Structural Basis for Mis18 Complex Assembly: Implications for Centromere Maintenance

The centromere, defined by the enrichment of CENP-A (a Histone H3 variant) containing nucleosomes, is a specialised chromosomal locus that acts as a microtubule attachment site. To preserve centromere identity, CENP-A levels must be maintained through active CENP-A loading during the cell cycle. A central player mediating this process is the Mis18 complex (Mis18α, Mis18ý and Mis18BP1), which recruits the CENP-A specific chaperone HJURP to centromeres for CENP-A deposition. Here, using a multi-pronged approach, we characterise the structure of the Mis18 complex and show that multiple hetero- and homo-oligomeric interfaces facilitate the hetero-octameric Mis18 complex assembly composed of 4 Mis18α, 2 Mis18ý and 2 Mis18BP1. Evaluation of structure-guided/separation-of-function mutants reveals structural determinants essential for Mis18 complex assembly and centromere maintenance. Our results provide new mechanistic insights on centromere maintenance, highlighting that while Mis18α can associate with centromeres and deposit CENP-A independently of Mis18ý, the latter is indispensable for the optimal level of CENP-A loading required for preserving the centromere identity.


Introduction
Faithful chromosome segregation during cell division requires bi-orientation of chromosomes on the mitotic spindle through the physical attachment of kinetochores to microtubules.
Kinetochores are large multiprotein scaffolds that assemble on a special region of chromosomes known as the centromere (Musacchio and Desai, 2017, Cheeseman, 2014, Catania and Allshire, 2014, Fukagawa and Earnshaw, 2014).Whilst centromeres in some organisms, such as budding yeast, are defined by a specific DNA sequence, in most eukaryotes, centromeres are distinguished by an increased concentration of nucleosomes containing a histone H3 variant called CENP-A (Fukagawa and Earnshaw, 2014, McKinley and Cheeseman, 2016, Stellfox et al., 2013, Black et al., 2010).CENP-A containing nucleosomes recruit CENP-C and CENP-N, two proteins that are part of the constitutive centromere-associated network (CCAN) and that recruits the rest of the kinetochore components at the centromeric region of the chromosome (Carroll et al., 2010, Kato et al., 2013, Weir et al., 2016).
Whilst canonical histone loading is coupled with DNA replication, CENP-A loading is not (Dunleavy et al., 2011).This results in a situation where, after S-phase, the level of CENP-A nucleosomes at the centromere is halved due to the distribution of existing CENP-A to the duplicated DNA (Jansen et al., 2007, Dunleavy et al., 2009).To maintain centromere identity, centromeric CENP-A levels must be restored.This is achieved through active CENP-A loading at centromeres (during G1 in humans) via a pathway that requires the Mis18 complex (consisting of Mis18a, Mis18b and Mis18BP1) and the CENP-A chaperone, HJURP (Jansen et al., 2007, Fujita et al., 2007, Dunleavy et al., 2009, Foltz et al., 2009, Barnhart et al., 2011).
The Mis18 complex can recognise and localise to the centromere, possibly through its proposed binding to CENP-C and/or other mechanisms which have not yet been identified (Dambacher et al., 2012, Stellfox et al., 2016, Moree et al., 2011).Once at the centromere, the Mis18 complex has been implicated in facilitating the deposition of CENP-A in several ways.There is evidence that the Mis18 complex affects DNA methylation and histone acetylation, which may facilitate CENP-A loading (Hayashi et al., 2004, Kim et al., 2012).But one of its most important and well-established roles of the Mis18 complex is the recruitment of HJURP, which binds a single CENP-A/H4 dimer and brings it to the centromere (Hu et al., 2011, Dunleavy et al., 2009, Barnhart et al., 2011).This then triggers a poorly understood process in which the H3 nucleosomes are removed and replaced with CENP-A nucleosomes.
The timing of CENP-A deposition is tightly regulated, both negatively and positively, by the kinases Cdk1 and Plk1, respectively, in a cell cycle-dependent manner (Silva et al., 2012, Spiller et al., 2017, Pan et al., 2017, McKinley and Cheeseman, 2014, Stankovic et al., 2017, Muller et al., 2014).Previous studies demonstrated that Cdk1 phosphorylation of Mis18BP1 prevents the Mis18 complex assembly and localisation to centromeres until the end of mitosis (when Cdk1 levels are reduced) (Spiller et al., 2017, Pan et al., 2017).Cdk1 also phosphorylates HJURP, which negatively regulates its binding to the Mis18 complex at the centromere (Muller et al., 2014, Stankovic et al., 2017, Wang et al., 2014).In cells, Plk1 is a positive regulator, and its activity is required for G1 centromere localisation of the Mis18 complex and HJURP.Plk1 has been shown to not only phosphorylate Mis18a/b and Mis18BP1, but it has also been proposed to interact with phosphorylated Mis18 complex through its polo-box domain (PBD) (McKinley and Cheeseman, 2014).
As outlined above, a central event in the process of CENP-A deposition at centromeres is the Mis18 complex assembly.The Mis18 proteins, Mis18a and Mis18b, possess a well-conserved globular domain called the Yippee domain (also known as the MeDiY domain; spanning residues 77-180 in Mis18a and 73-176 in Mis18b) and C-terminal a-helices .We and others previously showed that the Yippee domains of Mis18 proteins can form a hetero-dimer, while the C-terminal helices form a hetero-trimer with two Mis18a and one Mis18b.However, the full-length proteins form a hetero-hexameric assembly with 4 Mis18a and 2 Mis18b.This led to a proposed model, where the Mis18a and Mis18b mainly interact via the C-terminal helices to form a hetero-trimer, and two such heterotrimers interact via the Yippee hetero-dimerisation (Mis18a/Mis18b) or/and homo-dimerisation (Mis18a/Mis18a) to form a hetero-hexameric assembly (Nardi et al., 2016, Spiller et al., 2017, Pan et al., 2017, Pan et al., 2019).
Mis18BP1, the largest subunit of the Mis18 complex (1132 aa residues), is a multi-domain protein containing SANTA (residues 383-469) and SANT (residues 875-930) domains, which are known to have roles in regulating chromatin remodelling (Zhang et al., 2006, Aasland et al., 1996, Maddox et al., 2007).In-between these two domains resides the CENP-C binding domain (CBD) (Dambacher et al., 2012, Stellfox et al., 2016).In vivo, the CBD alone is not sufficient to recruit Mis18BP1 to the centromere and requires the N-terminus of the protein for proper localisation (Stellfox et al., 2016).We and others have previously shown that the Nterminal 130 amino acids of Mis18BP1 are sufficient for interaction with Mis18a/b through their Yippee domains, and Cdk1 phosphorylation of Mis18BP1 at residues T40 and S110 inhibits its interaction with Mis18a/b to form an octamer complex consisting of 2 Mis18BP1, 4 Mis18a and 2 Mis18b (Spiller et al., 2017, Pan et al., 2017).
Although the importance of the Mis18 complex assembly and function is well appreciated, structural understanding of the intermolecular interfaces responsible for the Mis18 complex assembly and their functions are yet to be identified.Here, we have characterised the structural basis of the Mis18 complex assembly using an integrative structure modelling approach that combines X-ray crystallography, Electron Microscopy (EM), Small Angle X-ray Scattering (SAXS), Cross-Linking Mass Spectrometry (CLMS), AlphaFold and computational modelling.By evaluating the structure-guided mutations in vitro and in vivo, we provide important insights into the key structural elements responsible for Mis18 complex assembly and centromere maintenance.

Structural basis for the assembly of Mis18a/b core modules
Mis18a and Mis18b possess two distinct but conserved structural entities, a Yippee domain and a C-terminal a-helix (Fig. 1a and S1a & b).Mis18a possesses an additional a-helical domain upstream of the Yippee domain (residues 39-76).Previous studies have shown that Mis18a Yippee domain can form a homo-dimer or a hetero-dimer with Mis18b Yippee domain whereas Mis18a/b C-terminal helices form a robust 2:1 heterotrimer (Subramanian et al., 2016, Spiller et al., 2017, Pan et al., 2017).Disrupting Yippee homo-or hetero-dimerisation in full-length proteins, while did not abolish their ability to form a complex, did perturb the dimerisation of Mis18a/b heterotrimer (Spiller et al., 2017).Contrarily, intermolecular interactions involving the C-terminal helices of Mis18a and Mis18b are essential for Mis18a/b complex assembly (Nardi et al., 2016).Overall, the available biochemical data suggest the presence of at least three independent structural core modules within the Mis18a/b complex: the Mis18a Yippee homo-dimer, the Mis18a/b Yippee hetero-dimer and the Mis18a/b Cterminal helical assembly.Here, we structurally characterised these modules individually and together as a holo-complex.
Mis18a Yippee homo-dimer: We previously determined a crystal structure of the Yippee domain in the only homologue of Mis18 in S. pombe (PDB: 5HJ0), showing that it forms a homo-dimer (Subramanian et al., 2016).To determine the structure of human Mis18 Yippee domains, we purified and crystalised Mis18aYippee (residues 77-190).The crystals diffracted Xrays to about 3 Å resolution, and the structure was determined using the molecular replacement method.The final model was refined to R and Rfree factors of 20.26% and 25.00%, respectively (Table S1 and Fig. 1b, PDB ID: 7SFZ).The overall fold of the Mis18aYippee is remarkably similar to the previously determined S. pombe Mis18Yippee homo-dimer structure (Subramanian et al., 2016).In brief, the monomeric Mis18aYippee is formed by two antiparallel b-sheets that are held together by a Zn 2+ ion coordinated via loops containing C-X-X-C motifs.
The Mis18aYippee dimerisation is mediated via a back-to-back arrangement of a 'threestranded' b-sheet from each monomer.
Mis18a/b C-terminal helical assembly: Previous studies have shown that recombinantly purified C-terminal α-helices of Mis18a and Mis18b form a hetero-trimer with 2 copies of Mis18a and 1 copy of Mis18b (Spiller et al., 2017, Pan et al., 2017).However, in the absence of high-resolution structural information, how Mis18 C-terminal helices interact to form a hetero-trimer and how the structural arrangements of a-helices influence the relative orientations of the Yippee domains, and hence the overall architecture of the Mis18a/b hexamer assembly, remained unclear.We purified Mis18a spanning aa residues 191 to 233 and Mis18b spanning aa residues 188 and 229 (Fig. 1a, S1a & b) and crystallised the reconstituted complex.The crystals diffracted X-rays to about 2.5 Å resolution.The structure was determined using single wavelength anomalous dispersion method.After iterative cycles of refinement and model building, the final model was refined to R and Rfree factors of 24.77% and 27.96%, respectively (Table S1, PDB ID: 7SFY).The asymmetric unit contained two copies of Mis18a/b hetero-trimer.The final model included Mis18a residues 191 to 231 in one copy, Mis18a residues 193 to 230 in the second copy, and Mis18b residues 190 to 223 (Fig. 1d).The two Mis18a helices interact in an antiparallel orientation, and one helix is stabilised in a slightly curved conformation.This arrangement results in a predominantly negatively charged groove that runs diagonally on the surface formed by the Mis18a helices (Fig. 1d &   e).This observation is consistent with the theoretically calculated pI of the Mis18a helix (pI=4.9).In contrast, the pI of the Mis18b helix is 8.32.This charge complementarity appears to facilitate the interaction with Mis18a, as a positively charged surface of the Mis18b helix snugly fits in the negatively charged groove of the Mis18a/a interface.A closer look at the intermolecular interactions reveals tight hydrophobic interactions along the 'spine' of the binding groove with electrostatic interactions 'zipping-up' both sides of the Mis18b helix (Fig 1e).The binding free energy calculated based on the buried accessible surface area suggests a nanomolar affinity interaction between the helices of Mis18a and Mis18b.It should be noted that the crystal structure presented here differs from the previously predicted models in terms of either the subunit stoichiometry (Pan et al., 2019) or the directional arrangement of individual subunits (Mis18a and Mis18b in parallel orientation with the 2 nd Mis18a in an antiparallel orientation (this work) vs all parallel (Pan et al., 2019)).

Multiple surfaces of Mis18a/b Yippee hetero-dimers contribute to the overall oligomeric assembly of the Mis18 complex.
Full-length Mis18a/b complex or the Mis18core complex (Mis18a -Mis18b -Mis18BP120-130) were not amenable for structural characterisation using X-ray crystallography possibly due to their intrinsic flexibility.Consistent with this notion, the SAXS profiles collected for the Mis18a/b DN (Mis18a residues 77-187 and Mis18b residues 56-183), Mis18a/b and Mis18core complexes suggest that these complexes possess an elongated shape with flexible features (Fig. S2, Table S2).Hence, to understand the overall assembly of the Mis18 complex we took an integrative structure modelling approach, combining the crystal structures of Mis18aYippee dimer and Mis18a/Mis18b C-terminal hetero-trimeric helical assembly together with the homology/AlphaFold modelling of Mis18aYippee/Mis18bYippee hetero-dimer, negative staining EM, SAXS and CLMS analysis of the Mis18core complex.
The negative staining electron micrographs of the Mis18core complex cross-linked using GraFix (Kastner et al., 2008) revealed a good distribution of particles (Fig. S3a).Particle picking, followed by a few rounds of 2D classifications revealed classes with defined structural features (Fig. S3b).Some of the 2D projections resembled the shape of a 'handset' of a telephone with bulkier 'ear' and 'mouth' pieces.Differences in the relative orientation of bulkier features of the 2D projection suggested conformational heterogeneity.The three-dimensional volumes calculated for the particles were similar (approximately 220 x 105 x 80 Å) and in agreement with the Dmax calculated from SAXS analysis (Fig. S2d).
We attempted to assemble the whole Mis18 complex using AlphaFold-multimer (AFM), with full length Mis18a (in purple), Mis18b (in pink) and two small region of Mis18BP1 (20-51 and 109-130; in salmon) (Evans et al., 2021).The AFM converged towards a structure with six Yippee domains stacked in a line-like arrangement in the Mis18aYippee-Mis18bYippee-Mis18aYippee-Mis18aYippee-Mis18bYippee-Mis18aYippee order and two triple helix bundles, each formed by C-terminal a-helices of 2 copies of Mis18b and 1 copy of Mis18b.However, the modelled two helical bundles had all three helices in a parallel orientation that is not supported by our crystal structures (Fig. 1d) and crosslinks (Fig. S2e).We modified the relative orientation of the helices to match the crystal structure by superposing the latter on the AFM model (Fig. 1f, 1g & S3d).Using crosslinks and docking we have added the N-terminal helices of the Mis18a.Cross-linking data indicates that these helices have multiple orientations with respect to the rest of the structure, contacting both Yippee domains and triple helix bundles.
The linker between the Yippee domain and the C-terminal helix is the shortest in Mis18b (Fig. 1a), further supporting the arrangement of the Yippee domains within the assembly.The integrative model of the Mis18 complex fits well in the EM map.Interestingly, the serial arrangement of the Yippee domains utilises the second Yippee dimerisation interface observed in the crystal packing of both human Mis18a Yippee and S. pombe Mis18 Yippee (Fig. S3d, highlighted by zoom in view).Accordingly, disrupting this interface by mutating Mis18a residues C154 and D160 (Fig. S3d) perturbed Mis18 oligomerisation as evidenced by SEC analysis (Fig. S3e).

Mis18a/b centromere localisation and new CENP-A loading.
Although the subunit stoichiometry and the arrangement of Mis18a/b C-terminal helices within the helical bundle proposed by Nardi et al. 2016 are different from the data presented here, the Mis18a residues (I201, L205, L212, L215 and L219) that were predicted by them to stabilise the helical bundle do indeed form the 'spine' of the hydrophobic core running along the triple helical bundle (Fig. 1d and e).Mutating these residues perturbed the ability of Mis18a tethered at an ectopic LacO site to facilitate CENP-A deposition at the tethering site (Nardi et al., 2016).However, how these Mis18a mutants perturb the oligomeric structure of the Mis18a/b C-terminal helical bundle and how this structural perturbation affects CENP-A loading at endogenous centromeres remain as open questions.
To address these questions, we first tested these mutants using in vitro amylose pull-down assays by mixing recombinantly purified WT and mutant His-MBP-Mis18b188-229 and His-SUMO-Mis18a191-233 proteins.Mutating these residues to Ala (Mis18aI201A/L205A and Mis18aL212A/L215A/L219A) or Asp (Mis18aI201D/L205D) abolished the ability of Mis18a a-helix to interact with Mis18b188-229 (Fig. S4a).SEC MALS analysis of His-SUMO tagged Mis18a188-233 showed that on its own, Mis18a WT protein can form a dimer, whilst introducing I201A/L205A or L212A/L215A/L219A results in both proteins forming a monomer (Fig. S4c).Co-immunoprecipitation (Co-IP) assays using an anti-Mis18a antibody were performed on cells where endogenous Mis18a was depleted, and Mis18a-mCherry was co-expressed with Mis18b-GFP to check for complex formation (Fig. S4b).In line with our in vitro pull-downs, the co-IPs using a Mis18a antibody revealed that Mis18aWT-mCherry interacted with Mis18b-GFP while Mis18aI201A/L205A and Mis18aL212A/L215A/L219A mutants did not (Fig. S4b).To evaluate the role of this interaction on centromere localisation of Mis18a and Mis18b and CENP-A deposition, these mutants were further tested in HeLa cells.
We then evaluated the impact of Mis18a mutants not capable of forming the C-terminal helical bundle on new CENP-A deposition.We did this by performing a Quench-Chase-Pulse CENP-A-SNAP Assay according to Jansen et al. (Jansen et al., 2007) (Fig. 2c).HeLa CENP-A-SNAP cells were depleted of endogenous Mis18a and rescued with either Mis18aWT or Mis18a mutants (Mis18aI20A1/L205A, Mis18aI201D/L205D and Mis18aL212A/L215A/L219A).The existing CENP-A was blocked with a non-fluorescent substrate of the SNAP, and the new CENP-A deposition in the early G1 phase was visualised by staining with the fluorescent substrate of the SNAP.Mis18aWT rescued new CENP-A deposition to levels compared to that of control siRNA (Fig. 2b and c).However, Mis18aI20A1/L205A, Mis18aI201D/L205D and Mis18aL212A/L215A/L219A abolished new CENP-A loading almost completely, indicating that the formation of the Mis18 triple helical bundle is essential for CENP-A deposition (Fig. 2c).

CENP-A, but efficient CENP-A loading requires Mis18b.
We again performed amylose in vitro pull-down assays, using His-SUMO-Mis18a191-233 WT and mutant His-MBP-Mis18b188-229 proteins, to assess the ability of Mis18b mutant to form a triplehelical bundle with Mis18a.Based on our X-ray crystal structure (Fig. 1d), we identified one cluster (L199/I203) in Mis18b and observed that mutating these residues to either Ala (Mis18bL199A/I203A) or Asp (Mis18bL199D/I203D) either reduced or abolished its ability to interact with Mis18a191-233 (Fig. 3a).Co-IP analysis using an anti-Mis18a antibody was performed on cells where endogenous Mis18b was depleted, and Mis18b-GFP was expressed along Mis18a-mCherry to check for complex formation.Western blot analysis showed that Mis18bWT could interact with Mis18a-mCherry and that the ability of Mis18bL199D/I203D to interact with Mis18a was reduced (Fig. 3a Interestingly, unlike the Mis18a mutants (Mis18aI20A1/L205A, Mis18aI201D/L205D and Mis18aL212A/L215A/L219A), Mis18bL199D/I203D did not abolish new CENP-A loading but reduced the levels only moderately.
Together, these analyses demonstrate that Mis18a can associate with centromeres and deposit new CENP-A independently of Mis18b.However, efficient CENP-A loading requires Mis18b.

Structural basis for centromere recruitment of Mis18a/b by Mis18BP1
Previous studies have shown that Mis18BP1 N-terminus (1-130 aa) is required to bind Mis18a/b (Spiller et al., 2017).However, how Mis18a/b Yippee domains recognise Mis18BP1 is not clear.Our structural analysis suggests that two Mis18BP1 fragments, a short helical segment spanning aa residues 110-130 (Mis18BP1110-130) and a region spanning aa residues 24-50 (Mis18BP124-50) interact with Mis18a Yippee domain and with an interface formed between Mis18a/b Yippee hetero-dimers, respectively (Fig. 4a).Mis18BP1110-130 binds at a hydrophobic pocket of the Mis18a Yippee domain formed by amino acids L83, F85, W100, I110, V172 and I175.This hydrophobic pocket is surrounded by hydrophilic amino acids E103, D104, T105, S169 E171 facilitating additional electrostatic interactions with Mis18BP1110-130 (Fig. 4a).Mis18BP124-50 contains two short b strands that interact at Mis18a/b Yippee interface extending the six-stranded-b sheets of both Mis18a and Mis18b Yippee domains.Notably, the two Cdk1 phosphorylation sites on Mis18BP1 (T40 and S110) that we and others have shown to disrupt Mis18 complex assembly (Spiller et al., 2017, Pan et al., 2017) lie directly within the Mis18 a/b binding interface predicted by this model, providing the structural basis for Cdk1 mediated regulation of Mis18 complex assembly.Consistent with this model, several crosslinks observed between Mis18BP1 and Mis18a and Mis18b map to these residues.Mutating the negatively charged amino acid cluster of Mis18a (E103, D104 and T105) that is juxtaposed to Mis18BP1110-130 in a TetR-eYFP-Mis18a vector (TetR-eYFP-Mis18aE103R/D104R/T105R) transfected in HeLa cells with an ectopic synthetic alphoid tetO array integrated in a chromosome arm significantly perturbed Mis18a's ability to recruit Mis18BP120-130-mCherry to the tethering site as compared to Mis18aWT (Fig. 4b).
Furthermore, we probed the effects of perturbing Mis18a-Mis18BP1 interaction on endogenous centromeres.We depleted Mis18a in a cell line that stably expresses CENP-A-SNAP and allows inducible expression of GFP-Mis18BP1 (McKinley and Cheeseman, 2014).
We then assessed the ability of transfected Mis18a-mCherry to co-localise with Mis18BP1 at centromeres.Depletion of Mis18a and simultaneous expression of either Mis18aWT-mCherry or Mis18aE103R/D104R/T105A-mCherry revealed that, unlike Mis18aWT, Mis18aE103R/D104R/T105A failed to localise at endogenous centromeres (Fig. 4c, middle panel).We also observed a slight decrease in the levels of GFP-Mis18BP1 at the centromere when Mis18aE103R/D104R/T105A was expressed as compared to Mis18aWT (Fig. 4c, right panel).Consistent with the observation of reduced centromeric Mis18a, when Mis18aE103R/D104R/T105A-mCherry is expressed, the quantification of new CENP-A deposition in HeLa cell expressing CENP-A-SNAP showed a significant reduction of new CENP-A deposition at the centromere indicating that the interaction of Mis18a with Mis18BP1 is essential for centromeric recruitment of the Mis18 complex and for CENP-A loading (Fig. 4d).

Discussion
Mis18 complex assembly is a central process essential for the recruitment of CENP-A/H4 bound HJURP and the subsequent CENP-A deposition at centromeres (Jansen et al., 2007, Fujita et al., 2007, Dunleavy et al., 2009).Thus far, several studies, predominantly biochemical and cellular, have characterised interactions and functions mediated by the two distinct structural domains of the Mis18 proteins, the Yippee and C-terminal a-helical domains of Mis18a and Mis18b (Spiller et al., 2017, Pan et al., 2017, Nardi et al., 2016, Stellfox et al., 2016).Some of the key conclusions of these studies include: (1) Mis18a/b is a hetero-hexamer made of 4 Mis18a and 2 Mis18b; (2) The Yippee domains and C-terminal a-helices of Mis18a and Mis18b have the intrinsic ability to homo-or hetero-oligomerise, and form three distinct oligomeric modules in different copy numbers -a Mis18aYippee homo-dimer, two copies of Mis18a/bYippee hetero-dimers and two hetero-trimers made of Mis18a/b C-terminal helices ( 2Mis18a and 1 Mis18b); (3) the two copies of Mis18a/bYippee hetero-dimers each bind one Mis18BP120-130 and form a hetero-octameric Mis18core complex (Mis18a/Mis18b/Mis18BP120-130: a Mis18a/b hetero-hexamer bound to 2 copies of Mis18BP120-130).However, no experimentally determined structural information is available for the human Mis18 complex.This is crucial to identify the amino acid residues essential for the assembly of Mis18a/b and the holo-Mis18 complexes and to determine the specific interactions that are essential for the localisation of Mis18 complex to centromeres and its function.
Here, we have taken an integrative structural approach that combines X-ray crystallography, electron microscopy and homology modelling with cross-linking mass spectrometry to characterise the structure of the Mis18 complex.Our analysis shows that Mis18a/b heterotrimer is stabilised by the formation of a triple helical bundle with a Mis18a/bYippee hetero-dimer on one end and Mis18aYippee monomer on the other.Two such Mis18a/b hetero-trimers assemble as a hetero-hexamer via the homo-dimerisation of the Mis18aYippee domains.The crystal structure of Mis18a/bC-term triple helical structure allowed us to design several separation of function Mis18a and Mis18b mutants.These mutations specifically perturb the ability of Mis18a or Mis18b to assemble into the helical bundle, while retaining their other functions, if there are any.Functional evaluation of these mutants in cells has provided important new insights into the molecular interdependencies of the Mis18 complex subunits.
Particularly, the observations that: (1) Mis18a can associate with centromeres and deposit CENP-A independently of Mis18b, and (2) depletion of Mis18b or disrupting the incorporation of Mis18b into the Mis18 complex, while does not abolish CENP-A loading, reduces the CENP-A deposition amounts, questions the consensus view that Mis18a and Mis18b always function as a single structural entity to exert their function to maintain centromere maintenance.
Whilst proteins involved in CENP-A loading have been well established, the mechanism by which the correct levels of CENP-A are controlled is yet to be thoroughly explored and characterised.The data presented here suggest that Mis18b mainly contributes to the quantitative control of centromere maintenance -by ensuring the right amounts of CENP-A deposition at centromeres -and maybe one of several proteins that control CENP-A levels.
Future studies will focus on dissecting the mechanisms underlying the Mis18b-mediated control of CENP-A loading amounts along with any other mechanisms involved.
Previous studies using siRNA to deplete Mis18a shows that is does not effect Mis18BP1 localisation and that Mis18BP1 can associate with centromeres independently of Mis18a (McKinley and Cheeseman, 2014).The separation of function Mis18a mutant unable to bind Mis18PB1, characterised here, shows that disrupting Mis18a-Mis18BP1 interaction completely abolishes Mis18a's ability to associate with centromeres and new CENP-A loading.This highlights that Mis18BP1-mediated centromere targeting is the major centromere recruitment pathway for the Mis18a/b complex.
Previously published work identified amino acid sequence similarity between the N-terminal region of Mis18a and R1 and R2 repeats of the HJURP that mediates Mis18a/b interaction (Pan et al., 2019).Deletion of the Mis18a N-terminal region enhanced HJURP interaction with the Mis18 complex.This led to speculation that the N-terminal region of Mis18a might directly interact with the HJURP binding site of the Mis18 complex and thereby modulating HJURP binding.Our work presented here strengthens this speculation and provides the structural justification.We show that the N-terminal helical region of Mis18a makes extensive contacts with the C-terminal helices of Mis18a and Mis18b that mediate HJURP binding.In the future, it will be important to address how and when the interference caused by the N-terminal region of Mis18a is relieved for efficient HJURP binding by the Mis18 complex.

Expression and purification of recombinant proteins
For crystallisation, both Mis18a/bC-term domains and Mis18aYippee were transformed and expressed in Escherichia coli BL21 (DE3) using the auto-inducible expression system (Studier, 2005).The cells were harvested and resuspended in the lysis buffer containing 30 mM Tris-HCl pH7.5, 500 mM NaCl, and 5 mM b-mercaptoethanol with protease inhibitor cocktails.The resuspended cells were lysed using the ultra-sonication method and centrifuged at 20,000 x g for 50 min at 4°C to remove the cell debris.After 0.45 µm filtration of the supernatant, the lysate was loaded into the cobalt affinity column (New England Biolabs) and eluted with a buffer containing 30 mM Tris-HCl pH7.5, 500 mM NaCl, 5 mM bmercaptoethanol, and 300 mM imidazole.The eluate was loaded into the amylose affinity column (New England Biolabs) and washed with a buffer containing 30 mM Tris-HCl pH7.5, 500 mM NaCl, and 5 mM b-mercaptoethanol.To cleave the His-MBP tag, on-column cleavage was performed by adding Tobacco Etch Virus (TEV) protease (1:100 ratio) into the resuspended amylose resin and incubated overnight at 4°C.The TEV cleavage released the untagged Mis18a/bC-term domains in solution, and the flow through fraction was collected and concentrated using a Centricon (Millipore).The protein was loaded onto a HiLoad™ 16/600 Superdex™ 200 column (GE Healthcare) equilibrated with a buffer containing 30 mM Tris-HCl pH7.5, 100 mM NaCl, and 1 mM TCEP.To further remove the contaminated MBP tag, the sample was re-applied into the amylose affinity column, and the flow-through fraction was collected and concentrated to 20 mg/ml for the crystallisation trial.SeMet (selenomethionine) incorporated Mis18a/bC-term domains were expressed with PASM-5052 auto-inducible media (Studier, 2005).The SeMet-substituted Mis18a/bC-term domains were purified using the same procedure described above.
The purification of His tagged Mis18aYippee employed the same purification method used for Mis18a/bC-term domains except for the amylose affinity chromatography step.The purified Mis18aYippee from the HiLoad™ 16/600 Superdex™ 200 chromatography was concentrated to 13.7 mg/ml with the buffer containing 30 mM Tris-HCl pH7.5, 100 mM NaCl, and 1 mM TCEP.
All other proteins were expressed in Escherichia coli BL21 (DE3) Gold cells using LB.After reaching an O.D. ~ 0.6 at 37 o C, cultures were cooled to 18 o C and induced with 0.35 mM IPTG overnight.The His-Mis18a/His-GFP-Mis18b complex was purified by resuspending the pellet in a lysis buffer containing 20 mM Tris-HCl pH 8.0 at 4 o C, 250 mM NaCl, 35 mM imidazole pH 8.0 and 2 mM b-mercaptoethanol supplemented with 10 µg/ml DNase, 1mM PMSF and cOmplete™ EDTA-free (Sigma).After sonication, clarified lysates were applied to a 5 ml HisTrap™ HP column (GE Healthcare) and washed with lysis buffer followed by a buffer containing 20 mM Tris-HCl pH 8.0 at 4 o C, 1 M NaCl, 35 mM imidazole pH 8.0, 50 mM KCl, 10 mM MgCl2, 2 mM ATP and 2 mM b-mercaptoethanol and then finally washed with lysis buffer.
The complex was then eluted with 20 mM Tris-HCl pH 8.0 at 4 o C, 250 mM NaCl, 500 mM imidazole pH 8.0 and 2 mM b-mercaptoethanol.Fractions containing proteins were pooled, and TEV was added (if needed) whilst performing overnight dialyses against 20 mM Tris-HCl pH 8.0 at 4 o C, 150 mM NaCl and 2 mM DTT.
His-GST-Mis18BP120-130 was purified in the same manner as above with the following modifications: the lysis and elution buffers contained 500 mM NaCl, whilst the dialysis buffer contained 75 mM NaCl.His-MBP-Mis18BP120-130 was purified using the same lysis buffer containing 500 mM NaCl and purified using amylose resin (NEB).Proteins were then eluted by an elution buffer containing 10 mM Maltose.

Interaction trials
Pull-down assays used to test the interaction between the C-terminus of Mis18a and Mis1b were performed by initially purifying the proteins through the cobalt affinity chromatography, as described for wild type proteins, and the eluted fractions were loaded into the amylose affinity resin, pre-equilibrated with a binding buffer consisting of 30 mM Tris-HCl pH7.5, 500 mM NaCl, and 5 mM b-mercaptoethanol.Amylose resins were washed with the binding buffer, and the proteins were eluted with a binding buffer containing 20 mM maltose.The fractions were subjected to SDS-PAGE analysis.
Pull-down assay using the amylose resin to test interactions between Mis18a/b and Mis18BP120-130 were done as described previously (Pan et al., 2017).Briefly, purified proteins were diluted to 10 µM in 40 µl binding buffer, 50 mM HEPES pH 7.5, 1 M NaCl, 1 mM TCEP, 0.01% Tween® 20.One third of the mixture was taken as input, and the remaining fraction was incubated with 40 µl amylose resin for 1 h at 4°C.The bound protein was separated by washing with binding buffer three times, and the input and bound fractions were analysed by SDS-PAGE.

Crystallisation, data collection, and structure determination
Purified Mis18a/bC-term domains and Mis18aYippee were screened and crystallised using the hanging-drop vapour diffusion method at room temperature with a mixture of 0.2 µl of the protein and 0.2 µl of crystallisation screening solutions.The crystals of Mis18a/bC-term domains were grown within a week with a solution containing 0.2 M magnesium acetate and 20% (w/v) PEG 3350.SeMet-substituted Mis18a/bC-term domains crystals were grown by the microseeding method with a solution containing 0.025 M magnesium acetate and 14% (w/v) PEG 3350.The crystals of SeMet-substituted Mis18a/bC-term domains were further optimised by mixing 1 µl of the protein and 1 µl of the optimised crystallisation solution containing 0.15 M magnesium acetate and 20% (w/v) PEG 3350.The crystals of Mis18aYippee were obtained in 2 M ammonium sulfate, 2% (w/v) PEG 400, and 100 mM HEPES at pH 7.5.The crystals of Mis18a/bC-term domains and Mis18aYippee were cryoprotected with the crystallisation solutions containing 20% and 25% glycerol, respectively.The cryoprotected crystals were flash-frozen in liquid nitrogen.Diffraction datasets were collected at the beamline LS-CAT 21 ID-G and ID-D of Advanced Photon Source (Chicago, USA).The data set were processed and scaled using the DIALS (Winter et al., 2018) via Xia2 (Winter et al., 2013).The initial model of Mis18a/bCterm domains was obtained using the SAD method with SeMet-derived data using the Autosol program (Terwilliger, 2000).The molecular replacement of the initial model as a search model against native diffraction data was performed using the Phaser program within the PHENIX program suite (Liebschner et al., 2019).The initial model of Mis18aYippee was calculated by molecular replacement method (Phaser) using yeast Mis18 Yippee-like domain structure (PDB ID: 5HJ0) (Subramanian et al., 2016) as a search model.The final structures were manually fitted using the Coot program (Emsley and Cowtan, 2004) and the refinement was carried out using REFMAC5 (Afonine et al., 2010).The quality of the final structures was validated with the MolProbity program (Chen et al., 2010).

SEC-MALS
Size-exclusion chromatography (ÄKTA-MicroTM, GE Healthcare) coupled to UV, static light scattering and refractive index detection (Viscotek SEC-MALS 20 and Viscotek RI Detector VE3580; Malvern Instruments) was used to determine the molecular mass of protein and protein complexes in solution.Injections of 100 µl of 2-6 mg/ml material were used.
His-SUMO-Mis18a188-233 (∂A280nm/∂c = 0.43 AU.ml.mg -1 ) WT and mutants were run on a Superdex 75 increase 10/300 GL size exclusion column pre-equilibrated in 50 mM HEPES pH 8.0, 150 mM NaCl and 1 mM TCEP at 22˚C with a flow rate of 1.0 ml/min.Light scattering, refractive index (RI) and A280nm were analysed by a homo-polymer model (OmniSEC software, v5.02; Malvern Instruments) using the parameters stated for the protein, ∂n/∂c = 0.185 ml.g -1 and buffer RI value of 1.335.The mean standard error in the mass accuracy determined for a range of protein-protein complexes spanning the mass range of 6-600 kDa is ± 1.9%.

SAXS
SEC-SAXS experiments were performed at beamline B21 of the Diamond Light Source synchrotron facility (Oxfordshire, UK).Protein samples at concentrations >5 mg/ml were loaded onto a Superdex™ 200 Increase 10/300 GL size exclusion chromatography column (GE Healthcare) in 20 mM Tris pH 8.0, 150 mM KCl at 0.5 ml/min using an Agilent 1200 HPLC system.The column outlet was fed into the experimental cell, and SAXS data were recorded at 12.4 keV, detector distance 4.014 m, in 3.0 s frames.Data were subtracted, averaged and analysed for Guinier region Rg and cross-sectional Rg (Rc) using ScÅtter 3.0 (http://www.bioisis.net),and P(r) distributions were fitted using PRIMUS (Konarev et al., 2003).
Ab-initio modelling was performed using DAMMIN (Svergun, 1999), in which 30 independent runs were performed in P1 or P2 symmetry and averaged.

Gradient fixation (GraFix)
Fractions from the gel filtration peak were concentrated to 1 mg/mL using a Vivaspin® Turbo (Sartorius) centrifugal filter, and the buffer exchanged into 20 mM HEPES pH8.0, 150 mM NaCl, and 2 mM DTT for GraFix (Kastner et al., 2008, Stark, 2010).A gradient was formed with buffers A, 20 mM HEPES pH 8.0, 150 mM NaCl, 2 mM DTT, and 5% sucrose and B, 20 mM HEPES pH 8.0, 150 mM NaCl, 2 mM DTT, 25% sucrose, and 0.1% glutaraldehyde using the Gradient Master (BioComp Instruments).500 µl of sample was applied on top of the gradient, and the tubes centrifuged at 40,000 rpm at 4ºC using a Beckman SW40 rotor for 16 h.The gradient was fractionated in 500 µl fractions from top to bottom, and the fractions were analysed by SDS-PAGE with Coomassie blue staining and negative staining EM.

Negative staining sample preparation, data collection and processing
Copper grids, 300 mesh, with continuous carbon layer (TAAB) were glow-discharged using the PELCO easiGlow™ system (Ted Pella).GraFix fractions with and without dialysis were used.Dialysed fractions were diluted to 0.02 mg/ml.4 µl of sample were adsorbed for 2 min onto the carbon side of the glow-discharged grids, then the excess was side blotted with filter paper.The grids were washed in two 15 µl drops of buffer and one 15 µl drop of 2% uranyl acetate, blotting the excess between each drop, and then incubated with a 15 µl drop of 2% uranyl acetate for 2 min.The excess was blotted by capillary action using a filter paper, as previously described (Scarff et al., 2018).
The grids were loaded into a Tecnai F20 (Thermo Fisher Scientific) electron microscope, operated at 200 kV, field emission gun (FEG), with pixel size of 1.48 Å. Micrographs were recorded using an 8k x 8k CMOS F816 camera (TVIPS) at a defocus range of -0.8 to -2µm.
Several rounds of 2D classification were employed to remove bad particles and assess the data, reducing the 14,840 particles to 5,540.These were used to generate three ab-initio models followed by homogeneous refinement with the respective particle sets.
Peptides were loaded at a flow rate of 0.3 µl/min and eluted at 0.2 µl/min or 0.25 µl/min using a linear gradient going from 2% mobile phase B to 40% mobile phase B over 109 or 79 min, followed by a linear increase from 40% to 95% mobile phase B in 11 min.The eluted peptides were directly introduced into the mass spectrometer.MS data were acquired in the datadependent mode with a 3 s acquisition cycle.Precursor spectra were recorded in the Orbitrap with a resolution of 120,000.The ions with a precursor charge state between 3+ and 8+ were isolated with a window size of 1.6 m/z and fragmented using high-energy collision dissociation (HCD) with a collision energy of 30.The fragmentation spectra were recorded in the Orbitrap with a resolution of 15,000.Dynamic exclusion was enabled with single repeat count and 60 s exclusion duration.The mass spectrometric raw files were processed into peak lists using ProteoWizard (version 3.0.20388)(Kessner et al., 2008), and cross-linked peptides were matched to spectra using Xi software (version 1.7.6.3)(Mendes et al., 2018) (https://github.com/Rappsilber-Laboratory/XiSearch)with in-search assignment of monoisotopic peaks (Lenz et al., 2018).Search parameters were MS accuracy, 3 ppm; MS/MS accuracy, 10ppm; enzyme, trypsin; cross-linker, EDC; max missed cleavages, 4; missing mono-isotopic peaks, 2. For EDC search cross-linker, EDC; fixed modification, carbamidomethylation on cysteine; variable modifications, oxidation on methionine.For sulfo-SDA search: fixed modifications, none; variable modifications, carbamidomethylation on cysteine, oxidation on methionine, SDA-loop SDA cross-link within a peptide that is also crosslinked to a separate peptide.Fragments b and y type ions (HCD) or b, c, y, and z type ions (EThcD) with loss of H2O, NH3 and CH3SOH.5% on link level False discovery rate (FDR) was estimated based on the number of decoy identification using XiFDR (Fischer and Rappsilber, 2017).
Scoring function for CLMS.A cross-link was considered satisfied if the Calpha-Calpha distance was less than 22Å.The final score was the fraction of satisfied cross-links.

Sampling.
To determine the structure of the Mis18 complex we used XlinkAssembler, an algorithm for multi-subunit assembly based on combinatorial docking approach (Schneidman- Duhovny andWolfson, 2020, Inbar et al., 2005).The input to XlinkAssembler is N subunit structures and a list of cross-links.First, all subunit pairs are docked using cross-links as distance restraints (Schneidman-Duhovny et al., 2005).Pairwise docking generates multiple docked configurations for each pair of subunits that satisfy a large fraction of cross-links (> 70%).Second, the combinatorial assembler hierarchically enumerates pairwise docking configurations to generate larger assemblies that are consistent with the CLMS data.
XlinkAssembler was used with 11 subunits to generate a model for Mis18a/b: initial hexamer structure based on AlphaFold (Jumper et al., 2021), two Mis18aYippee domains as well as four copies of the two helices in the Mis18a N-terminal helical region (residues 37-55 and 60-76).

Cell culture and transfection
The cell line HeLa Kyoto, HeLa 3-8 (having an alphoid tetO array integrated into one of its The TetR-eYFP tagged proteins were transfected using the XtremeGene-9 (Roche) transfection reagent according to the manufacturer's protocol.The HeLa 3-8 cells attached on to the coverslip in a 12-well plate were transfected with the corresponding vectors (500 ng) and the transfection reagent diluted in Opti-MEM (Invitrogen) followed by incubation for 36-48 h.

Generation of monoclonal antibodies against Mis18a/Mis18b
Lou/c rats and C57BL/6J mice were immunized with 60 µg purified recombinant human
Micrographs were acquired at the Centre Optical Instrumentation Laboratory on a DeltaVision Elite™ system (Applied Precision) or Nikon Ti2 inverted microscope.Z stacks were obtained at a distance of 0.2 µm and were deconvolved using SoftWoRx, or AutoQuant software, respectively, followed by analysis using ImageJ software.The intensity at the tethering site was obtained using a custom-made plugin.Briefly, the CENP-A signal at the tethering site (eYFP) was found for every z-section within a 7-square pixel box.The mean signal intensity thus obtained was subtracted from the minimum intensities within the section, which was then normalised with the average CENP-A intensities of the endogenous centromeres.The values were obtained from a minimum of three biological repeats.Statistical significance of the difference between normalised intensities at the centromere and tethering region was established by a Mann-Whitney U two tailed test using Prism 9.1.2.

SNAP-CENP-A assay and quantification
SNAP-CENP-A quench pulse labelling was done as described previously (Jansen et al., 2007).Briefly, the existing CENP-A was quenched by 10 µM SNAP Cell® Block BTP (S9106S, NEB).The cells were treated with 1 µM STLC for 15 h for enriching the mitotic cell population, and the newly formed CENP-A was pulse labelled with 3 µM SNAP-Cell® 647-SiR (S90102S, NEB), 2 h after release from the STLC block (early G1).After pulse labelling, the cells were washed, fixed and processed for immunofluorescence.Images were obtained using DeltaVision Elite™ system (Applied Precision), deconvolved by SoftwoRx and processed by Image J. The average centromere intensities were obtained using a previously described macro CraQ (Bodor et al., 2012).Briefly, the centromeres were defined by a 7x7 pixel box  f) Model of the Mis18core complex generated using partial structures determined using X-ray crystallography and AlphaFold2 (Jumper et al., 2021) and cross-linking restrained molecular docking in EM maps.Mis18BP1 shown in salmon, Mis18a in purple and Mis18b in light pink.
g) Histograms show the percentage of satisfied or violated cross-links for structures modelled using MODELLER (Sali and Blundell, 1993).
, right panel).To assess the contribution of Mis18b for the centromere association and function of Mis18a, we evaluated the Mis18b mutant (Mis18bL199D/I203D), which cannot form the triple helical assembly with Mis18a, in siRNA rescue assays by expressing Mis18b-GFP tagged proteins in a mCherry-Mis18a cell line (McKinley and Cheeseman, 2014).Depletion of endogenous Mis18b and simultaneous transient expression of Mis18bWT-GFP led to co-localisation of Mis18b with Mis18a at centromeres (Fig. 3b, S4d & S4e).Under these conditions, Mis18bWT-GFP levels at centromeres were comparable to that of the control siRNA.Whereas Mis18bL199D/I203D failed to localise at the centromeres.Strikingly, Mis18bL199D/I203D perturbed centromere association of Mis18a only moderately (Fig 3b,).This suggests that Mis18a can associate with centromeres in a Mis18b independent manner.Next, we assessed the contribution of Mis18b for CENP-A deposition in the Quench-Chase-Pulse CENP-A-SNAP assay described above.Endogenous Mis18b was depleted using siRNA, and Mis18bWT and Mis18bL199D/I203D were transiently expressed as GFP-tagged proteins in HeLa cells expressing CENP-A-SNAP.Mis18bWT rescued new CENP-A deposition to comparable levels to the ones observed in the control siRNA-Mis18b WT condition (Fig 3c).
Mis18a/b protein complex, 5 nmol CpG (TIB MOLBIOL, Berlin, Germany), and an equal volume of Incomplete Freund's adjuvant (IFA; Sigma, St. Louis, USA).A boost injection without IFA was given 6 weeks later and three days before fusion of immune spleen cells with P3X63Ag8.653myeloma cells using standard procedures.Hybridoma supernatants were screened for specific binding to Mis18a/b protein complex and also for binding to purified GST-Mis18b protein in ELISA assays.Positive supernatants were further validated by Western blot analyses on purified recombinant human Mis18a/b complex, on cell lysates from Drosophila S2 cells overexpressing human Mis18a and on HEK293 cell lysates.Hybridoma cells from selected supernatants were subcloned at least twice by limiting dilution to obtain stable monoclonal cell lines.Experiments in this work were performed with hybridoma supernatants mouse anti-Mis18a (clone 25G8, mouse IgG2b/ƙ) and rat anti-Mis18b (clone 24C8; rat IgG2a/ƙ).
using a reference channel, and the corresponding mean signalling intensity at the data channel was obtained by subtracting the minimum intensities within the selection.The values plotted were obtained from a minimum of three independent experiments.Statistical significance of the difference between normalised intensities at the centromere region was established by a Mann-Whitney U test using Prism 9.1.2.

Figure 1 :
Figure 1: Mis18a/b Contains Two Independent Structural Domains that can Oligomerise.

Figure 2 :
Figure 2: Mis18a Mutations Disrupting the Mis18a/b Triple Helical Assembly Result in Loss

Figure 3 :
Figure 3: Mis18a Associates with Centromeres in a Mis18b-Independent Manner but

Figure 4 :
Figure 4: Disrupting the Mis18BP1 Binding Interface of Mis18a Prevents its Centromere