by microscopy. First, KT localization of Δ313-455 and Δ313-507 displayed no discernable difference in localization compared to wild-type (figure 2b, right). Our nuclear localization signal (NLS) analyses (nls-mapper.iab.keio.ac.jp/cgi-bin/NLS_Mapper_y.cgi; see Material and methods section) indicated residues of 337-345 (SSNKRRKLD, score 9.0) and/or that of 599N-625L (score 6.9) contain NLSs. However, previously we have shown that the mutation of 340-KRRK-343 to alanine residues   royalsocietypublishing.org/journal/rsob Open Biol. 12: 210274 does not affect the KT localization of the mutant [26]. Therefore, this region of 313-507 is not essential for the KT localization of Spc105. Truncation of 313-638 or 313-709 completely abrogated KT localization of Spc105. According to our analysis, spc105 599-625 may harbour an NLS, and deletion of this signal may have abrogated KT localization of the mutants expressing spc105 Δ313-638 or spc105 Δ313-709 . We introduced SV40-NLS (NLS SV40 ) at the N-terminus of spc105 Δ313-638::GFP to check if nuclear localization of this mutant rescues its loading at the KTs (figure 2b, bottom left). However, we did not observe any KT-specific localization.
Subsequently, we wanted to check whether these truncation mutants can support cell viability. To address this, we induced sporulation/meiosis in heterozygous diploid strains expressing these truncated molecules of spc105 (Δ313-455 or Δ313-507). We observed that they were able to complement the deletion of endogenous SPC105 (spc105Δ). We can conclude that the domain of 313-507 is not essential for any activity of Spc105 that contributes to cell viability. On the contrary, Spc105 mutants with either 313-638 or 313-709 truncated could not rescue the viability of spc105Δ. Even fusion of SV40-NLS (NLS SV40 ) at the N-terminus of spc105 Δ313-638 did not rescue its ability to support the cell viability in the absence of wild-type Spc105 (figure 2b right and bottom left). This dataset reveals that the proper localization of Spc105 at the KT is essential for its proper function. They also infer that the lack of nuclear localization cannot explain the lethality of the (spc105 Δ313-638 ) mutant.
We confirmed these observations using the plasmid-shuffle assay (data not shown). Therefore, we hypothesized that the domain of Spc105 housed within amino acid 507-638 directly interacts with Kre28. Deletion of this domain abrogates the interaction resulting in delocalization of Spc105 from the KTs.
To biochemically confirm the results of the 2-hybrid and localization experiments, we immunoprecipitated GFPlabelled versions of Spc105 from strains expressing either Spc105 455::GFP (FL) or spc105 Δ313-638::GFP (Δ313-638) and examined if both molecules interact with Kre28 (figure 2c). Immunoprecipitation followed by immunoblot analysis demonstrated that even though Spc105 455::GFP binds Kre28-5xFlag, the mutant of spc105 Δ313-638::GFP is unable to do so, which indicates that a domain harboured within Spc105 507-638 is essential for its interaction with Kre28 and subsequently its recruitment at the KTs. ac.uk/psipred, also see electronic supplementary material, figure S3A). To check which domain of Kre28 is essential for its loading at the KT and interaction with Spc105, we created yeast strains that express GFP-fused Kre28 fragments from the ADH1 promoter. We examined their localization in a diploid yeast strain where one genomic copy of KRE28 is deleted, and the other allele is tagged with mCherry at its C-terminus (figure 3b). As expected, GFP-Kre28 FL localizes at the KT. On the contrary, we could not detect the localization of GFP-kre28 Δ127-182 or fragments with larger truncations at the KTs. When overexpressed from prADH1, GFP-fused versions of Kre28 FL and its truncations revealed a high cytoplasmic GFP signal (figure 3b). Therefore, we performed similar experiments with the SV40-NLS at the C-terminus of GFP-kre28 Δ127-182 . Even in this case, we did not see any KT localization (data not shown). It should be noted that the GFP tagged Kre28 FL and its truncations were expressed at similar levels (figure 3c). Does Kre28 delocalization affect the cell viability when yeast cells express the mutants in the absence of wild-type Kre28? To test this, we sporulated these diploid strains and isolated haploid spores. We observed that the segregants over-expressing truncated versions of kre28 rescued the deletion of endogenous KRE28 (kre28Δ). However, the segregants expressing truncated kre28 molecules from their native promoter (KRE28pr) could not complement genomic KRE28 deletion (figure 3d ). These data indicate that full-length Kre28 is essential for binding with Spc105 and its interaction with the Mtw1 complex. However, truncated kre28 mutants with defective KT localization were able to sustain cell viability when overexpressed. We backcrossed viable spores with the parent strain (YEF473) to avoid background mutations. The segregants isolated from those crosses were subjected to further experiments. Figure 2. (Overleaf.) Kre28 N terminus interacts with the helix-rich mid strand region of Spc105. (a) Yeast two-hybrid assay between Kre28 and CC (Spc105 455-709 ) or RWD (kinetochore-binding domain, Spc105 709-917 ). The plate pictures of synthetic dextrose with histidine dropouts (-HIS) are shown (medium stringent interaction). No growth was observed on plates of adenine dropout (-ADE, high stringent interaction, data not shown). Kre28 FL and other kre28 fragments were fused to Gal4 activation domain (GAD, Prey). RWD and CC fragments of Spc105 were fused to Gal4-binding domain (GBD, Bait). Swapping the fragments between GAD and GBD exhibited background growth with kre28 1-80 fragment on -ADE plates. For controls, please see electronic supplementary material, figure  We observed slower colony growth among the segregants expressing only the kre28 truncation mutants (figures 3d and 5b). The slow growth suggested that these mutants have a high propensity of chromosome missegregation, and this will affect their viability. Among these mutants, we chose kre28 Δ127-182 for further analysis because this is the smallest truncation that has a significant defect in KT localization. To check whether kre28 Δ127-182 affects the biorientation of sister KTs, we tagged Ndc80 and Spc105 individually with mCherry and studied KT biorientation in kre28 truncation mutant (figure 4a,b). We did not find any noticeable defects in the bipolar attachment of metaphase KTs in this mutant (electronic supplementary material, figure S2F). However, we observed a significant decrease (approx. 61%) in Spc105 recruitment in the truncation mutants (figure 4b) with a much smaller change (approx. 24%) in Ndc80 localization (figure 4a).
To further characterize the correlation between Kre28 and Spc105 recruitment to the KT, we varied the amount of either Kre28 or Spc105 per the KTs by exploiting variable Secondary structure of kre28 120-240 royalsocietypublishing.org/journal/rsob Open Biol. 12: 210274 Kre28 FL kre28 D127-182 Ndc80-mCherry intensity <0.0001  royalsocietypublishing.org/journal/rsob Open Biol. 12: 210274 expression of the respective protein using the galactoseinduced GALL promoter and quantified the amount of Spc105 and Kre28, respectively, per bioriented KT (figure 4c(i),(ii)). This quantification revealed that the amounts of KT localized Kre28 and Spc105 are mutually correlated. As the number of molecules of either protein per kinetochore increases, so does the number of molecules for the other. As expected, both numbers saturate at approximately 8 molecules per KT, which is close to the maximal number of Ndc80 complex molecules per yeast KT [27]. Given that Spc105 can localize to KTs even in strains over-expressing Kre28 fragments, these results strongly suggest that Kre28 positively contributes to Spc105 interactions with the Mtw1 complex. We also found that the number of Ndc80 molecules per KT was slightly lower in cells with KTs containing small numbers of Kre28 molecules ( figure 4c(iii)). This result is consistent with our prior work, which showed that a reduction in Spc105 molecules per KT similarly lowers the number of Ndc80 molecules per KT [27].
Kre28-mCh   royalsocietypublishing.org/journal/rsob Open Biol. 12: 210274 Finally, we performed the immunoblot assay following RFP-trap experiments to assess the severity of impairment in the interaction between kre28 Δ127-182 and Spc105. We observed that the co-precipitation of kre28 Δ127-182 with Spc105-mCherry was significantly reduced (approx. 40%; figure 4d; see the figure legend and Material and methods section for details of band intensity calculation and normalization). Moreover, we detected the protein level of Spc105 was reduced in the presence of kre28 Δ127-182 (approx. 69%, figure 4d, panel of anti-DsRed blot). This set of observations imply that the coiled domain Kre28 containing amino acid residue 127-182 plays a significant role in the binding of Kre28 and Spc105. They also suggest that Kre28 plays a role the maintaining the stability of Spc105 protein.

Effect of Kre28 truncation on spindle assembly checkpoint and error correction pathway
Spc105 contains short linear interaction motifs known as MELT motifs that, when phosphorylated by the Mps1 kinase, serve as the scaffold for checkpoint components [17,18]. Moreover, the evolutionarily conserved RVSF motif present in the N-terminus of Spc105 acts as the primary binding motif of protein phosphatase I (PP1) that dephosphorylates the MELT repeats to silence the SAC [20]. Therefore, Spc105 delocalization may affect SAC activation and silencing. We first studied SAC signalling in cells over-expressing kre28 Δ127-182 by treating cell cultures with the MT poison nocodazole and performing flow cytometry to quantify cellular DNA content. The flow cytometry revealed that SAC strength was not discernably different in strains expressing Kre28 FL and kre28 Δ127-182 , as indicated by the arrest of the cell population with 2 N DNA content (figure 5a). However, we have previously shown that this assay cannot detect smaller defects in SAC signalling [6]. To examine the efficacy to detect smaller defects in SAC and that of the error correction process in kre28 mutants, we subjected them to a low dose of another MT poison, benomyl. At its dosage used in this assay, benomyl destabilizes MTs and forces yeast cells to rely on a combination of effective error correction and SAC signalling to ensure chromosome biorientation and accurate chromosome segregation [6]. The strains that express kre28 with larger truncations (Δ127-201, Δ1-201, and Δ1-126) demonstrated growth defects even in non-selective growth media (YPD). We also performed the benomyl-sensitivity assay using a strain where SV40-NLS is fused to Kre28 127-182 at the C-terminus. Fusion of NLS also did not alleviate the effect of kre28 truncation, suggesting that the afore-mentioned phenotypes are not caused by delocalization of kre28 from the nucleus (figure 5b bottom). The growth defects of the kre28 truncation mutants in benomyl-containing media strongly imply that their ability to activate a robust SAC and/or error correction pathway is compromised. Reduced Spc105 loading in the KT of these mutants can explain the observed deficiencies in SAC signalling, error correction or both.
Previous studies showed that the SAC is essential to ensure the fidelity of meiotic segregation [28,29]. Therefore, delocalization of kre28 may also result in defects in meiotic processes. To examine this aspect, we produced zygotes by crossing a and α haploid strains that over-express truncated kre28 in the absence of any wild-type Kre28. However, these zygotes had severe growth defects (figure 5c). We did not find any nuclear fusion defects during mating of these haploid cells (data not shown). Furthermore, when we sporulated these slow-growing homozygous diploid strains (as shown in one of the panels of homozygosis in figure 5c) and dissected tetrads, we observed a significant decrease in the number of viable segregants (figure 5d, right panel). In control experiments, we did not observe any defects in heterozygous diploid formation (figure 5c, left panel) or the growth of viable segregants after subsequent sporulation ( figure 5d, left panel). We have tabulated our observations for the truncation mutants of Kre28, which we studied here (table 1). The observation of truncated kre28 mutants affecting meiosis in yeast is similar to what was previously observed in mouse oocyte meiosis [30].

Discussion
Here we have elucidated how Spc105 and Kre28 are localized and aligned in the MT attachment sites of the bioriented KTs of budding yeast cells. Kre28-C localizes approximately 48 nm away from the N-terminus of Ndc80. This is consistent with observations reported previously with Zwint1, the human orthologue of Kre28 [22]. Our FRET data are also in accordance with previously published biochemical and cell biological studies [14,15,22]. Interestingly, we could not determine the FRET between Kre28-C and components of the Mtw1 complex because strains expressing the fluorescent fusions were inviable. The fluorescent tags probably disrupt interactions between Spc105/Kre28 and the Mtw1 complex, which results in synthetic lethality [14]. Most strikingly, we observed that the stoichiometry of Kre28 to Spc105 is 1 : 1, whereas it was previously thought to be 2 : 1 [12].
Results of the yeast two-hybrid assays involving Kre28 FL and Spc105 455-709 were consistent with previously published data from Yanagida lab [15,25]. However, it was unclear why the yeast two-hybrid assay did not work between Kre28 FL and spc105 455-917 (spc105 CC+RWD ). It may be the case that CC + RWD (Spc105 455-917 ) fusion with GAL4-binding domain (GBD_C1) did not fold in a way that they can interact with Kre28. On the other hand, it is also possible that RWD interfering with the binding of CC and Kre28 has an unknown physiological significance. Although we obtained yeast two-hybrid interaction between kre28 1-201 and spc105 455-709 , we did not detect any KT localization of kre28  . It was also not sufficient to support cell viability in the absence of wild-type Kre28. Much to our surprise, even after taking the data of Herzog lab into consideration, when we maintained the two binding sites of Spc105 (kre28 128-169 and kre28 229-259 ) in our kre28 cassette (kre28 Δ1-126 ) [14], we did not see any KT localization. After considering these data, we can conclude that full-length Kre28 is essential for its localization and for full KT recruitment of Spc105.
Despite being an essential and conserved component of the KT, the structure of Kre28 remains unknown, and the absence of a structure prevents a clear understanding of the implications of our data. Therefore, to provide a structural context to our interaction-mapping and KT localization experiments, we used an implementation of Alphafold2 to predict structures of protein complexes using Google royalsocietypublishing.org/journal/rsob Open Biol. 12: 210274 Colabfold (https://colab.research.google.com/github/sokrypton/ColabFold/blob/main/beta/AlphaFold2_advanced. ipynb [31]). Using Colabfold, we predicted the structure of a heterodimer of Kre28 FL and the C-terminal region of Spc105 451-917 (electronic supplementary material, figure  S3B). The N-terminal boundary of Spc105 used in the structure prediction was chosen based on the start of the structured region within Spc105. Alphafold predicts a 'T' shaped architecture for the Kre28-Spc105 451-917 heterodimer with high confidence for most residues predicted to possess secondary structure (scores 97%-23%, electronic supplementary material, figure S3B, C). The Pairwise Alignment Error matrix suggests with high confidence that extensive interactions occur among residues in the N-and C-terminal halves of the two proteins. These interactions give rise to two distinct domains within the heterodimer, but the relative organization of these domains cannot be inferred (electronic supplementary material, figure S3D). The structure correlates very well with the CC predictions for Kre28 and Spc105 (electronic supplementary material, figures S3A, S3E). It is also consistent with our limited FRET data. We detected significant and similar FRET between Kre28-mCherry and either Spc105 709::GFP or Spc105-GFP (figure 1). In the predicted structure, the separation between Kre28-C and Spc105 709 and Spc105-C is similar (4.6 nm and 4.8 nm, respectively, electronic supplementary material, figure S3F). Interestingly, we did not detect any FRET between Kre28-mCherry and a Spc105 455::GFP , even though the model predicts a separation of 7.1 nm between these two points (electronic supplementary material, figure S3F). However, the model also predicts that the donor and acceptor fluorophores will be in two different domains, and the relative configuration of the two domains is unknown (electronic supplementary material, figure S3D). Therefore, the actual separation between Kre28-C and Spc105 455 is likely to be greater than 10 nm. The predicted model is also in agreement regarding the residues in Kre28 identified by a previous study to be critical in mediating its interaction with Spc105 [14]. In conclusion, the extensive interaction interface between the two proteins with involvement from residues over nearly the entire length of Kre28 predicted by the model can explain why we could not eliminate the interaction between Spc105 and Kre28 using extensive truncations in either protein.
Our data of Spc105 protein becoming destabilized in the kre28 Δ127-182 mutant (figure 4d, panel of anti-DsRed blot) imply that impairment of Kre28-Spc105 interaction significantly affects Spc105 integrity of expression. This implication is in accordance with a previously reported study by Zhang et al. [32] where they depleted Zwint1 by RNAi and demonstrated that the cellular protein level of hSpc105 is depleted. A study in the fission yeast S. pombe also revealed that KT proteins like Spc105 are surveilled by protein quality control machinery that includes Hsp70, Bag102, the 26S proteasome, Ubc4, and the ubiquitin-ligases Ubr11 and San1 [33]. In that study, the authors suggested that cells employ this mechanism to maintain the homeostasis of nuclear components and genomic integrity. We came across another indirect evidence of protein quality regulation of Spc105 and Kre28 from Yong-Gonzales and colleagues [34]. They showed that both of these proteins become sumoylated by Smc5-Smc6 complex and deleterious mutations in Smc5-Smc6 complex leads to chromosome loss [34]. Consistent with the above-mentioned studies, the Biggins lab also discovered Mub1/Ubr2 ubiquitin ligase complex to be a part of a quality control mechanism that monitors KT protein Dsn1 [35]. A similar mechanism probably controls Spc105 and/or Kre28 levels in S. cerevisiae. Does Kre28 act as a chaperone to stabilize the recruitment of Spc105 at the KT? Some of the previous studies argue against this hypothesis. In human KT, Zwint1 is dispensable for the interaction between hSpc105 and Mis12 complex [15,16]. While performing ex-vivo KT assembly experiments, Biggins lab showed that Ipl1 phosphorylation of Dsn1, which triggers outer KT assembly, also recruits Kre28, which should be specific to mitotic cells [36]. Contrastingly, we observed a similar level of Kre28 at the KTs at every stage in the cell cycle, including the G1-S phase (unbudded and small budded cells, data not shown), which implies that Kre28 loading at the KT takes place at the same time point as loading of Spc105. Combining our observations with those from Herzog lab, we can conclude that kre28 127-182 contributes to the main interaction between Kre28 and Spc105 and kre28 229-259 contributes to interaction with the Mtw1 complex. However, full-length Kre28 is essential for proper binding with Spc105, their mutual recruitment and their activity at the KTs. Kre28 may also become phosphorylated by Ipl1, which can trigger its association with Spc105 and subsequently their loading at the KTs.
Does Kre28 have a specific function in SAC and error correction or during meiosis? The results of the functional assays (figure 5) clearly show that the delocalization of Kre28 from the KT impairs the processes. However, all these phenotypes may be linked with the delocalization of Spc105. A similar conclusion has been noted for Zwint1 [22,32,37]. Our experiment where we observed a significant number of inviable segregants after sporulation of homozygous diploid strains expressing only truncated kre28 (kre28 Δ127-182 ) are also in agreement with a study done by Dong Woo Seo and colleagues [30]. According to that study, Zwint1 depletion results in erroneous chromosome alignment and a high frequency of aneuploidy during mice oocyte meiosis. Our study suggests that this function of Zwint1 is also conserved in yeast Kre28.

Plasmid and strain construction
The strains and plasmids used in this study are documented in tables 2 and 3, respectively. Yeast strains containing multiple genetic modifications were constructed by standard yeast genetic techniques. GFP (S65T) and mCherry fusion of proteins were used to localize KTs by fluorescence microscopy. The C-terminal tags like GFP, mCherry, 5xFlag and gene deletion cassettes like spc105Δ::NAT and kre28Δ:: NAT were introduced at the endogenous locus through homologous recombination of PCR amplicons [38]. A 7-amino acid linker (sequence: 'RIPGLIN') bridges the tags (GFP, mCherry, or 5xFlag) from the C-termini of the tagged proteins. Earlier, we observed that the intensity of mCherry-tagged KT proteins varies significantly from one transformant to another for the same strain, due to inherent variability of the mCherry brightness. Therefore, to construct all the FRET strains with Ndc80, Stu2, Nsl1, Kre28 and Ask1-mCherry, we crossed a specific mCherry strain with haploid strains of all GFP-fused Spc105 alleles and sporulated the heterozygous diploids to obtain the desired segregants.
To construct the yeast strains with internally tagged Spc105 mentioned in electronic supplementary material, figure S1, and the strains with truncated Spc105 mentioned in figure 3, first we used BstEII digest of pRS305 chimera or StuI digest of pRS306 chimera of Spc105-GFP fusion alleles to transform AJY3278 (SPC105/Δ::NAT) that was later sporulated to obtain the haploid segregants expressing only GFP fusion copy of Spc105. To construct some these strains, first we deleted the genomic copy of SPC105 in a strain already supplemented with the wild-type SPC105 gene expressed from centromeric plasmid yCP50 (URA3). Then, the pRS315 chimera containing Spc105-GFP alleles were transformed in that strain. After that, the strains expressing only Spc105-GFP alleles were generated by negative selection for yCP50 on 5-FOA plates.
The construction of chimeras for over-expression of Kre28 FL or truncations was achieved by cloning N-terminal GFP tagged fusions of Kre28 in a centromeric plasmid pRS414, where KRE28 ORF is flanked by ADH1 (alcohol dehydrogenase 1) promoter and GFP ORF at the upstream and CYC (Cytochrome C) terminator at the downstream (obtained from Dr Mara Duncan's lab, department of Cell and Developmental Biology, University of Michigan). Kre28 fragments were cloned in BamHI-PstI sites. SV40-NLS was cloned within PstI-SalI. For wild-type expression, Kre28 FL and its truncations were cloned in BamHI-PstI of a pRS305 plasmid. They were expressed as N-terminal GFP tagged fusions by its own promoter and terminator. BstEII digests of these chimeras are transformed in AJY3298 (kre28Δ:: NAT/KRE28-mCherry-Hyg) to check for their localization. To create diploid zygotes, two strains of a and α mating types are mixed with each other and spotted on YPD plate which was incubated for 3-4 h at 30°C. To induce sporulation, diploid yeast cells were grown in YPD overnight to stationary phase. Next day cells were pelleted down and resuspended with starvation media (0.1% yeast extract, 1% potassium acetate, 0.025% dextrose) and incubated 4-5 days at RT.

Yeast two-hybrid assay
We performed yeast two-hybrid experiments by co-transforming both of prey ( pGAD_C1) and bait ( pGBD_C1) chimera in strain AJY3802 (PJ69A) [39]. Then, we streaked two of the transformants for each prey-bait pair on synthetic dextrose plates of histidine dropout (-HIS) and dropout of histidine and adenine (-HIS-ADE). Plates were incubated in 32°C for at least 3 days.

Benomyl-sensitivity assay
This experiment was performed as described previously [6,26]. Starting from 0.1 OD 600 of log-phase cultures, we prepared 10-fold serial dilutions and frogged or spotted them on YPD and YPD containing 20 µg ml −1 and 30 µg ml −1 benomyl. At least two biological replicates were used, and spotting was repeated twice for each set of experiments. The plates were incubated at 32°C and pictures were taken royalsocietypublishing.org/journal/rsob Open Biol. 12: 210274 Table 2. Strains used in this study. royalsocietypublishing.org/journal/rsob Open Biol. 12: 210274 after 2 (YPD) to 3 (YPD + Benomyl) days. For space limitations, we showed only YPD + Ben 30 plates. For all spotting assays with benomyl, we used YEF473 or strain expressing Spc105 222::GFP as wide-type ( positive control) and mad2Δ or sgo1Δ as negative controls. Before spotting, we grew the strains expressing truncated kre28 or Kre28 FL control in synthetic dextrose media (Sd-Trp) without Tryptophan to culture only cells carrying the pRS414 chimera. As shown in the plate images, we also used strains expressing Kre28 FL -mCherry along with truncated kre28 as controls which did not show any discernable difference in growth, compared to wild-type.

Microscopy and image acquisition and analyses
A Nikon Ti-E inverted microscope with a 1.4 NA, 100X, oil-immersion objective was used for experiments mentioned in the paper [40]. A ten-plane Z-stack was acquired (200 nm separation between adjacent planes). To measure Ndc80 and Spc105-mCherry, an extra 1.5 x opto-var lens was used. We measured total fluorescence of KT clusters (16 KTs in metaphase) by integrating the intensities over a 6 × 6 region centred on the maximum intensity pixel. We used median intensity of pixels immediately surrounding the 6 × 6 area to correct for background  royalsocietypublishing.org/journal/rsob Open Biol. 12: 210274 fluorescence. The calculation was performed using semiautomated MATLAB programs as described earlier [41]. FRET, high-resolution colocalization, fluorescence distribution analyses and analyses of the images were performed as previously described [7,8,27,40,42]. While measuring proximity ratio, we considered any value below 0.10 as no FRET (mean of the data marked as black), range between 0.10 and 0.3 as medium to low FRET (average of the data marked as red) and any values above 0.3 as high FRET (mean of the data marked as red).

Titration of Kre28 and Spc105 proteins levels and quantification of Kre28, Spc105 and Ndc80 intensities
We grew the strains with prGALL-SPC105 or prGALL-KRE28 in the presence of raffinose (2%). On the day of the assay, we supplemented the media with variable amounts of galactose as discussed previously (2%, 1.5%, 0.5%, 0.2%, 0.1% and 0.05%) [27]. We determined the number of Kre28, Spc105 and Ndc80 from their intensities as reported previously royalsocietypublishing.org/journal/rsob Open Biol. 12: 210274 [27]. We first deduced the fluorescence intensities of Kre28-mCherry, Ndc80-GFP and Spc105-GFP from bioriented KTs. We used AJY939 (Ndc80-GFP, Spc25-mCherry) as a reference to obtain the intensities for known number of Ndc80 and Spc25 molecules at the bioriented KTs. AJY939 was cultured under same imaging conditions as the experimental strains, and the calibration data were acquired throughout the duration of this study. This calibration accounted for alteration in the microscope and imaging technique set-up over time.
We used the values of Ndc80-GFP and Spc25-mcherry to determine the number of molecules of Spc105-GFP and Kre28-mCherry that were loaded in the bioriented KTs.

GFP-trap and RFP-trap assay to pull down Spc105 and immunoblot assay
We used AJY6273 and AJY6274 for GFP-trap experiments. As mentioned in the strain list, AJY6273 expresses Spc105 455::GFP from a centromeric plasmid (pRS315) and genomic SPC105 allele is deleted. The truncation of 313-638 affects the cell viability; hence in AJY6374, the genomic SPC105 allele was intact and spc105 Δ313-638 was expressed from a centromeric plasmid (pRS315). We grew both the strains in synthetic dextrose media devoid of leucine (SD-LEU) to maintain the centromeric plasmid before harvesting them for cell lysis. For RFP-trap assays, we used AJY3483 and AJY3484 (see strain list for detailed information on their genotypes). We grew the strains in SD-TRP (synthetic dextrose devoid of tryptophan) media till late log-phase before harvesting the cells. We lysed the cells by glass-beads in the presence of buffer H 0.15 (25 mM HEPES of pH 8.0, 2.0 mM MgCl 2 , 0.1 mM EDTA of pH 8.0, 0.5 mM EGTA-KOH of pH 8.0, 15% glycerol, 0.1% IGEPAL-CA-630 and 150 mM KCl), supplemented with 0.2 mM PMSF, protease inhibitor cocktail and phosphatase inhibitor cocktails [43]. We isolated clear lysates of the strains and from there, we incubated equal amount of lysates with pre-equilibrated beads of GFP-trap (Chromotek, gta-20) or RFP-Trap (Chromotek, rta-20) overnight. Next day, we washed the beads with post IP wash buffer (composition as mentioned above) with and without 2 mM Dithiothreitol, before boiling them in the presence of 1xSDS loading buffer. After subjecting the proteins through SDS-PAGE, we transferred them to nitrocellulose membrane which we blocked with 5% skimmed milk in 1x phosphate-buffered saline-Tween (PBS-T, 137 mM NaCl, 10 mM phosphate, 2.7 mM KCl, 0.05% Tween 20). We probed the blot with Mouse anti-GFP (JL8, Living Colors, Takara, 1 : 3000) or Mouse anti-Ds-Red (Santa Cruz Biotechnology, sc-390909, 1 : 2000), Mouse α-Flag M2 (Sigma-Aldrich, 1 : 5000) and HRP conjugated secondary anti-Mouse (Sigma-Aldrich, 1 : 10 000). ECL (Thermo Scientific) treatment was used to develop the western blot. We exposed the anti-Ds-Red blots to X-ray films. Anti-GFP and anti-Flag blots were imaged by C600 imager (Azure Biosystems).
We calculated the band intensities by ImageJ in RFP-TRAP assay. Then, we normalized both Spc105 and Kre28 intensities of the truncation mutants with those of the full-length proteins. After that, we normalized Kre28 IP intensities with that of normalized Kre28 input values. Following that, we divided the Kre28 IP intensities with Spc105 IP intensities. We mentioned those values as fold difference at the bottom of the immunoblot panels (figure 4d ).

Flow cytometry
We performed flow cytometry as described previously [6,26]. For strains expressing Kre28_FL or truncated kre28, we started with overnight inoculums grown in Sd-Trp and shifted to YPD to grow till early to mid-log phase before supplementing the media with Nocodazole (final concentration 15 µg ml −1 ) or DMSO control. We collected cell samples at 0, 1, 2, 3 h postdrug treatment and fixed them with 70% ethanol before storing them at 4°C overnight. Next day, after removing the Ethanol, treated the samples with bovine pancreatic RNase (Millipore Sigma, final concentration 170 ng ml −1 ) at 37°C for 6 h-overnight in RNase buffer (10 mM Tris pH 8.0, 15 mM NaCl). After that, we removed the RNase and resuspended the cells in 1x PBS. We treated the samples with propidium iodide (Millipore Sigma, final concentration 5 mg ml −1 in PBS) for at least 1 h at room temperature before analysing them using the LSR Fortessa machine (BD Biosciences) in Biomedical research core facility, University of Michigan medical school. We analysed and organized the data using FlowJo software (FlowJo_V10.7.1_CL).

Prediction of coiled-coil domains in Kre28 and spc105 451-917
We predicted the CC of full-length Kre28 and spc105 451-917 by inserting their amino acid sequences on the website of https://embnet.vital-it.ch/software/COILS_form.html [46]. We used MTIDK matrix for the prediction. We downloaded the output in postscript format and further configured by adobe illustrator.