Methylation of CENP-A/Cse4 on arginine 143 and lysine 131 regulates kinetochore stability in yeast

Abstract Post-translational modifications on histones are well known to regulate chromatin structure and function, but much less information is available on modifications of the centromeric histone H3 variant and their effect at the kinetochore. Here, we report two modifications on the centromeric histone H3 variant CENP-A/Cse4 in the yeast Saccharomyces cerevisiae, methylation at arginine 143 (R143me) and lysine 131 (K131me), that affect centromere stability and kinetochore function. Both R143me and K131me lie in the core region of the centromeric nucleosome, near the entry/exit sites of the DNA from the nucleosome. Unexpectedly, mutation of Cse4-R143 (cse4-R143A) exacerbated the kinetochore defect of mutations in components of the NDC80 complex of the outer kinetochore (spc25-1) and the MIND complex (dsn1-7). The analysis of suppressor mutations of the spc25-1 cse4-R143A growth defect highlighted residues in Spc24, Ndc80, and Spc25 that localize to the tetramerization domain of the NDC80 complex and the Spc24-Spc25 stalk, suggesting that the mutations enhance interactions among NDC80 complex components and thus stabilize the complex. Furthermore, the Set2 histone methyltransferase inhibited kinetochore function in spc25-1 cse4-R143A cells, possibly by methylating Cse4-K131. Taken together, our data suggest that Cse4-R143 methylation and Cse4-K131 methylation affect the stability of the centromeric nucleosome, which is detrimental in the context of defective NDC80 tetramerization and can be compensated for by strengthening interactions among NDC80 complex components.


Introduction
Nucleosomes containing the centromeric histone H3 variant CENP-A mark the region on eukaryotic chromosomes where the kinetochore is assembled, which connects the centromeric DNA to the spindle microtubules [reviewed in (McAinsh and Marston 2022)].Proper assembly of this macromolecular complex is crucial for correct chromosome segregation in mitosis and meiosis, and errors in this process can lead to aneuploidy.In higher eukaryotes, the centromeric chromatin is a broad domain (regional centromere) that is composed of CENP-A nucleosomes interspersed with canonical, H3-carrying nucleosomes (Allshire and Karpen 2008).The CENP-A nucleosomes are generally thought to be akin to conventional nucleosomes with the DNA wrapped left-handed around the histone octamer core, though a righthanded nucleosomal structure also has been proposed (Furuyama and Henikoff 2009;Diaz-Ingelmo et al. 2015).In contrast to higher eukaryotes, the yeast Saccharomyces cerevisiae has compact kinetochores that assemble on a single centromeric nucleosome (called a "point" centromere) and attach on their other end to a single microtubule (Furuyama and Biggins 2007).The fact that one nucleosome per chromosome is sufficient for faithful chromosome segregation in yeast may be attributed to the fact that Cse4 has a longer N-terminus than CENP-A homologs (135 amino-acids, Cse4N) that is essential for centromere function (Stoler et al. 1995;Keith et al. 1999).In earlier work, we and others showed that Cse4N interacts with two components of the constitutive centromere-associated network (CCAN) complex, Okp1 CENP-Q and Ame1 CENP-U , and thus recruits CCAN to the centromere (Anedchenko et al. 2019;Fischböck-Halwachs et al. 2019).Furthermore, we identified post-translational modifications in Cse4N, the methylation of arginine 37 (R37me) and acetylation of lysine 49 (K49ac), as negative regulators of this interaction (Samel et al. 2012;Anedchenko et al. 2019).Also, we found that phosphorylation of S33 regulates the deposition of Cse4 at the centromere (Hoffmann et al. 2018) (Fig. 1a).Cse4 carries other phosphorylation sites (Boeckmann et al. 2013), and it is ubiquitinated and sumoylated (Hewawasam et al. 2010;Ranjitkar et al. 2010;Ohkuni et al. 2020).
The structure of CCAN bound to the Cse4-containing nucleosome (termed CCAN-Cenp-A Nuc ) shows that CCAN assumes a "Y"-shaped structure that clasps onto the nucleosome (Yan et al. 2019).On its nucleosome-distal side, CCAN interacts with the MIND (Mis12/Mtw1) complex (Hornung et al. 2014), which itself forms an elongated, Y-shaped rod and contains the proteins Fig. 1.CENP-A Cse4 is methylated at arginine 143 and lysine 131.a) Overview of the amino acid sequence of Cse4.R143me and K131me sites are located in the α-N-helix and C-terminal to the Cenp-A N helix, respectively.Cse4 is phosphorylated at S33 (S33ph), methylated at R37 (R37me), and acetylated at K49 (K49ac).b) R143 corresponds to R52 in canonical H3.Pairwise alignment of yeast H3 and Cse4 was generated with EMBOSS needle.c) Top, location of Cse4-R143 (red) and -K130 (magenta) in the context of CCAN-Cenp-A Nuc (PDB 6QLD).K130 is shown, because K131 is not visible in the structure.Bottom, view of the α-N and Cenp-A N helices and the location of R143 and K130 (as a proxy for K131) of CCAN-Cenp-A Nuc .The CCAN subunits were omitted for clarity.
On the chromatin-distal side, MIND interacts with the main microtubule-binding entity of the kinetochore, the NDC80 complex (NDC80c) (Dimitrova et al. 2016).NDC80c is a heterotetrameric complex consisting of the proteins Ndc80, Nuf2, Spc24, and Spc25 that associate as two rod-like heterodimers (Ndc80/ Nuf2, Spc24/Spc25) to form an elongated complex of ca.600 Å in length (Janke et al. 2001;Wigge and Kilmartin 2001;Cheeseman et al. 2006;DeLuca et al. 2006;Wei et al. 2007).The two heterodimers each have an α-helical, coiled-coil shaft and globular domains at the ends (Wei et al. 2005;Ciferri et al. 2008), of which the Ndc80 and Nuf2 globular domains are microtubule-binding modules (Wei et al. 2007;Ciferri et al. 2008) and those of Spc24 and Spc25 bind components of the MIND complex in the inner kinetochore (Malvezzi et al. 2013;Hornung et al. 2014).The two α-helical coiled coils join end to end and form a tetramer junction (Valverde et al. 2016).The interactions of NDC80c with the microtubules and other kinetochore subcomplexes are tightly regulated by control mechanisms such as the spindle-assembly checkpoint (Santaguida and Musacchio 2009).
In this work, we describe two novel modifications on CENP-A/ Cse4 in S. cerevisiae, the methylation of arginine 143 (R143me) and methylation of lysine 131 (K131me).These residues are located in the core domain of the centromeric nucleosome, close to the DNA duplex that is partially unwrapped in the centromeric nucleosome (Yan et al. 2019;Migl et al. 2020).We found that the absence of Cse4-R143me enhanced the centromeric defect of mutations in NDC80c (spc25-1) and MIND (dsn1-7).A screen for suppressors revealed intragenic mutations in SPC25 as well as mutations in SPC24, NDC80, and SET2, which encode the histone methyltransferase for H3 K36 (Strahl et al. 2002).Genetic analysis suggests that Set2 acts at the kinetochore by methylating Cse4-K131.The location of the suppressor mutations in NDC80c indicates that spc25-1 destabilizes the tetramerization junction and that the suppressor alleles strengthen interactions among the NDC80c components and thus suppress growth defects.Altogether, we propose that the absence of Cse4-R143me and -K131me affects the stability of the centromeric nucleosome, which is detrimental in the context of a compromised NDC80 complex.

Yeast strains and plasmids
The S. cerevisiae strains and plasmids used in this study are listed in Supplementary Tables 1 and 2 (Supplementary File 1), respectively.Yeast was grown and manipulated according to standard procedures (Sherman 1991).Yeast was grown on full medium (YPD) and selective minimal plates (YM).Alleles of cse4 were generated by site-directed mutagenesis in CSE4-bearing plasmids, and correct mutation was verified by sequence analysis.If plasmid-borne cse4 alleles were tested, strains were constructed that carried cse4Δ, and cell viability was maintained with a URA3-CSE4 plasmid (AEY2781, 5064).The cse4 allele was introduced on a HIS3-carrying plasmid, and the URA3-CSE4 plasmid was removed by counter-selection on medium containing 5-fluoro-orotic acid (5-FOA).
To test the effect of H3 K36A, strains were constructed that carried deletions of the cassettes HHT1-HHF1 and HHT2-HHF2 and whose viability was maintained by a URA3-marked plasmid carrying HHT1-HHF1 (AEY7040, AEY7042, and AEY7045).TRP1-marked plasmids with hht1-K36A-HHF1 or HHT1-HHF1 as a control were transformed into the strains, and the URA3-marked HHT1-HHF1 plasmid was removed by counter-selection on 5-FOA.The hht1-K36A-HHF1 plasmid was obtained from a library of yeast strains carrying plasmid-borne histone mutations (Nakanishi et al. 2008) by amplifying the plasmid from the appropriate strain in Escherichia coli.The presence of H3 K36A was verified by sequence analysis.
Genomic integration of cse4 alleles was performed by cloning the allele on a URA3-marked integrative vector and introducing it into the strain by integrative transformation followed by loopout on 5-FOA.For genetic crosses, cse4 was marked with HisMX by integration of the selection marker downstream of the open reading frame.

Mass spectrometric analysis of Cse4
Purification and analysis of 3xHA-tagged Cse4 from yeast cells were performed as described (Samel et al. 2012) (see Supplementary File 1 for details).

Screen for suppressors of spc25-1 cse4-R143A and whole-genome sequence analysis
The spc25-1 cse4-R143A strain (AEY6838) was spread at a density of 3 × 10 6 cells on YPD plates.Plates were UV-irradiated with 3000 µJ/ cm 2 (wave length 254 nm) in a Stratalinker 2400 UV Crosslinker and subsequently incubated for 4-5 days at 34°C in the dark.This UV dosage gives a survival rate of ∼60% when cells are grown at 23°C.Surviving colonies were re-streaked and grown on YPD at 23°C.Growth at different temperatures was determined by plating serial dilutions of the strains on YPD and incubating the plates at different temperatures for 3 days.
For sequencing of the spc25 and cse4 alleles of the suppressor strains, the genes were PCR-amplified, and the PCR products were subjected to DNA sequencing.In spc25-1, we observed a silent mutation (Spc25-Q38, CAA to CAG) in addition to the reported L25P exchange (Wigge and Kilmartin 2001).
For whole-genome sequence analysis (WGS), strains were grown individually, and groups of five strains were mixed in equal quantities of cells (OD 600 equivalents).DNA was extracted from 10 such pools, sequencing libraries were prepared, and sequencing was conducted on an Illumina NovaSeq6000 instrument (paired-end sequencing, 150 bp) to obtain ∼5 × 10 6 read pairs per pool.Reads were aligned to the reference genome (SacCer3) using Bowtie2, and single-nucleotide polymorphisms (SNPs) were called using deepSNV (Gerstung et al. 2012).The presence of suppressor mutations in individual strains was verified by sequence analysis of PCR-amplified alleles.Sequencing reads were deposited in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) at http://www.ncbi.nlm.nih.gov/sraunder accession no.PRJNA933081.
To verify whether mutations are causative for the suppression of spc25-1 cse4-R143A, the segregation of temperature resistance with the mutation was determined in backcrosses with a spc25-1 cse4-R143A parental strain and sequencing of PCR-amplified alleles.Furthermore, for spc24-D37A, spc24-S59L, and ndc80-L681W, five temperature-resistant and five temperature-sensitive segregants from the backcross were pooled and subjected to WGS.All temperature-resistant and no temperature-sensitive segregants carried the respective suppressor mutation.

CENP-A Cse4 is methylated on arginine 143 and lysine 131
Here, we report the identification of two novel modifications in residues of Cse4 that are within the core region of the centromeric nucleosome, rather than in the N-terminus.In purifications of Cse4 from S. cerevisiae cells that were analyzed by gel-enhanced liquid chromatography mass spectrometry (GeLCMS) (Samel et al. 2012), we identified mono-methylation of arginine 143 (R143me1) and mono-methylation of lysine 131 (K131me1) (Fig. 1a, Supplementary Fig. 1 in Supplementary File 1).Specifically, collision-induced dissociation fragmentation analysis of trypsindigested Cse4 identified the peptide 131-KYTPSELALYEIR-143 as monomethylated.Inspection of the MS/MS spectrum of the parent ion allowed us to uniquely assign the methylation site to R143 (Supplementary Fig. 1a in Supplementary File 1).Furthermore, fragmentation analysis identified the same peptide as being monomethylated at K131 (Supplementary Fig. 1b in Supplementary File 1).Of note, Cse4-K131 can be ubiquitinated in vitro by the E3 ubiquitin ligase Psh1, but whether this occurs in vivo was not determined (Hewawasam et al. 2010).
It is interesting to consider the position of these modifications in the context of the CCAN-bound centromeric nucleosome (CCAN-Cenp-A Nuc ) (Yan et al. 2019).The R143 methylation site corresponds to R52 in histone H3 (Fig. 1b) and lies within the α-N helix of Cse4 that is in close proximity to the DNA of the Cse4 nucleosome.However, in contrast to the canonical H3 nucleosome, where the corresponding helix is wedged between two DNA helices (Fig. 1c) (Luger et al. 1997), the DNA in CCAN-Cenp-A Nuc is more loosely wrapped (Migl et al. 2020), and the α-N helix of Cse4 is not wedged between the helices.Rather, a helix N-terminal to α-N in one of the Cse4 units of the nucleosome (aa 111-129, termed Cenp-A N ) is inserted between the unwrapped DNA duplex and DNA gyre (Yan et al. 2019).Notably, the K131 methylation site lies two amino acid C-terminal to this helix, but only K130 (Fig. 1b), not K131 itself, is visible in the CCAN-Cenp-A Nuc structure.The alignment of Cse4 with H3 suggests that Cse4-131 is equivalent to K27 in H3.It should, however, be noted that H3 does not contain an equivalent α-helix, and this must therefore be interpreted with caution.Due to their proximity to the DNA at the entry/exit site of the DNA in the nucleosome, both R143 and K131 methylation can be hypothesized to affect the interaction of CENP-A with the DNA in a left-handed nucleosome.How a right-handed nucleosome configuration would be affected by these modifications remains to be seen (Furuyama and Henikoff 2009).

Cse4-R143A causes a synthetic growth defect with spc25-1 and dsn1-7
We next sought to determine how the absence of methylation on Cse4-R143 and -K131 affects Cse4 function at the centromere.For this, strains were created that carried alleles of CSE4 in which these residues were replaced by alanine (cse4-R143A, cse4-K131A).Neither mutation caused an appreciable growth defect or temperature sensitivity (Figs.2a, e, and 3c), showing that these modifications as such are not required for cell viability.
We subsequently tested whether cse4-R143A causes synthetic growth defects when combined with mutations in kinetochore components that reduce centromeric function (see below for cse4-K131A).Interestingly, cse4-R143A strongly enhanced the growth defect of cells carrying a mutation in the NDC80 component Spc25 (spc25-1, Fig. 2a) (Wigge and Kilmartin 2001).Specifically, the spc25-1 allele alone causes slightly reduced growth at 30°C and is unable to grow at 37°C, and the additional cse4-R143A mutation strongly enhanced the growth defect at 30° C.This result indicates that Cse4-R143me becomes important when Spc25 function is compromised.
The growth defect of spc25-1 cse4-R143A suggested that the chromosome segregation defect of spc25-1 is enhanced in these cells.To test this, we examined the stability of minichromosomes (plasmids) in cells that were wild type (wt) or carried one or both mutations.As expected, spc25-1, but not cse4-R143A alone, increased the plasmid loss rate, indicating a compromised centromere function of the plasmid.Importantly, cse4-R143A further exacerbated the plasmid maintenance defect of spc25-1, thus underscoring the notion that it enhanced the chromosome segregation defect (Fig. 2b).This indicated that Cse4-R143 methylation affects kinetochore function when Spc25 function is compromised.
We furthermore investigated how cse4-R143A affected cellcycle progression in spc25-1 cells.Incubation of spc25-1 cells at the semi-permissive temperature of 30°C for 3 hours had little effect on the proportions of cells with 1n or 2n DNA content, as determined by measuring the DNA content by FACS analysis (Fig. 2c).Importantly, spc25-1 cse4-R143A cells were arrested with a 2n DNA content under this condition, indicating that Cse4-R143A caused a defect in centromere function and chromosome segregation at the G2/M boundary.
The phenotypes observed above for cse4-R143A point towards a function for R143 in kinetochore stability.We were therefore interested to see whether other amino-acid substitutions can enhance the effect or can cause a phenotype of the Cse4-R143 substitution alone, without spc25-1.A replacement of R143 with the negatively charged glutamate (cse4-R143E) caused a similar phenotype as cse4-R143A in that the mutation alone had no effect on viability but enhanced the defect of spc25-1.cse4-R143Q, a replacement with a neutral charge, caused a slightly less pronounced defect in spc25-1 than the other two cse4 alleles (Fig. 2d).Thus, at least these, and perhaps all mutations in Cse4-R143, only cause a defect in the context of a compromised kinetochore, as is the case in the spc25-1 background.
The genetic interaction of cse4-R143A with spc25-1 was surprising, because Spc25 is a component of NDC80c (Wigge and Kilmartin 2001), which interacts on its chromatin-proximal side with the MIND complex and connects the inner kinetochore to the Dam ring and the microtubule.Spc25 thus is thought to be spatially distant from the Cse4-containing nucleosome (McAinsh and Marston 2022).A comprehensive genetic analysis of cse4-R143A with mutations in genes encoding kinetochore components revealed only one other mutation to have a genetic interaction with cse4-R143A, dsn1-7 (Nekrasov et al. 2003), in that the cse4-R143A dsn1-7 double mutants showed an enhanced temperature-sensitive growth defect (Fig. 2e).Dsn1 is a component of the MIND complex (De Wulf et al. 2003), which links NDC80c to CCAN in the inner kinetochore.Notably, cse4-R143A did not enhance growth defects of mutations in many other components of the inner and outer kinetochore (  Discussion).Cse4-R143A also did not display genetic interactions with other MIND, CCAN, or CBF3 complex components, or with Cbf1.This was in striking contrast to our earlier analysis of mutations in the Cse4 N-terminus, which showed highly selective genetic interactions with components of CCAN and with Cbf1, thus highlighting the functional difference of Cse4-R143 methylation with R37 methylation, K49 acetylation, and S33 phosphorylation (Samel et al. 2012;Hoffmann et al. 2018;Anedchenko et al. 2019).

Genetic screen for suppressors of the temperature sensitivity of spc25-1 cse4-R143A
The synthetic genetic interactions of cse4-R143A with spc25-1 and dsn1-7 were surprising given the assumed spatial distance between the respective proteins within the architecture of the kinetochore, raising the question of the mechanistic basis of the defect.To investigate this, we characterized intra-and extragenic suppressors of the temperature-sensitive growth defect of spc25-1 cse4-R143A.We surmised that such mutations would provide information about the defect of spc25-1, and thus by inference, also of cse4-R143A.The expectation was that some suppressors would specifically suppress spc25-1 cse4-R143A, whereas others would suppress spc25-1 alone.Operationally, spc25-1 cse4-R143A cells were UV-mutagenized, plated, and incubated at the semirestrictive temperature of 34°C for 4-5 days.This temperature was chosen in the hope of reducing the likelihood of isolating revertants of spc25-1, which carries a point mutation [CTG to CCG, L25P (Wigge and Kilmartin 2001)] and was therefore expected to easily revert back to wild type.FACS was conducted as in Fig. 2c.The experiment was carried out simultaneously with that presented in Fig. 2c.For further controls, see Supplementary Fig. 3a in Supplementary File 1. c) cse4-K131A partially suppressed the growth defect of spc25-1 cse4-R143A.Analysis was performed with plasmid-borne cse4 alleles as in Fig. 2d.
A total of 148 mutants were isolated that showed different degrees of suppression of the temperature-sensitive growth defect, ranging from poor growth at 34°C to moderate growth at 37°C.To distinguish between revertants and intragenic suppressors, the spc25 alleles of these strains were PCR-amplified and sequenced.Indeed, 29 of the 148 strains showed a reversion of the spc25-1 mutation (see Supplementary Fig. 2 in Supplementary File 1 for a flow chart of the analysis).Interestingly, 66 strains carried an intragenic suppressor mutation.They will be described below.The remaining 53 suppressors still carried the original spc25-1 allele (spc25-L25P).They were further investigated for whether they carried a mutation of the cse4-R143A allele, since reversion or other mutations in CSE4 in principle might also suppress the temperature sensitivity of the double mutant.Reversion is expected to be rare, because the mutation carries two base substitutions (CGA to GCA).The analysis showed that all 53 remaining suppressors still carried cse4-R143A.
Next, the 50 strongest suppressors were subjected to wholegenome sequencing (WGS) in order to identify the causative mutation.For economic reasons, they were grouped in 10 pools of five strains depending on their strength of growth at elevated temperatures, and the pools were subjected to WGS and SNP analysis.Here, we describe mutations in SET2, SPC24, and NDC80 as causative for the suppression, which accounts for 25 of the 50 mutants (Supplementary Fig. 2 in Supplementary File 1).The analysis of other mutants will be described elsewhere.

set2Δ suppresses the centromere defect of spc25-1 cse4-R143A
Interestingly, one of the spc25-1 cse4-R143A suppressor mutants was found to carry a mutation in SET2 (set2-P253S, CCA to TCA).Set2 is a histone lysine methyltransferase that associates with the elongating RNA polymerase II and methylates H3 K36 during transcription elongation (Krogan et al. 2003;Xiao et al. 2003).This serves to recruit the Rpd3S histone deacetylase complex, which removes acetylation marks on chromatin in the open reading frames (ORFs) after transcription and thus suppresses intragenic transcription emanating from cryptic transcription start sites within the ORF (Carrozza et al. 2005;Keogh et al. 2005).More recently, Set2 has been described to methylate H3 K37, and this protects the genome from spurious DNA replication (Santos-Rosa et al. 2021).
The function of Set2 as a chromatin modifier made it an interesting candidate to pursue.We tested whether the deletion of SET2 (set2Δ) caused the same phenotype as the point mutation isolated in the screen.Indeed, set2Δ suppressed the growth defect of spc25-1 cse4-R143A at semi-permissive temperatures, and the growth defect of spc25-1 alone was not suppressed (Fig. 3a).Furthermore, the arrest of spc25-1 cse4-R143A cells in G2/M at the semi-permissive temperature was suppressed by set2Δ (Fig. 3b, Supplementary Fig. 3a in Supplementary File 1).These results indicated that the absence of Set2 function alleviated the centromeric defect caused by Cse4-R143A in spc25-1 strains.
Since Set2 methylates H3 K36 (Strahl et al. 2002;Krogan et al. 2003;Li et al. 2003), a first hypothesis was that the suppression of spc25-1 cse4-R143A is the result of the absence of H3 K36me and concomitant changes in transcription, for instance of genes encoding kinetochore components.Of note, published transcriptome datasets did not reveal changes in the transcription of kinetochore components in mitotic cells in set2Δ (Venkatesh et al. 2016), though Set2 contributes to the repression of NDC80 transcription during meiosis (Chen et al. 2017;Chia et al. 2017).To further test a possible effect of H3 K36 methylation, we constructed spc25-1 cse4-R143A strains in which H3 K36 was exchanged for alanine (H3 K36A), in order to imitate the unmethylated state.We observed that H3 K36A improved the growth of both spc25-1 and spc25-1 cse4-R143A at the semi-permissive temperature (Supplementary Fig. 3b in Supplementary File 1).This effect therefore was distinct from that of set2Δ, which only suppressed spc25-1 cse4-R143A, arguing against H3 K36 methylation as the causative mechanism for the suppression.
Since we have identified Cse4-K131 to be methylated (Supplementary Fig. 1b in Supplementary File 1), we considered the possibility that Set2 could methylate K131 and that the absence of this methylation would specifically suppress the growth defect of spc25-1 cse4-R143A.To this end, we constructed an allele of CSE4 in which both R143 and K131 were replaced by alanine (cse4-R143A-K131A) and tested how this affected the temperature sensitivity of spc25-1.Significantly, spc25-1 cse4-R143A-K131A strains showed improved growth compared to spc25-1 cse4-R143A.Furthermore, cse4-K131A alone in spc25-1 did not improve growth (Fig. 3c).Thus, with respect to its suppression phenotypes, cse4-K131A resembles the effect of set2Δ.A speculative interpretation of this result therefore is that Set2 methylates Cse4-K131 and that this methylation negatively impacts centromeric function.Notably, this was the only phenotype observed for cse4-K131A, and this mutation showed no synthetic defects with several mutations in kinetochore components (Supplementary Table 3 in Supplementary File 1).Alternatively, based on the observation that Set2 regulates a non-coding transcript that controls NDC80 expression in meiosis (Chen et al. 2017;Chia et al. 2017), it may similarly affect mitotic centromere function by regulating an as yet unknown non-coding transcript.

Suppressor mutations of spc25-1 cse4-R143A are located in the tetramer junction and the Spc24-Ndc80 coiled coil of the NDC80 complex
In the screen for suppressors of the temperature sensitivity of spc25-1 cse4-R143A, we isolated 66 second-site mutations in Table 1.Synthetic genetic interactions of cse4-R143A with mutations in genes encoding kinetochore components a .

Allele
Synthetic phenotype with cse4-R143A Additional phenotype caused by cse4-R143A in combination with the indicated allele of the gene encoding the respective kinetochore component.-, no additional phenotype observed.
SPC25 representing 9 different alleles (Table 2).Furthermore, 24 isolates carried mutations in SPC24 (6 alleles) or NDC80 (3 alleles, Table 3; see Fig. 4a for temperature resistance of selected mutants).Like Spc25, Spc24 and Ndc80 are components of NDC80c (Fig. 4b) (Wigge and Kilmartin 2001;De Wulf et al. 2003), and it therefore seems most likely that these mutations are causative for the suppression of the spc25-1 cse4-R143A temperature sensitivity.To test this, the strains with the mutations spc24-D37A, spc24-S59L, and ndc80-L681W were backcrossed to a spc25-1 cse4-R143A parent strain, and five temperature-resistant and five temperature-sensitive segregants (as a control) were pooled and subjected to WGS.In all three cases, the temperatureresistant but not the temperature-sensitive segregants carried the respective spc24 or ndc80 mutation, indicating that they indeed were responsible for the suppression.Furthermore, backcrosses of ndc80-L681W, spc24-S45I, and spc24-S59L to a spc25-1 and a wt strain showed (1) that these mutations suppressed the temperature sensitivity of spc25-1 alone (without cse4-R143A) and ( 2) that the mutations themselves did not cause a growth defect.
The location of these mutations, as well as of the spc25 intragenic suppressors, in the respective proteins provides interesting insights in the context of the structure of NDC80c.NDC80c is a tetramer with two heterodimeric subcomplexes (Ndc80/Nuf2 and Spc24/Spc25) that connect end to end via a tetramerization domain (Fig. 4c) (Janke et al. 2001;Wigge and Kilmartin 2001;Cheeseman et al. 2006;DeLuca et al. 2006;Wei et al. 2007).The structure of a shortened "dwarf" version of yeast NDC80c shows two intertwined α-helical coiled coils with contacts between the heterodimers (Fig. 4d) (Valverde et al. 2016).The mutated residue in the spc25-1 allele, L25, is located in a bundle of three α-helices formed by Spc25, Spc24, and Ndc80 within the junction of NDC80c, and the mutation to proline can be hypothesized to disrupt the α-helix and thus to disturb the tetramerization of NDC80c.Interestingly, positions Spc24-D37, S45, and Ndc80-S667, which are highlighted by the respective suppressor alleles, are all located within the three-helix bundle structure formed by Spc24 and Spc25 with Ndc80 (Fig. 4d).This suggests that the mutations strengthen the interaction with the mutant Spc25-L25P protein and thus are able to suppress the temperature-sensitive growth defect.Furthermore, Spc25-Q28, which is highlighted by an intragenic suppressor, is also located in this region, suggesting that the mutation Q28L improves interaction to the closest protein, Ndc80.Also, the residues Spc24-S59 and Ndc80-L681 lie within the coiled-coil region formed by the two proteins, again indicating that the suppressor mutations stabilize the interaction between the two proteins and thus suppress the kinetochore defect of spc25-L25P.
It is noteworthy that among all extragenic suppressors isolated here, we did not find any mutations in NUF2, which encodes the fourth component of NDC80c.This may be due to the fact that spc25-1 does not directly affect the interaction to Nuf2 and such suppressor mutations in NUF2 hence may not exist.
Two residues highlighted by the suppressors, Spc25-G117 and -L123, are not present in the "dwarf" NDC80c structure (Valverde et al. 2016).Pairwise alignment shows that Spc25-G117 is equivalent to Q132 of human Spc25, and L123 lies in a yeastspecific short loop that is located between K133 and S134 of the human protein (Supplementary Fig. 4a in Supplementary File 1).These residues lie within the coiled-coil region of Spc25 and Spc24 in the Ndc80 ΔN-bonsai structure (Supplementary Fig. 4b in Supplementary File 1) (Ciferri et al. 2008).As above, the location of the respective suppressor mutations in the α-helical stalk formed by Spc25 and Spc24 suggests that they improve the binding between the two proteins and thus can compensate for the spc25-L25P mutation.Spc24-S134 is not present in either structure.
Our interpretation of the suppressor alleles that mutated Spc25-L25P to another amino acid is that these mutations (1) abrogate the helix-breaking proline and/or (2) cause an enhanced interaction with Spc24 and Ndc80.
Taken together, the analysis of the spc25-1 cse4-R143A suppressors suggests that they enhance kinetochore function by strengthening protein-protein interactions among the components of NDC80c.By inference, this suggests that the absence of Cse4-R143 methylation contributes to kinetochore failure, possibly by affecting the centromeric nucleosome, and that this can be compensated for by improving the stability of NDC80c.

Discussion
Post-translational modifications on histones are important regulators of chromatin function.Here, we describe two new modifications on the centromeric histone H3 variant CENP-A/Cse4 in S. cerevisiae, methylation of Cse4-R143 and methylation of Cse4-K131.Mutation of Cse4-R143 caused a synthetic growth defect and enhanced minichromosome loss and a chromosome segregation defect and a defect in G2/M phase transition when the function of Spc25, a component of NDC80c, was compromised.Furthermore, cse4-R143A caused a defect when combined with a mutation in DSN1, which encodes a component of the MIND complex (Euskirchen 2002;Nekrasov et al. 2003).We found several mutations in SPC24, NDC80, and SPC25 itself to suppress the defect of spc25-1 cse4-R143A, and their location within the structure of NDC80c suggests that they act by strengthening protein-protein interactions among these three components.This in turn indicates that the centromeric defect of spc25-1 cse4-R143A cells is the result of enhanced kinetochore failure when Cse4-R143A is mutated, possibly because it affects the centromeric nucleosome.Furthermore, the Set2 methyltransferase had a negative function Table 2. Intragenic suppressors of spc25-1.

Allele
Number of isolates spc25-L25P-T81I (Hs L106) at the centromere, because set2Δ partially suppressed the spc25-1 cse4-R143A defect, possibly via methylation of Cse4-K131.Of note, the defects caused by the absence of Cse4-R143 or -K131 methylation were markedly different from those of the PTMs in Cse4N, which inhibit the interaction between Cse4N and the Okp1 CENP-U and Ame1 CENP-Q components of the CCAN complex (Anedchenko et al. 2019).
It was surprising that cse4-R143A showed a specific defect with a mutation in SPC25, since Spc25 is located at the outer kinetochore and therefore is thought to be physically distant from the Cse4 nucleosome within the architecture of the kinetochore.From a formal genetic point of view, this genetic interaction could be interpreted as a physical contact between Spc25 and the Cse4 nucleosome.However, this seems unlikely, given our current knowledge of the overall structure of the kinetochore (McAinsh and Marston 2022).Rather, the suppressor mutations in SPC24, SPC25, and NDC80 indicate a role for Cse4-R143 in the stability of the centromeric nucleosome.Cse4-R143 lies in the α-N helix of Cse4, which is at the entry-exit site of the DNA that is partially unwrapped in the Cse4 nucleosome compared to a canonical nucleosome (Luger et al. 1997).Methylation at this site may enhance unwrapping, thus may destabilize the nucleosome and therefore is expected to counteract centromere function.Similarly, Cse4-K131 is located in the Cenp-A N helix of Cse4, which is wedged between the DNA strands, and methylation at this site may further pry the helices apart and affect nucleosome stability.One possibility is that these PTMs in the core region of the centromeric nucleosome help destabilize the nucleosome during DNA replication and help the replication fork move through the centromeric region.They thus may act in concert with R37 methylation and K49 acetylation in the N-terminus of Cse4, whose abundance increases in the S-phase (Anedchenko et al. 2019), concomitantly with the brief dissociation of the kinetochore from centromeric sequences during replication (Kitamura et al. 2007).
Of note, both arginine and lysine residues can carry more than one methyl group (symmetric and asymmetric arginine dimethylation and di-and trimethylation of lysine), but the mass spectrometric analysis of Cse4 showed only mono-methylation for R143 and K131.This, however, does not exclude that either residue also exists in higher methylation states in the cell.
We were furthermore surprised that cse4-R143A does not show genetic interactions with alleles in other genes encoding NDC80 components.One explanation for this is that the alleles ndc80-1, nuf2-61, and spc24-1 may be the result of a kinetochore defect that is mechanistically distinct from spc25-1, for instance, a destabilization of the interaction with the microtubule.This cannot be ascertained at this point, because they all carry multiple amino acid substitutions, unlike spc25-1, which carries a single exchange, L25P (Nekrasov et al. 2003).
It is also noteworthy that cse4-R143A caused a defect with dsn1-7, because Dsn1 is a direct interaction partner of NDC80c (Hornung et al. 2014;Kudalkar et al. 2015).A molecular interpretation of this is that the mutation in DSN1 destabilizes its interaction with the chromatin-proximal side of NDC80c, which causes an enhanced defect upon concomitant destabilization of the centromeric nucleosome by Cse4-R143 mutation.
The observation that R143 and K131 of Cse4 are methylated raises the question which methyltransferases are responsible for the modifications.While this remains to be seen for R143, our genetic analysis of K131 suggests that it is methylated by Set2, which is known for methylation of H3 K36 and K37 (Strahl et al. 2002;Santos-Rosa et al. 2021).K131 lies within a sequence of Cse4 that shows limited sequence similarity to residues in the N-terminus of H3 surrounding K27 (Fig. 1b).However, H3 K27, a prominent methylation site in higher eukaryotes (Cao et al. 2002), is not known to be methylated in S. cerevisiae.We speculate that Cse4-K131 represents a non-canonical target of Set2.Notably, the related methyltransferase Set1, whose best-known target residue is H3 K4 (Briggs et al. 2001;Roguev et al. 2001), also has a function at the kinetochore in that it methylates residues within the Dam1 kinetochore protein, which is part of the Dam1 ring complex that multimerizes around and slides along the microtubules during chromosome segregation (Westermann et al. 2006), and this Dam1 methylation counteracts phosphorylation by the Ipl1/Aurora kinase (Zhang et al. 2005).
Our discovery of two new PTMs in the core region of the centromeric nucleosome in yeast raises the question whether these modifications are conserved in the CENP-A homologs in higher eukaryotes.So far, several phosphorylation sites, as well as acetylation and ubiquitination, have been described on CENP-A (Fukagawa 2017).Importantly, Cse4-R143, but not Cse4-K131, is conserved in CENP-A (as well as in H3, Supplementary Fig. 5 in Supplementary File 1), and it thus is possible that this residue is also methylated in CENP-A.

Fig. 2 .
Fig. 2.The absence of Cse4-R143 methylation enhances the centromeric defect of spc25-1 and dsn1-7.a) Mutation of Cse4-R143 to alanine caused slow growth in the spc25-1 background.Strains with the indicated genotypes were serially diluted and spotted on YPD medium, and plates were incubated for 3 days and the indicated temperatures.b) cse4-R143A enhanced the plasmid maintenance defect of spc25-1.Error bars give SD of three independent experiments.*Significant difference, P < 0.05.n.s., not significant.c) spc25-1 cse4-R143A cells arrested at the G2/M phase of the cell cycle at the restrictive temperature.Cells were grown to early logarithmic phase at 23°C and shifted to 30°C for 3 h.DNA content as measured by FACS analysis is shown.d) cse4-R143E and -R143Q enhanced the growth defect of spc25-1.Serial dilutions of cse4Δ strains with the cse4 alleles on a plasmid were spotted on YPD plates and incubated for 2 days at the indicated temperatures.e) cse4-R143A enhanced the growth defect of dsn1-7.Serial dilutions of the indicated strains were spotted on YPD plates and growth for 2 days at the indicated temperatures.

Fig. 4 .
Fig. 4. Suppressor mutations of spc25-1 cse4-R143A are located in the tetramer junction and the Spc24-Ndc80 coiled coil of the NDC80 complex.a) Suppression of the growth defect of spc25-1 cse4-R143A by mutations in NDC80 and SPC24.Serial dilutions of the strains were spotted on full medium and grown for 3 days at the indicated temperatures.b) Schematic of the S. cerevisiae kinetochore.c) Schematic of NDC80c illustrating the subcomplexes of Nuf2 with Ndc80 and Spc24 with Spc25.d) Localization of residues mutated in second-site suppressors of the spc25-1 cse4-R143A temperature-sensitive growth defect in the structure of the NDC80c e-dwarf structure (PDB 5TD8, (Valverde et al. 2016)).Color schematic of proteins as in b).The respective residues are shown as yellow spheres.The suppressor mutations localize to the tetramerization junction and the Spc24-Spc24 stalk.Image generated using PyMol.

Table 1 )
. Perhaps most surprising was the absence of phenotypes with mutations in other NDC80 components, notably Spc24, the most intimate physical interactor of Spc25 (Janke et al.