{"paper_id":"2463d97f-eebd-482f-8437-6a9e3b71c65d","body_text":"- R X U Q D O \u0003 1 D P H\nStructural\ndynamics in the CENP-A nucleosome im-\npacted by protein-protein interactions with centromere\nprotein N †\nAbhik Ghosh Moulick, a Sylvia Erhardt, b Wolfgang Wenzel,a and Mariana Kozlowska a∗\nNoncanonical nucleosomes at the centromere contain the histone variant CENP-A, playing a crucial\nrole in chromosome segregation. CENP-A is highly regulated and its centromeric organization is par-\ntially regulated by the centromere protein N (CENP-N). Despite its importance, the protein–protein\ninteractions within the complex formed between CENP-A nucleosomes and CENP-N remain poorly\nunderstood at a molecular level. Here, we employ the SIRAH coarse-grained molecular dynamics\n(MD) simulations to investigate the binding interface and structural rearrangements of proteins in\nthe CENP-A nucleosome complexed with CENP-N. We aim to assess the stability of the CENP-A\nnucleosome and the change in its plasticity upon the CENP-N binding. By the set ofµs-long MDs,\nwe reveal enhanced flexibility in the N-terminal region of CENP-A and stabilization of its RG loop in\nthe complex with CENP-N. This is demonstrated to have rather minor effects on the overall stability\nof the nucleosome and changing its compactness. Nevertheless, the data suggest the binding of\nCENP-N allosterically changes the conformational states of CENP-A and impacts its interactions\nwith other proteins in the histone core. A distance-based contact map analysis further elucidates\nkey residues mediating the interaction between CENP-A and CENP-N, while umbrella sampling\nsimulations quantify their binding free energy, which remains challenging to measure experimentally.\n1 Introduction\nIn eukaryotic cells, DNA is compacted within the nucleus through\na series of structural levels of chromatin, where the nucleosome\ncore particle (NCP) serves as fundamental unit 1. NCP typically\ncomprises 147 bp long duplex DNA wrapped approximately 1.7\ntimes around positively charged histone octamer protein com-\nplexes, which are spaced by so called linker DNA , that facilitate\nhigher-order structure2. Canonical nucleosomes consist of an oc-\ntamere of two copies of the four canonical histones H2A, H2B,\nH3, and H4. Histones are highly conserved and consist of a glob-\nular core region with a smaller fraction of disordered, positively\ncharged segments that extend outwards from the core, called the\nN-terminal histone tails. These tails are highly dynamic, modu-\nlated by a large number of post-translational modifications and\ninvolved in chromatin organization and chromosome condensa-\ntion. 1,3–6. Further complex processes, including chromatin loop-\ning, formation of topologically associating domains, and com-\na Institute of Nanotechnology, Karlsruhe Institute of T echnology (KIT), Kaiserstraße 12,\n76131 Karlsruhe, Germany; E-mail: mariana.kozlowska@kit.edu\nb Molecular Cell Biology Of Animals, , Karlsruhe Institute of T echnology (KIT), Kaiser-\nstraße 12, 76131 Karlsruhe, Germany\n† Supplementary Information available: [details of any supplementary information\navailable should be included here]. See DOI: 00.0000/00000000.\npartmentalization into chromosome territories, progressively or-\nganize the genome, allowing it to be efficiently packed within the\nnucleus 7,8.\nVariations in nucleosome composition, differences in DNA se-\nquences, linker DNA length, and various post-translational mod-\nifications, influence NCP-NCP interactions, and thereby folding\nof chromatin, changing its higher-order structure and gene reg-\nulation in virtually all biological processes, including DNA re-\npair, replication, and gene expression 9–11. For example, cancer-\nassociated mutations in linker histone have been reported to\ndisrupt nucleosome stacking and decompaction of higher-order\nchromatin structures, which in turn result in irregular gene ex-\npression and contribute to oncogenic transformations12. Changes\nin canonical nucleosomes via histone variants are pivotal in the\nepigenetic process by shaping the identity of specific regions in\nthe genome, for example, centromeres. Centromeres act as plat-\nforms for the assembly of kinetochores 13,14, which are large pro-\ntein complexes that mediate the attachment of spindle micro-\ntubules to chromosomes during mitosis and meiosis, maintain-\ning proper chromosome segregation. Centromeric DNA in hu-\nman cells are composed of repetitive AT-rich α-satellite DNA se-\nquences 15,16, along with the presence of the histone H3 variant\nCENP-A. At centromeric chromatin, CENP-A replaces canonical H3\nin a subset of nucleosomes, thus defining the site of kinetochore\n+ P V S O B M \u0001 / B N F \r \u0001 < Z F B S > \r \u0001 < W P M \u000f > \r\n1–16 | 1\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 10, 2025. ; https://doi.org/10.1101/2025.11.10.686994doi: bioRxiv preprint \n\nformation. Thereby , CENP-A acts as an epigenetic marker17–19 for\ncentromere localization and kinetochore formation.\nIn total, sixteen inner kinetochore proteins are associated with\ncentromeric nucleosome, which collectively known as the consti-\ntutive centromere-associated network (CCAN) 20–22. The specific\nbinding of NCPs containing CENP-A with CCAN occurs by complex\nprotein-protein interactions (PPIs), where CENP-N and CENP-C\ndirectly recognize the CENP-A nucleosome necessary for further\nkinetochore assembly and kinetochore segregation 23–25. While\nCENP-N acts as a reader and locator of CENP-A, anchoring the\nCCAN complex26 specifically to CENP-A–containing nucleosomes,\nCENP-C binds CENP-A nucleosomes for recruiting and organiza-\ntion of other CCAN components 27,28. Moreover, CENP-N was\nreported to facilitate the stacking of CENP-A–containing nucleo-\nsomes and the formation of nucleosomal arrays through contacts\nbetween its α6 helix and the DNA of neighboring nucleosomes.\nThis leads to the formation of densely packed chromatin at cen-\ntromeric regions 29. Cryo-EM studies and biophysical analyses\nfurther confirmed that these protein–protein and protein–DNA in-\nteractions are key elements underlying the formation of higher-\norder centromeric structures 29. Hydrogen/deuterium exchange\n(HX) coupled to mass spectrometry experiments additionally re-\nvealed that the N-terminal domain of CENP-N adopts a folded\nconformation, with the first 200 residues forming the major inter-\nface with CENP-A nucleosomes30. Consistently , structural studies\nshowed that human CENP-N (residue GLU3, THR4, and THR7)\nconfers binding specificity through interactions with the L1 loop\nof CENP-A, particularly its exposed RG motif (ARG80–GLY81),\nwhich are further stabilized by electrostatic contacts with nu-\ncleosomal DNA. Several positively charged residues of CENP-N\n(ARG44, LYS45, ARG11, LYS81, LYS148, and ARG169, ARG170),\nlocated proximal to the DNA backbone, likely form stabiliz-\ning interactions with the DNA phosphate groups 25. Overall,\nthe functionality of such noncanonical nucleosomes depends on\ntheir intrinsic structural flexibility 26,30 and on conformational\nchanges stimulated by CCAN protein binding and DNA interac-\ntions 25,30–32. Despite these understandings, detailed dynamical\nstudies of how specific CENP-A–CENP-N interactions guide cen-\ntromere function at the level of a single nucleosome and beyond\nremain elusive. Above all, quantitative data on binding energet-\nics are still lacking from both experiments and simulations, yet\nsuch information is important for understanding whether CENP-\nN binding is essentially constitutive or dynamic, and how it com-\npares in strength to other nucleosome-protein interactions. These\nenergetic insights, along with structural observations, help to bet-\nter understand how protein–protein and protein–DNA interac-\ntions regulate nucleosome function at the centromere.\nThe structual changes of canonical nucleosomes, for example\ntheir loop formation33,34, DNA breathing and unwrapping6,34–36,\ntwist defects 36, nucleosome sliding 37, as well as some structural\ncharacteristics of histone variant nucleosomes 38,39 have been\nstudied to date through different computational methods6,40 with\ndiverse structural resolutions 41–43. Here, molecular dynamics\nand the use of enhanced sampling simulations 44,45 are the most\napplicable in the field. MD simulations have already been em-\nployed for histone variant nucleosome. For example, Kohestani et\nal. 38 elucidated the molecular mechanism by which H2A.B leads\nto a less compact nucleosome state, thereby increasing genetic\naccessibility and gene transcription. Bowerman et al. 46 reported\naltered dynamics and allosteric pathways mediated by changes in\nL1-loop interactions between the two H2A.Z histone copies. Kono\net al. 44 performed free energy calculation based on unwrapping\nof superhelical turn of CENP-A-containing nucleosome and re-\nvealed that the lowest free energy corresponds to the state where\n16 to 22 base pairs were unwrapped. Still, studies related to the\ninteraction of histone variant nucleosomes with other biological\nmacromolecules present in the nucleus remain obscure.\nDue to the large system size of NCPs and biological complexes\nthey may form by interacting with other proteins, gaining long-\ntimescale conformational insights and PPIs through atomistic\nsimulations are computationally expensive. Therefore, coarse-\ngrained (CG) models 47–51 serve as a useful alternative for ex-\nploring such systems. While polymer-based CG models 50,52–55\nand mesoscopic models 43,56–58 are often used for nucleosome\nand chromatin modeling, higher resolution CG force fields (FF)\nlike SIRAH (Southamerican Initiative for a Rapid and Accurate\nHamiltonian) 59,60 or MARTINI61 are capable to decipher protein-\nprotein and protein-DNA interactions with a higher accuracy .\nMARTINI uses largely a bottom-up strategy for bonded interac-\ntions and a top-down strategy for non-bonded interactions dur-\ning parameterization, while SIRAH is mostly bottom-up, physics-\nbased CG FF with long-range electrostatics, thus enabling cap-\nturing of hydrogen bond-like interactions and other interactions\nthat are typically absent in less fine FFs. In addition, owing to its\nparameterization scheme, SIRAH permits unbiased simulation of\nthe secondary structure that often possesses constraints in other\nmodels, and it was demonstrated to correctly reproduce PPI 62\nand protein-DNA interactions 63–65 in comparison to experimen-\ntal observations. Furthermore, the SIRAH FF was recently applied\nto understand protein-DNA interactions and dynamics of canoni-\ncal nucleosomes in comparison to all-atom simulations 65,66.\nDue to adequate quality of SIRAH in describing interactions\nbetween biological macromolecules and structural dynamics of\nbiologically relevant complexes 67–70, we selected this FF for in-\ndepth characterization of protein binding and dynamics of the\nhistone variant nucleosome system with CENP-A towards under-\nstanding of its CENP-N–driven structural dynamics and binding\nenergetics. Using the SIRAH FF, we first assess nucleosome sta-\nbility in the presence and absence of CENP-N and characterize the\nbinding interface between CENP-N, CENP-A, and DNA. We then\nanalyze the conformational flexibility of CENP-N in both its free\nand bound states, providing microscopic details of its structuring\nupon NCP binding. Finally , we quantify the PPIs binding strength\nthrough umbrella sampling (US) simulations. With this approach,\nwe aim to gain mechanistic insights into how CENP-N contributes\nto CENP-A nucleosome stability and recognition, providing a bet-\nter understanding of the functionality and structural changes of\ncentromeric NCPs.\n2 | 1–16\n+ P V S O B M \u0001 / B N F \r \u0001 < Z F B S > \r \u0001 < W P M \u000f > \r\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 10, 2025. ; https://doi.org/10.1101/2025.11.10.686994doi: bioRxiv preprint \n\nDNACENP-AH4H2AH2BCENP-N\n(a) (b)\nSHL 0SHL 1SHL 2\nSHL 3SHL 4SHL 5SHL 6\nSHL 7(c)\nFig. 1 (a) The all-atom structure of the CENP-A-containing nucleosome along with the CENP-N protein obtained in cryo-EM29 and further refined\nusing atomistic simulations (this work). The histone core and CENP-N protein are visualized using New Cartoon and four histone pairs are marked in\nred, green, yellow and cyan for CENP-A, H2B, H4 and H2A, respectively, while DNA is visualized using QuickSurf and is marked in grey. The CENP-N\nprotein is marked in salmon. (b) The coarse-grained representation of the CENP-A nucleosome with bonded CENP-N as mapped from its refined\nall-atom structure. (c) The visualization of superhelical location (SHL) sites marked over nucleosomal DNA. Each SHL typically spans approximately\n10 base pairs, ranging from SHL0 to SHL±7.\n2 Methods\n2.1 System preparation\nWe considered the cryo-EM structure of CENP-A nucleosome\nin complex with the CENP-N protein taken from protein data\nbank (PDB) under ID 7U46 29. The histone part of this cryo-\nEM structure has missing histone tails, and CENP-N protein has\nsome missing residues, see Table S1 in Supplementary Informa-\ntion (SI). These missing residues and histone tails were mod-\neled in the present work using AlphaFold3 71. Further struc-\nture refinements were performed by means of atomistic molec-\nular dynamics simulations using Amber99SB-ILDN force field 72\nand TIP3P water model 73. Simulations were performed in the\nGROMACS simulation package 74, version 2019.2. At first, the\nNCP system was immersed in a cubic box of water of dimensions\n167.35×167.35×167.35Å\n3\n. The system was neutralized by adding\n149 Na+ ions and minimized for 50000 steps using the steepest\ndescent algorithms. Further equilibration proceeded in two steps\nusing NVT, and later NPT, ensembles keeping position restraint on\nheavy atoms. It was carried out at 300 K and 1 bar using the V-\nrescale thermostat and Parrinello-Rahman barostat with isotropic\npressure coupling. Short-range van der Waals and Coulomb in-\nteractions were truncated at 10 Å. Long-range electrostatics were\ntreated using the Particle Mesh Ewald (PME) method with a cu-\nbic interpolation order of 4 and a Fourier grid spacing of 1.6 Å.\nNeighbor lists were updated every 10 steps using the Verlet cutoff\nscheme with a grid-based search. The production run of all-atom\nsimulation was performed for 10 ns with 2 fs integration time\nstep employing periodic boundary conditions in all directions. All\nbonds involving H atoms were constrained using the LINCS algo-\nrithm. The final structure obtained from the atomistic simulation\n(presented in Fig.1a) was used for further CG simulations.\n+ P V S O B M \u0001 / B N F \r \u0001 < Z F B S > \r \u0001 < W P M \u000f > \r\n1–16 | 3\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 10, 2025. ; https://doi.org/10.1101/2025.11.10.686994doi: bioRxiv preprint \n\nTable 1 Summary of the systems setup used for the CG-MD simulations with SIRAH.\nSystem studied CENP-A Nucleosome + CENP-N CENP-A Nucleosome Only CENP-N\nPBC Box (Å) 192.1×181.3×157 171.3×161.5×139.8 115.4×108.8×94.2\nNo. of CG beads 67454 48078 14788\nNo. of CG solvent molecules 14620 10222 3269\nNo. of Na+ ions 536 399 101\nNo. of Cl- ions 387 247 104\nSalt concentration (M) 0.15 0.15 0.15\n2.2 Coarse-grained MD simulations\nCG-MD simulations were performed for two main NCP systems:\n(i) CENP-A nucleosome with bounded CENP-N protein (this sys-\ntem was labeled as \"CENP-A Nucleosome + CENP-N\") and (ii)\nCENP-A nucleosome without CENP-N protein (labeled as \"CENP-\nA Nucleosome\"). The second system was prepared by removing\nCENP-N from the final structure obtained from the atomistic simu-\nlation of CENP-A Nucleosome + CENP-N (see Section 2.1). Simu-\nlations were performed following the computational setup in the\nprevious work of simulating canonical NCP system 66. All-atom\nrefined structures were protonated using PDB2PQR 75 server at\nneutral pH assumption following the AMBER naming scheme nec-\nessary for SIRAH. The structures obtained were mapped into CG\nform using SIRAH 2.2 tools (Fig. 1b). After mapping, the com-\nplexes were solvated in a PBC box with SIRAH WT4 water type76.\nAll systems were neutralized and the physiological conditions\nwere modeled by adding Na + and Cl - ions at a salt concentra-\ntion of 0.15 M. The required number of ions, PBC box size, and\nnumber of solvent CG beads are listed in Table 1. The simula-\ntion box was defined using a truncated octahedral shape with a\n20 Å distance between the solute and the box edges to ensure\nthat the complexes did not interact with their periodic images.\nEnergy minimization was performed in two steps: (i) first, for\n50000 steps with the protein side chains minimization while re-\nstraining the backbone, and (ii) second, for 5000 steps with the\nentire system minimized without constraints. In both cases, the\nsteepest descent algorithm was utilized. The side chain equilibra-\ntion enhances structural stability of protein molecule by avoiding\nsignificant distortions to the secondary structure. Subsequently ,\nthe solvent molecules were equilibrated around the complex by\nrunning a 5 ns simulation with harmonic restraints applied to all\nCG beads. The temperature of the system was maintained at 300\nK using a V-rescale thermostat77 and the constant isotropic pres-\nsure of 1 bar using Parrinello-Rahman barostat 78, respectively .\nBefore production run, an additional run of 10 ns was performed\nkeeping restrain on protein, allowing the DNA to relax in the in-\nterface. Finally , unrestrained production run was carried out for\n10 µs. The time step for the simulation was maintained at 20\nfs. The electrostatic interactions were computed using the PME\nmethod with a 12 Å cutoff and a grid spacing of 2 Å. Van der\nWaals interactions were treated with a 12 Å cutoff. Three sepa-\nrate replicas of each system were simulated for a duration of 10\nµs by starting from different random seeds. The solvated CENP-\nN protein without the NCP was simulated for three independent\nreplicas using the same protocol as for other complexes, exclud-\ning the DNA relaxation step. The system specific information is\nspecified in Table 1.\nSIRAH-Backmap tools 79 were used for backmapping CG-MD\ntrajectories into atomistic resolution. The backmapping pro-\ncedure involves reconstructing atomistic positions on a per-\nresidue basis 80, preserving the geometric structure (internal co-\nordinates), followed by protonation and minimization using the\natomistic force field ff14SB atomistic force field 81 in Amber-\nTools82 tleap module.\n2.3 Analysis of nucleosome dynamics\nTo characterize the dynamical properties of the CENP-A nucleo-\nsome and its structural changes upon the specific protein bind-\ning simulated with CG-MD, we applied a set of analyses ex-\nplained here. Global stability was monitored through the root-\nmean-square deviation (RMSD), while local flexibility was cap-\ntured by root-mean-square fluctuations (RMSF). For both analy-\nses, we considered the backbone CG beads of both histone and\nDNA. RMSD and RMSF were calculated relative to the energy-\nminimized structure explained in Section 2.2. Changes in the\noverall compaction of the nucleosome were quantified using the\nradius of gyration (Rg), which is a mass-weighted root-mean-\nsquare distance of all beads from the center-of-mass (COM). To\nprobe sequence- and region-specific histone–DNA interactions,\nwe computed contact maps between protein and DNA beads. Fol-\nlowing earlier work 83, native contacts were defined when the\nspecified beads (GC for protein representing carbon atom, PX\nfor DNA representing backbone phosphate atom) were within\n7 Å. In the SIRAH CG representation, the direct identification\nof specific interactions such as hydrogen bonds or salt bridges\nis not feasible due to the reduced resolution of the model. In-\nstead, residue–residue interactions were characterized using this\ndistance-based contact definition. For each residue pair, the av-\nerage contact population value was calculated as the fraction of\nsimulation frames in which the contact was present. An average\ncontact value of 1.0 indicates that the residues remained in con-\ntact throughout the entire trajectory , representing a highly sta-\nble interaction. Contacts with values below 0.4 were classified\nas transient or weak, indicative of higher flexibility in that re-\ngion. The average contact values between protein residues were\nfurther visualized in the form of 2D contact maps, where the\naxes correspond to residue indices and the color scale indicates\nthe stability of the contacts across the trajectory . A dark/intense\ncolor (the average contact value close to 1.0) shows that the two\nresidues stayed in contact throughout the trajectory (stable, per-\nsistent interaction) while lighter color (less than 0.4) shows weak\nor transient contacts. All analyses were performed using GRO-\nMACS tools, while MDAnalysis was utilized to calculate the con-\ntact map 84.\n4 | 1–16\n+ P V S O B M \u0001 / B N F \r \u0001 < Z F B S > \r \u0001 < W P M \u000f > \r\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 10, 2025. ; https://doi.org/10.1101/2025.11.10.686994doi: bioRxiv preprint \n\nThe secondary structure of proteins was calculated using\nSIRAH secondary structure tool. It uses the positions of CG\nbackbone beads to approximate the backbone geometry defined\nas a structured element in terms of helical and extended re-\ngions (i.e. β-strands), or unstructured element defined as a coil.\nSince the SIRAH CG geometry of proteins retains enough infor-\nmation about local backbone curvature and spacing, these CG-\nbased geometric rules provide an accurate distinction between\nhelices from sheets 59. To extract dominant modes of motion in\nhistone CENP-A upon CENP-N binding and to construct free en-\nergy landscapes (FEL), representing thermodynamic map of con-\nformational space of a protein, the principal component analy-\nsis (PCA) was employed. It is capable to reveal the functionally\nrelevant conformational states of molecules during the trajecto-\nries and reduce dimensionality that helps in identifying configu-\nrational spaces with a limited number of degrees of freedom. In\nthis method, a 3N × 3N covariance matrix of positional fluctua-\ntions of CG beads relative to every other coordinates over time is\nconstructed based on simulated trajectory . Thus, it captures the\ncorrelated motions of coordinates, and the diagonalization of this\nmatrix yields eigenvectors, representing principal motion direc-\ntions, and the corresponding eigenvalues, indicating the magni-\ntude of fluctuations along these directions. The trajectory is then\nprojected onto these eigenvectors to derive the principal compo-\nnents (PCs). The first two PCs describe the dominant large-scale\nmotions and are oftenemployed to construct a two-dimensional\nFEL. It is based on the estimation of the joint probability den-\nsity function (P(PC1,PC2)) obtained from a histogram of PCs and\ndefined as ∆GFEL (x, y) = −kBT ln(P(x, y)/Pmax), where Pmax is the\nprobability of the most probable state, kB is Boltzmann constant\nand T is the temperature. The PCA and FEL were calculated us-\ning MDAnalysis Python library 84,85 providing input coordinates\nof CG beads. Together, the analyses performed provide a compre-\nhensive view of nucleosome stability , flexibility , as well as collec-\ntive dynamics.\nThe errors for the quantities calculated were estimated where\napplicable. The error of the average is given by s/√\nn, where s is\nthe standard deviation of the mean values over the replicas, and\nn represents the number of replicas.\n2.4 Umbrella Sampling simulations\nTo understand PPIs and binding free energy between CENP-A pro-\ntein in an NCP and CENP-N CCAN protein, US simulations us-\ning SIRAH FF were performed 62,86. The initial structure of the\ncomplex was taken from the final structure obtained from the\natomistic simulations as discussed above. It was placed in a box\nof 320 × 200 × 160Å\n3\n, so that the complex was aligned parallel\nto the X-axis. The system was solvated, neutralized, and mini-\nmized as per the prior simulation protocol, explained in Section\n2.2. Equilibration was then performed under NVT and NPT con-\nditions, maintaining consistency with the MD simulation proto-\ncol. To initiate the US simulation, at first we performed steered\nmolecular dynamics (SMD) simulation to generate initial confor-\nmations for US windows. The CENP-N protein was pulled away\nfrom its initial position near the CENP-A side of the NCP into the\nbulk solvent along the X-axis, applying a harmonic pulling po-\ntential with a force constant of 1000 kJ mol−1 nm−2 with a pulling\nrate of 0.0001 nm/ps. To prevent drifting of the system along the\nreaction coordinate, a harmonic restraint with force constant of\n20 kJ mol−1 nm−2 for whole nucleosome, except CENP-N protein,\nwas applied. Positions of the molecular system simulated were\nsaved during the course of pulling with 60 structures generated\nas US windows for separate MD simulations. Each window struc-\nture was saved every 0.1 nm from the initial position of CENP-N\nand up to a COM distance between the CENP-A and CENP-N pro-\nteins of 7.95 nm. From a COM distance of 8.05 to 14.65 nm, the\nspacing was fixed at 0.2 nm. The schematic of US procedures is\ndepicted in Fig. S1.\nEach system in the US window was independently further sim-\nulated for 25 ns in thr NPT ensemble followed by an additional\nMD simulation for 50 ns with a V-rescale thermostat and Par-\nrinello–Rahman barostat. A harmonic bias potential with a force\ncosntant of 1000 kJ mol−1 nm−2 was applied to each window. This\nvalue was selected within the range (500–2000 kJ/mol· nm) re-\nported in previous SIRAH umbrella sampling studies62,63 on pro-\ntein–protein interactions and was further optimized to achieve\nadequate histogram overlap between adjacent windows, ensur-\ning sufficient sampling across the reaction coordinate. Fig. S2\ndepicts overlap of histograms from different windows obtained\nalong reaction coordinate. Finally , the potential of mean force\n(PMF) was derived using the weighted histogram analysis method\n(WHAM) 87,88 as implemented in GROMACS to eliminate the in-\nfluence of the applied bias. Statistical errors were estimated\nby using default Bayesian bootstrapping algorithm built into the\nWHAM program 88.\nThe binding free energy (∆G) between proteins, indicating the\nstrength and the character of PPIs was calculated as the differ-\nence in free energy between the bound and unbound states. The\nunbound state was defined as a conformational state where in-\nteractions between the proteins are completely diminished, cor-\nresponding to a plateau (near-zero interaction) in the PMF plot.\n∆G can be calculated mathematically as follows:\n∆G = (−kBT ln\nZ bound\ne−Φi/kBT ) − (−kBT ln\nZ unbound\ne−Φi/kBT ) (1)\nΦi represents the PMF value associated with the ith bin along the\nreaction coordinate. The error in the binding energy is calcu-\nlated by propagating the errors89 from the minimum PMF (bound\nstate) and the plateau region (unbound state) using the following\nequation:\nσ∆G =\nq\nσ 2\nmin + σ 2\nplateau (2)\nHere σmin is the error at the minimum PMF and σplateau is the\naverage error in the plateau region.\n3 Results\n3.1 Stability of CENP-A nucleosome upon CENP-N binding\nSince the function of NCPs depends on their structural flexibility ,\ninteractions with other NCPs or molecules, as well as upon chang-\ning environment conditions, we aimed to analyze the stability of\nCENP-A nucleosomes in the presence and absence of CENP-N pro-\n+ P V S O B M \u0001 / B N F \r \u0001 < Z F B S > \r \u0001 < W P M \u000f > \r\n1–16 | 5\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 10, 2025. ; https://doi.org/10.1101/2025.11.10.686994doi: bioRxiv preprint \n\n180°\n(a)\n(b)\n(c)\nCENP-A Nucleosome\nCENP-A Nucleosome + CENP-N Protein\nFig. 2 (a) The average root mean square deviation and (b) radius of gyration of the nucleosome with and without histone tails considered and\nthe nucleosomal DNA alone. Structural parameters for the CENP-A nucleosome with and without CENP-N protein are marked in red and black,\nrespectively. (c) The structure of the systems simulated after 10µs CG-MD simulations: the CENP-A nucleosome + CENP-N protein complex (top),\nand CENP-A nucleosome (bottom). Two views of the same complex are depicted.\ntein. For that, we calculated RMSD and Rg using data obtained\nfor three independent replicas (see more details in Fig. S3 and\nFig. S4). The RMSD has been calculated for both the histone\nprotein core and nucleosomal DNA separately . We further cate-\ngorized the RMSD calculations for the histone core based on the\ninclusion and omission of the residues present in histone tails,\nwhich are generally inherently flexible, thus, hindering the un-\nderstanding of subtle changes in other protein regions. The re-\nspective data are labeled as \"with tails\" and \"without tails\" in Fig.\n2. The time evolution of RMSD for all cases is presented in Fig.\nS3 while time dependent RMSD of only histone tails are depicted\nin Fig. S5 .\nThe average RMSD plot (Fig. 2a) shows that considering hi-\nstone tails in the RMSD calculation, the fluctuation is higher in\nthe system with bounded CENP-N to the CENP-A system (RMSD\nof 7.51±0.33 Å, see marked in red) as compared to the unbound\nCENP-A system (6.94±0.10 Å, in black). It is connected to the in-\nteractions induced by the presence of CENP-N explained further.\nThe RMSD of systems without the consideration of the histone\ntails is lower, indicating higher histone core stability . In addition,\nthe RMSD of the histone core in the presence of CENP-N is even\nslightly smaller, i.e., 5.12±0.08 Å while in the absence of CENP-N\nthe average RMSD is 5.34± 0.08 Å. This indicates a possible sta-\nbilization of the nucleosome upon CENP-N binding. The average\nvalue of RMSD of nucleosomal DNA in the presence and absence\nof CENP-N is 5.45± 0.21 Å and 5.63± 0.08 Å, respectively (Fig.\n2a). It suggests a slight rigification of some base pairs that are\nlocated near the CENP-N binding site, permitting their smaller\nstructural deviations. Together with a slight histone core struc-\ntural stabilization, it provides evidence for possible NCP plasticity\nchanges, which we discuss in the following section.\nThe average radius of gyration for both systems is depicted in\nFig. 2b. The presence of CENP-N protein does not alter larger\nstructural moves and the shape of the NCP. Time evolution of Rg\nis shown in Fig. S4. The final structure of the NCPs after the sim-\nulation is visualized in Fig. 2c. The upper panel depicts the CENP-\nA nucleosome with bound CENP-N, while the lower panel shows\nthe unbounded CENP-A NCP. Distinct orientations are displayed\nfor each system to convey the complete molecular architecture\nand assembly . The figure demonstrates that the systems consid-\nered in this study retained their structural integrity throughout\nthe microsecond-scale CG-MD simulations.\n3.2 CENP-N binding interface: contacts with CENP-A and\nDNA\nCENP-N interacts specifically with the CENP-A histone and nu-\ncleosomal DNA, forming extensive contacts that are critical for its\nbinding specificity and centromere function. Fig. 3a highlights the\ninteraction interface between DNA, CENP-A, and CENP-N. The L1\nloop of CENP-A is shown in red, while the DNA region, spanning\nSHL 2 to 3.5 (see Fig. 1c) that participates in binding, is high-\nlighted in yellow. Residues of CENP-N interacting with CENP-A\n6 | 1–16\n+ P V S O B M \u0001 / B N F \r \u0001 < Z F B S > \r \u0001 < W P M \u000f > \r\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 10, 2025. ; https://doi.org/10.1101/2025.11.10.686994doi: bioRxiv preprint \n\nSHL 2-3.5L1 loop\nCENP-N residueswith DNACENP-N residueswith CENP-A\n(a)(b)\n(d)\n(c)\nFig. 3 (a) The binding interface involving residues from CENP-A, CENP-N, and DNA bases was obtained from the cryo-EM structure. Red indicates\nthe L1 loop of CENP-A; yellow highlights DNA from SHL 2 to 3.5, the region interacting with CENP-N. Green denotes the CENP-N amino acids that\ninteract with CENP-A, while blue marks those interacting with the DNA SHL. (b) Contact map showing interactions between CENP-N and CENP-A\nresidues. (c) Contact map showing interactions between CENP-N residues and DNA bases. (d) Binding interface obtained from CG simulation with\ncolor coding as in panel (a).\nand DNA are shown in green and blue, respectively . This repre-\nsentation provides a structural overview of the binding sites iden-\ntified from experimental cryo-electron microscopy analysis 25. To\nevaluate the stability of experimentally observed residue contacts\nunder dynamic conditions, we analyzed these interactions over\n10 µs of CG-MD simulations, from which 8 µs were considered\nfor data analysis.\nIn Fig. 3b, the contact map between CENP-N and CENP-A\nresidues over 4000 snapshots from CG-MD simulations is de-\npicted. This contact analysis (see Methods section for de-\ntails) confirms that CENP-A residues forming the RG loop (i.e.,\nARG80–GLY81) interact with CENP-N, consistent with cryo-EM\ndata, and suggests stable contacts at this interface during CG-\nMD. ARG80 of CENP-A interacts strongly with residue ASN145 of\nCENP-N, resulting in the average contact value of 0.79. Its inter-\naction with residues PRO144 and GLN146 from CENP-N is weaker\n(contact value of 0.21 and 0.11 respectively), indicating a broader\ncontact region. Residue GLY81 stands out with strong and mul-\ntiple contacts, notably with residues ASN145 and GLN146 (both\nwith an average contact population of 1.00), and significant inter-\nactions with residues PHE8, PRO144, and TYR147, pointing to a\nhighly engaged interface. The average contact population values\nof all participating residue pairs are listed in Table S2.\nTo maintain its main function, CENP-N should primarily bind to\nthe CENP-A protein in the histone core23,90. However, it addition-\nally shows the interaction with the DNA as depicted in Fig. 3a,\ntherefore, the respective contact map between the CENP-N pro-\ntein and the DNA was calculated (see Fig. 3c). The CG-MD simu-\nlations denote several persistent contacts between the DNA bases\nand CENP-N residues. DNA base G98 exhibits strong interactions\n+ P V S O B M \u0001 / B N F \r \u0001 < Z F B S > \r \u0001 < W P M \u000f > \r\n1–16 | 7\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 10, 2025. ; https://doi.org/10.1101/2025.11.10.686994doi: bioRxiv preprint \n\n180°\n(a)\n(b)\nΔRMSF >0 ΔRMSF <0\n1\n1\n2\n3\n2\n(c)\n(d)\nFig. 4 Root mean square fluctuations of the CENP-A protein in the presence (red) and absence (black) of the CENP-N protein. RMSF values were\naveraged over three independent replicas for both systems. The cyan-shaded areas denote CENP-A residues that exhibit higher RMSF in the presence\nof CENP-N, while the green-shaded areas mark residues with lower RMSF. These residues are further highlighted in (b), where the final CG-MD\nstructure is depicted with the color coding consistent with the RMSF plot. The free energy landscape of CENP-A along PC1 and PC2 shown in the\npresence (c) and absence (d) of CENP-N. Representative structures of CENP-A corresponding to each minimum are depicted in panels outside the\nfigure. ∆GFEL is provided in kcal/mol.\nwith ARG169 and LEU168 of CENP-N, with average contact pop-\nulation of 0.99 and 0.73, respectively . The high frequency of con-\ntact formation indicates the attractive character and stability of\nthe CENP-N binding at this interface (see Table S3). Similarly ,\nG99 exhibits stable contacts with LYS81 and VAL82, both showing\nmaximum contact values of 1.00, along with a weaker interaction\nwith TRP83. The adjacent base A100 also forms strong contacts,\nparticularly with LYS148 (1.00), TYR147 (0.85), and again with\nLYS81 (0.97), indicating a consistent role of this region in DNA\nrecognition. A43 base from DNA strand-2 (DNA base number\n190 in Fig.3c) strongly interacts with ARG44, LYS45, and GLU46,\nwith contact values exceeding 0.75, while C44 (DNA base num-\nber 191 in Fig.3c) forms multiple contacts with MET18, ASN19,\nand LYS45, further supporting weak interaction with ARG44. T49\n((DNA base number 196 in Fig.3c)) also shows a modest contact\nwith ASN19. These interactions (see values listed in Table S3)\nhighlight key regions of CENP-N that stably associate with the\nDNA, suggesting their role in mediating nucleosome binding and\nstabilizing the DNA–protein interface in the CENP-A nucleosome.\nThe binding interface formed by CENP-A, CENP-N, and DNA is\ndepicted in Fig. 3d, which represents the final snapshot taken af-\nter 10 µs-long MD simulation. The amino acid residues from the\nproteins as well as the DNA nucleic bases are shown in surface\nrepresentation, using the same color scheme as described in Fig.\n3a.\n3.3 CENP-A structural stability upon CENP-N binding\nThe CG-MD simulations indicate that the overall structure of the\nCENP-A-containing NCP remains largely unchanged upon CENP-N\nbinding (Fig. 2). To examine potential structural changes specif-\nically in the CENP-A protein within the histone core in the pres-\nence and absence of CENP-N, the RMSF of CENP-A was calculated\n(see Fig. 4a). The region near the L1 loop of CENP-A (specifically\nresidues CYS75, VAL76, LYS77, PHE78, THR79, ARG80, GLY81,\nVAL82) possesses lower fluctuations in the CENP-N bound state\n(data in red) compared to the unbound state (data in black), in-\ndicating that CENP-N binding stabilizes this region. Fluctuations\nin the C-terminal residues (Res 134-Res 140) remain largely un-\nchanged. However, the N-terminal residues exhibit higher value\nof RMSF in the presence of CENP-N (exceeding 1 Å). Such an\nobservation supports that CENP-N binding induces increased flex-\nibility at the N-terminal region, potentially facilitating conforma-\ntional adjustments required for further stability . We marked the\nCENP-A residues showing higher and lower RMSF in the pres-\nence of CENP-N relative to the CENP-N unbound state in cyan and\ngreen, respectively . These residues are shown in Fig. 4b (with re-\nspective regions indicated in Fig. 4a), which represents the final\nsnapshot taken after 10 µs-long MD simulation.\n8 | 1–16\n+ P V S O B M \u0001 / B N F \r \u0001 < Z F B S > \r \u0001 < W P M \u000f > \r\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 10, 2025. ; https://doi.org/10.1101/2025.11.10.686994doi: bioRxiv preprint \n\nTable 2 Secondary structure content (in %) for CENP-N protein regions in bound and free states. Errors are indicated as mean± error.\nRes 1-50 Res 51-200 Res 201-295\nCategory Bound Free Bound Free Bound Free\nStable ss: Helix 52.67± 3,75 49.37± 2.37 30.67± 1.01 32.13± 1.67 30.60± 0.99 33.07± 2.20\nStable ss: Extended 5.03± 0.88 4.70± 2.36 23.07± 1.90 22.30± 1.22 8.20± 1.56 5.73± 0.95\nStable ss: Helix + Extended 57.70± 3.85 54.07± 3.45 53.74± 2.16 54.43± 2.06 38.80± 1.84 38.80± 2.40\nCoil 42.30±3.96 45.97± 1.60 46.23± 1.31 45.57± 1.19 61.17± 0.67 61.17± 1.80\nTo further understand the CENP-A conformational change upon\nthe CENP-N binding, we performed the PCA analysis as described\nin the Methods section. In Fig. 4c,d, the FEL plots along two\nprincipal components (PC1 and PC2) for CENP-A histone in the\npresence as well as absence of CENP-N are visualized. In both\ncases, we observe distinct energy basins indicating distinct stable\nconformational states of CENP-A. In the absense of CENP-N, the\ntwo stable conformational states of CENP-A are structurally sim-\nilar and are separated by the low energy barrier. However, the\nnumber of basins upon the CENP-N binding increases (Fig. 4c),\nrepresenting stronger structural changes. Such conformational\nchanges likely reflect the specific interactions between CENP-A\nand CENP-N, stabilizing additional conformational states and re-\nvealing a more complex free energy landscape. The identified rep-\nresentative structures corresponding to each energy basin for both\ncases are depicted in panels outside in Fig. 4c,d. They were iden-\ntified by picking conformations at the minima of the FEL spanned\nby PC1 and PC2 shown in Fig. 4c,d. We used backmapped all-\natom structures to better visualize the conformational changes\nacross the different basins. We see that the stable helical regions\nremain largely unchanged between structures, while the loop re-\ngions exhibit distinct conformations in each minimum. In the\nCENP-N unbound case (Fig. 4d), the representative structures of\nCENP-A also show variations in loop conformations while main-\ntaining the helical structure.\n3.4 Free and bound states of CENP-N\nTo explore whether CENP-N and CENP-A coevolve to support\nNCP’s function in the centromere region, we analyzed the struc-\ntural dynamics of CENP-N both in its complex with the CENP-\nA-containing nucleosome (labeled as bound CENP-N) and in its\nisolated state (labeled as free CENP-N). Hence, we performed\nan additional CG-MD simulation of free CENP-N as explained in\nthe Methods section. Fig. 5 shows the set of analyses conducted\nto demonstrate the structural differences. Using RMSF calcula-\ntion depicted in Fig. 5a, differences in fluctuations of specific\nregions of CENP-N in its both states are visualized. The RMSF\nplot clearly demonstrates that residues 1–200 become more or-\ndered (stabilized) upon nucleosome binding (see data in red),\nwhereas residues 201–295 display consistently higher RMSF val-\nues in both states. Thus, in the bound case, CENP-N exhibits in-\ncreased flexibility in its C-terminal residues res216-res255 (cyan\ncolored region in Fig. 5a) which are known to be disordered\naccording to previous experimental studies 30. In contrast, the\nN-terminal residues display lower RMSF values in the bounded\nstate of CENP-N, indicating significant stabilization due to interac-\ntions with CENP-A and nucleosomal DNA. The N- and C-terminal\nresidues are highlighted in the final conformation of CENP-N in\nboth the bound and free states in panels outside Fig. 5a. Both the\nCG representation, as well as the backmapped all-atom structures\nof the protein are visualized. In the CG representation, the CG\nbeads of each amino acid are colored according to their RMSF\nvalues as indicated in the color bar. Regions of the molecule\nwith relatively higher RMSF values, indicating greater flexibil-\nity , are shown in red, while regions representing relatively lower\nfluctuations are shown in blue. The RMSF coloring of CENP-N\nwas generated using a user-defined color scale in VMD 91. This\ncoloring reflects relative differences in flexibility across the pro-\ntein rather than absolute RMSF values. The N- and C-terminal\nresidues are marked within dashes circles in the CG representa-\ntion, while they are highlighted in pink over the protein’s sec-\nondary structure in the atomistic representation. The N-terminal\nresidues 1–50 of CENP-N show a more intense red color of the\nCG beads in the free state of the protein, indicating their higher\nfluctuations. In the bound state, these fluctuations are signifi-\ncantly reduced, showing medium relative RMSF. In contrast, the\nC-terminal residues (201–295) display similar color patterns in\nboth free and bound states, suggesting comparable fluctuations,\nwhich result in a higher flexibility of this region present in both\nstates of CENP-N.\nThe backmapped all-atom structures reveal that several\nresidues from the N-terminus of CENP-N adopt loop-like confor-\nmations in both states, despite largely belonging to helical re-\ngions. By contrast, the C-terminal tail predominantly forms loop\nstructures in both bound and free states. Henceforth, it is im-\nportant to calculate the secondary structure (ss) elements of the\nCENP-N protein in both the bound and free states across MD sim-\nulation replicas. At first, we performed the ss analysis for protein\nat both free and bound state using SIRAH ss tool as explained in\nMethods section. Fig. 5b shows the average ss content, along\nwith error bars from three independent replicas, for both sys-\ntems. We also calculated the time evolution of the ss content\nfor the bound and free states of CENP-N (see Fig. S6) over the\nlast 8 µs. Both results revealed no significant changes in the ss\nbetween the two states. Considering the differing flexibility of\nparticular regions of the protein, we identified three different re-\ngions that were studied separately in more detail. The first region\nwas located between residue 1 to 50, which showed the strongest\nchanges in the RMSF behavior upon binding. The second region\n(residue 51 to 200), demonstrating no significant RMSF changes\nupon binding and the third region (residue 201 to 295), which re-\nmain disordered indicating higher RMSF in both bound and free\nstate (Fig. 5a). The secondary structure for these three differ-\nent regions is given in Table 2, where helix and extended regions\nrepresent stable ss, while coil represents unstructured ss. From\nthese data, we see that the first region shows increased contri-\n+ P V S O B M \u0001 / B N F \r \u0001 < Z F B S > \r \u0001 < W P M \u000f > \r\n1–16 | 9\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 10, 2025. ; https://doi.org/10.1101/2025.11.10.686994doi: bioRxiv preprint \n\n  \n(a)\n(b) (c)\nBound CENP-N Bound CENP-N\nFree CENP-N\nFree CENP-N\nRMSF of GC bead\nFig. 5 (a) Root mean square fluctuations of the bound (red) and free (black) CENP-N protein, averaged over three independent replicas. . The\ncyan-shaded areas denote CENP-N residues that exhibit higher RMSF in bound state, while the green-shaded areas mark residues with lower RMSF\nwith respect to free state. Final structures of both free and bound CENP-N are shown at CG and backmapped all-atom resolutions. In all-atom\nrepresentation, the N- or C-terminal residues are highlighted in pink, while the rest of the protein is shown in yellow. b) Percentage of secondary\nstructure elements observed over the simulation trajectory for both bound and free CENP-N. (c) Average RMSD plots of unbound and free CENP-N\nfor three regions: overall, residues 1–200, and residues 201–295. Error bars represent standard deviations from three independent replicas.\nbution of stable ss from 54.07% to 57.70% (especially for helix\ncontent, which changes from 49.37% to 52.67%) upon binding\nas compared to the free state of the protein. In the other regions,\nthe stable ss content versus coil does not change markedly: less\nthan 1% of stable ss decreases or remains unchanged, with over-\nall flexibility largely unaffected upon binding with CENP-A and\nnucleosomal DNA.\nSince experimental studies, such as HX exchange 30, have re-\nported that the N-terminal domain of CENP-N adopts a folded\nconformation upon interaction with CENP-A, particularly involv-\ning the first 200 residues, we examined the combined dynamics\nof the ordered region (residues 1–200) and the disordered re-\ngion (residues 201–295) in CENP-N through RMSD calculation.\nAt first, we calculated the RMSD of whole CENP-N in its free and\nbound states, see Fig. 5c. The overall RMSD was significantly\nhigher for free CENP-N than compared to the bound CENP-N with\nthe average RMSD of 14.32±0.44 Å and 9.98±0.93Å respectively .\nAt the same time, ordered residues (res 1-200) show a decrease\nin RMSD from 11.18± 0.45Å to 5.57± 0.22 Å upon nucleosome\nbinding, whereas the flexibility of the disordered region remains\nsimilar in both conditions. To further assess the compactness of\nCENP-N considering residues in these different regions, we calcu-\nlated the corresponding Rg (see Fig. S7). Rg is slightly smaller\nin the bound state of CENP-N, however, the average Rg for all\nthree cases (entire CENP-N, the ordered region, and the disor-\ndered region), indicates less differences than the average RMSD.\nThus, while region-specific RMSD analysis reveals local stabiliza-\ntion of the N-terminal residues upon nucleosome binding, the\noverall Rg remains largely similar, indicating that CENP-N retains\nits global size and shape while undergoing local conformational\nadjustments.\n3.5 CENP-N binding free energy\nTo quantify the strength of the PPIs between the CENP-A pro-\ntein in the NCP and the modulating CENP-N protein, the binding\nenergy was calculated using the US method. Here, we consid-\n10 | 1–16\n+ P V S O B M \u0001 / B N F \r \u0001 < Z F B S > \r \u0001 < W P M \u000f > \r\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 10, 2025. ; https://doi.org/10.1101/2025.11.10.686994doi: bioRxiv preprint \n\n180°\nξ = 8.0 nm\nBound\nξ = 5.45 nm\n180°\nUnbound\nξ = 14.45 nm\n(a)\n(b)\n(c)\n(d)\n(a)\nFig. 6 (a) The PMF curve representing the binding free energy between the CENP-A and CENP-N proteins in the centromere NCP obtained using\nthe umbrella sampling method. The error bars calculated using Bayesian bootstrapping algorithm are depicted in grey. The 3D representation of the\nbinding site and the proteins involved is depicted using CG beads, along with the nucleosomal DNA depicted in white surface representation. Different\nstructures representing the COM reaction coordinate,ξ, of: (b)ξ = 5.45 nm, (c)ξ = 8.0 nm, and (d)ξ = 14.45 nm, are visualized for clarity. Bead\ncolors represent the relative RMSF fluctuations, where red beads indicate regions of CENP-A and CENP-N with the highest fluctuations, while blue\nbeads indicate the smallest fluctuations. The RMSF coloring was generated using a user-defined color scale in VMD91. Two views of the same system\nare depicted at the selected COM reaction coordinate.\nered the reaction coordinate, ξ, as the COM distance between the\nCENP-A protein and the COM of the ordered part of CENP-N (res\n1-200). The choice of the ordered region of CENP-N was guided\nby the RMSF and RMSD analyses depicted in Fig. 5a,c, which\nshowed that residues 201–295 remain disordered irrespective of\nnucleosome binding. Focusing on the ordered regions ensures\nthat the sampled conformational space reflects meaningful inter-\nactions and avoids contributions from highly flexible segments.\nThe resulting PMF plot along with the calculated error bars is\npresented in Fig. 6a. The binding energy of the complex can\nbe calculated from the difference between the highest and lowest\nvalues of the average PMF curve (details in the Method section).\nThus, the binding free energy between CENP-A and CENP-N is\nconcluded to be -7.92±0.99 kcal/mol.\nTo visualize the structural changes of CENP-A and CENP-N at\nthree different ξ values, the representative structures of the sys-\ntem calculated were further inspected. To do so, the CG beads\nof all residues in both proteins were visualized and colored ac-\ncording to their relative RMSF values, calculated from the re-\nstrained trajectories at the corresponding ξ value. A more in-\ntense red color indicates significant fluctuations of molecular en-\ntity representing the bead, while a more intense blue color rep-\nresents smaller fluctuations. At ξ = 5.45 nm, corresponding to\nthe bound state between the proteins, we observe higher fluctu-\nations in the disordered region of CENP-N that is not involved in\nthe direct protein-protein contact (Fig. 6b). At ξ = 8.0 nm, fluc-\ntuations at both terminals of CENP-N increase (Fig. 6c), while in\nthe unbound state (ξ = 14.45 nm) they display much stronger\nfluctuations (Fig. 6d). The unbound state of CENP-N essentially\nrepresents the free state of CENP-N, where a substantial portion\nof the protein exhibits a disordered nature. For CENP-A, fluctua-\ntions increase in the unbound state compared to the bound state,\nalthough the increase is less significant compared to CENP-N since\nthe protein is bounded inside the histone core. The higher struc-\ntural stability of both proteins upon their binding is clearly visible\nfrom this analysis and complements the dependencies observed\nin the unbiased MD simulations.\n4 Discussion\nThe plasticity of the centromeric NCP is a key factor in facilitat-\ning its function 92. Its interaction with binding proteins directs\nchromatin folding and the compaction of genetic information,\nprocesses that are essential for accurate chromosome segregation\nand the maintenance of genome stability 11. Here, we examine\nthe binding of CENP-N and the resulting structural changes in the\nNCP. Overall, our study investigates the PPIs within the NCP, with\n+ P V S O B M \u0001 / B N F \r \u0001 < Z F B S > \r \u0001 < W P M \u000f > \r\n1–16 | 11\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 10, 2025. ; https://doi.org/10.1101/2025.11.10.686994doi: bioRxiv preprint \n\na particular focus on how CENP-N binding influences the dynam-\nics and stability of the CENP-A nucleosome. Because nucleosome\ndynamics occur over extended timescales, we employed a CG-MD\nto capture these interactions and evaluate how the underlying\nprotein–protein associations contribute to nucleosome function-\nality .\nThe RMSD analysis of NCPs without histone tails reveals no\nsignificant structural differences upon the CENP-N binding with\nrather small structural stabilization of 0.22 Å (Fig. 2a), indi-\ncating that the core histone fold remains structurally stable ir-\nrespective of the CENP-N presence. However, when histone tails\nare included in this analysis, a modest increase in RMSD in the\npresence of CENP-N is observed, suggesting that the histone tails\nmay adopt more dynamic or altered conformations upon CENP-\nN binding. Interestingly , the radius of gyration remains largely\nunchanged across all comparisons, including those performed for\nhistone cores with and without tails and the DNA (Fig. 2b). This\nindicates that despite local flexibility changes, particularly in the\nhistone tails, the overall compactness of the nucleosome remains\nconstant. These results suggest that the role of the CENP-N pro-\ntein is not to further compact individual nucleosomes, but rather\nto facilitate inter-nucleosomal interactions that were reported to\ndrive centromeric chromatin folding and organization 29.\nStarting with the binding interface involving DNA bases, CENP-\nA amino acids and CENP-N amino acids reported in cryo-EM 29,\nwe performed several replica MD simulations each of 10 µs,\nwhich revealed stable residue-residue contacts between the NCP\nand CENP-N. PPIs between CENP-A and CENP-N were shown to\ndemonstrate a high specificity of the interaction interface en-\nriched in polar, charged, and aromatic side chains, suggesting\ndiverse binding interactions. The RG loop of CENP-A within its\nL1 loop, ARG80 and GLY81 emerge as central points of contact\nin the simulation based on the average contact values (see Tab.\nS2), which were also suggested by the cryo-EM study of Chit-\ntori et al. 25. Considering the chemical nature of the individ-\nual contributing residues, we found that ARG80 of CENP-A, with\nits positively charged guanidinium group, interacts strongly with\nASN145 of CENP-N and moderately with PRO144 and GLN146.\nThese residues possess polar or partially polar side chains, en-\nabling electrostatic and hydrogen bond-based interactions. Even\nif the SIRAH CG FF cannot properly treat subtle hydrogen bonds,\nit is known to adequately represent electrostatic interactions,\nwhich are essential in proper mimicking the PPIs of the complex\nstudied. Notably , GLY81 of CENP-A exhibits a high degree of flexi-\nbility and forms extensive contacts with ASN145, GLN146, PHE8,\nPRO144, and TYR147 of CENP-N, supporting the idea that this\nposition serves as a hub for both polar and hydrophobic inter-\nactions. THR79, a polar uncharged residue of CENP-A, interacts\nstrongly with THR4 of CENP-N, with an average contact value\nof 0.99 (see Table S2), likely forming a hydrogen-bond network\nvia their side-chain hydroxyl groups. Additionally , a hydropho-\nbic VAL82 participates in both polar (ASN145, GLN146) and aro-\nmatic (PHE8, TYR147) contacts with CENP-N, suggesting van der\nWaals and hydrophobic interactions may also contribute. More-\nover, a negatively charged ASP83 forms notable contacts with\npolar aromatic TYR147 (average contact value 0.55), which are\nlikely mediated through hydrogen bonding or electrostatic attrac-\ntion. These findings indicate that the binding interface between\nCENP-A and CENP-N is stabilized through a combination of di-\nverse interactions, which together contribute to a high PPIs speci-\nficity and strength of the CENP-A/CENP-N interaction, which was\ncalculated to be -7.92±0.99 kcal/mol based on umbrella sam-\npling simulations performed in this work. Interestingly , while\ncryo-EM studies25 have previously identified GLU3, THR7 as par-\nticipating in the interaction interface between CENP-N and the\nRG loop of CENP-A, the contact-based analysis performed based\non CG-MD simulations in the present work did not reveal stable\nor significant interactions involving these residues.\nThe interaction between the nucleosomal DNA and CENP-N\nwas also probed using a distance cutoff approach. The DNA bases,\nprimarily belonging to SHLs 2 to 3.5, and amino acids from CENP-\nN involved in the interaction are listed in Table S3. This obser-\nvation is consistent with previous experimental studies 25. The\nsimulation suggests that several arginine and lysine residues from\nCENP-N, such as ARG169, LYS81, LYS148, ARG44, and LYS45, in-\nteract with DNA, indicating that the interaction is predominantly\nelectrostatic. The distance-based interaction analysis also identi-\nfied additional amino acids likely to participate in DNA binding.\nFor example, LEU168 and VAL82 were detected to interact with\nDNA through hydrophobic contacts, while GLU46 and TYR147\nfrequently contact DNA bases (average contact value 0.76 and\n0.85 respectively), suggesting potential hydrogen-bond-like inter-\nactions in the coarse-grained representation. MET18 and ASN19\nare also recognized as potential DNA binding residues due to their\nability to form hydrophobic interactions and hydrogen bonds, re-\nspectively .\nTo monitor any allosteric changes in the histone core upon\nbinding of CENP-N, we focused on the conformational arrange-\nment of the CENP-A protein in the presence and absence of CENP-\nN. The RMSF analysis of CENP-A in Fig. 4a revealed sequence-\ndependent differences in the structural flexibility upon CENP-N\nbinding. The L1 loop region of CENP-A exhibits reduced RMSF\nin the presence of CENP-N, whereas in the absence of CENP-N,\nthis region displays markedly increased flexibility . This suggests\nthat binding of CENP-N confers structural stabilization to the L1\nloop, potentially through direct interactions. RMSF analysis fur-\nther shows that the N-terminal region of CENP-A exhibits higher\nflexibility when bound to CENP-N compared to its unbound state.\nThis suggests that interaction with CENP-N enhances the dynamic\nbehavior of the CENP-A N-terminus. In parallel, contact map base\nanalysis (Fig. S8) revealed a reduction in the average number of\ncontacts formed by residues 13–22 of CENP-A with histone H2A\nwhen CENP-N is present (Fig. S8a). Conversely , in the absence\nof CENP-N, this region engages in more persistent contacts with\nH2A (Fig. S8c) with participating residues marked within a circle.\nOverall, this study indicates that the binding of CENP-N alloster-\nically disrupts contacts between the CENP-A N-terminal tail and\nhistone H2A, thereby enhancing the conformational freedom of\nthe N-terminus.\nThe enhanced fluctuations of the CENP-A N-terminal tail (see\nFig. 4a) observed in our simulations suggest that this region relies\non interactions with specific binding partners for protein stabiliza-\n12 | 1–16\n+ P V S O B M \u0001 / B N F \r \u0001 < Z F B S > \r \u0001 < W P M \u000f > \r\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 10, 2025. ; https://doi.org/10.1101/2025.11.10.686994doi: bioRxiv preprint \n\ntion when bound to CENP-N. This dynamic behavior is consistent\nwith experimental observations93, highlighting the functional im-\nportance of the N-terminus of CENP-A for centromere functional-\nity and kinetochore assembly . For example, the N-terminus of\nCENP-A was shown to contribute to the stabilization of the cen-\ntromere binding protein CENP-B by direct interaction94,95. More-\nover, it acts as a recruiter of key kinetochore proteins such as\nCENP-C and CENP-T at both ectopic sites and endogenous cen-\ntromeres in Schizosaccharomyces pombe and human cells 96,97.\nThus, the enhanced fluctuations of the N-terminal tails of CENP-\nA, as captured by RMSF in Fig. 4a, may serve as a signature of its\nactive engagement in modulating interactions essential for cen-\ntromere function.\nFurthermore, CENP-A’s unstructured N-terminal tail bears post-\ntranslational modifications98. Several studies have demonstrated\nthat for instance phosphorylation of serine 7, 16 or 18 within the\nCENP-A N-terminal domain significantly influences centromeric\nchromatin structure and function 99–101. These residues lie near\nthe region that shows altered contact behavior in CG simulations\n(see residue 13 to 22 in Fig. S8), suggesting that CENP-N binding\nmay modulate the exposure or accessibility of phosphorylation\nsites, thus indirectly impacting downstream chromatin remodel-\ning or signaling events (see Fig. S8 a). Therefore, the observed\ndestabilization of local contacts (Fig. S8 a) and increased flexi-\nbility of the N-terminus of CENP-A (residue 13 to 22 in Fig. 4a)\nupon CENP-N binding is not only structurally plausible but also bi-\nologically meaningful, potentially contributing to regulating cen-\ntromere function via a modulation of post-translational modifica-\ntion, recruitment of kinetochore proteins, or higher-order chro-\nmatin structure formation.\nTo understand the molecular determinants of centromere as-\nsembly , it is essential to quantify the binding energetics of the\nCENP-A–CENP-N complex. Binding energy not only reflects over-\nall stability but also reveals how different regions of a multiva-\nlent protein contribute to interaction, providing mechanistic in-\nsight that cannot be captured by static structural measurements\nalone. Although, calculating binding energy experimentally for\nthis nucleosome-like system is challenging to date due to the mul-\ntivalent and context-dependent nature of their interactions.\nFinally , the calculated binding free energy between the CENP-A-\ncontaining NCP and the CENP-N protein from CG-MD simulations\nin our work captures the strength and complexity of the interac-\ntion (Fig. 6a), while further analyses of trajectories generated\nuncovers the split nature of CENP-N: the N-terminal domain be-\ncomes ordered upon binding to the NCP, whereas the C-terminal\nregion remains largely disordered (Fig. 6b-d). A similar observa-\ntion was also reflected during the independent CG-MD simulation\nof CENP-N in its bonded and a free state as discussed in Section\n3.4. RMSF base study supports the split nature of CENP-N (see\nFig. 5a), where N- and C-terminals of CENP-N remain disordered\nin the free state, while the N-terminal becomes ordered in the\nbound state with the NCP. Such a localized structural ordering\nin our simulations aligns well with experimental HX exchange\ndata that analyzed the structural dynamics of the N-terminal do-\nmain of CENP-N bound to CENP-A nucleosome, as well as in its\nfree state30. The experimental study reported substantial HX pro-\ntection throughout the N-terminal with up to 200 residues upon\nbinding to the CENP-A nucleosome, indicating a transition to a\nmore stable and ordered conformation. In our work, MD simula-\ntions show that the C-terminal region of CENP-N (Res 201-295)\nremains largely disordered, consistent with the experimental ob-\nservation that residues 209–240 are not involved in binding 25.\nNotably , the increased order in the bound state in our simulations\ndoes not involve major changes in the secondary structure (Fig.\n5b) for the whole protein, with no global alterations observed\nover multiple replicas. However, regional differences are evident:\nthe N-terminal segment (residues 1–50) exhibits increased stable\nsecondary structure upon binding, suggesting its role in mediat-\ning interactions with CENP-A and nucleosomal DNA. The mid-\ndle segment (residues 51–200) maintains a balanced distribution\nof structured and unstructured elements (see Table 2, showing\nminimal response to binding. By contrast, the C-terminal region\n(residues 201–295) is predominantly disordered in both states,\nconsistent with its reported higher flexibility . Despite the local\nchanges described, the overall Rg of CENP-N remains largely un-\nchanged (a decrease of around 1.07± 0.86 Å was obtained, see\nFig. S7), indicating that the global size and shape of the pro-\ntein are mostly maintained. Taken together, these findings indi-\ncate that binding of CENP-N to the NCP primarily stabilizes the\nN-terminal region of CENP-N, while the central and C-terminal\nregions remain largely unaffected. This study suggests that the\nbinding-induced conformational changes in CENP-N are driven\nby localized ordering rather than large-scale structural rearrange-\nments.\n5 Conclusions\nIn this study , we provide molecular insights into the interac-\ntion between the CENP-A nucleosome and one of its key bind-\ning partners, CENP-N, using a combination of biased and unbi-\nased molecular dynamics simulations. Notably , the binding in-\nterface predicted by cryo-EM studies remains conserved in the\ncoarse-grained simulations performed, linking structural and dy-\nnamical perspectives of this biological complex. In addition,\nCENP-N association enhances conformational fluctuations in the\nN-terminal region of CENP-A, a feature likely to facilitate fur-\nther protein–protein interactions essential for centromere assem-\nbly and function. A split nature of CENP-N was observed in unbi-\nased MD simulations, as well as upon binding to the nucleosome\ncore particle revealed by umbrella sampling method. The multi-\ntude of analyses and dependencies reported in this work supports\ndiverse experimental observations, thus demonstrating the SIRAH\ncoarse-grained force field reliably captures the long microsecond-\nscale dynamics of this centromeric complex.\nFinally , binding free energy analyses provided a quantitative\nperspective on CENP-A–CENP-N association in NCPs, revealing in-\nsights into promising directions for the regulation of centromere\nfunction through post-translational modifications, chromatin re-\nmodelers, or the rational engineering of synthetic centromeres.\nTogether, our findings provide a critical mechanistic understand-\ning in the characterization of protein-protein and protein-DNA in-\nteractions, deepening our perception of microscopic processes in\ncentromeres and offering a foundation for future studies aimed at\n+ P V S O B M \u0001 / B N F \r \u0001 < Z F B S > \r \u0001 < W P M \u000f > \r\n1–16 | 13\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 10, 2025. ; https://doi.org/10.1101/2025.11.10.686994doi: bioRxiv preprint \n\ntargeted control of centromere functionality . Subsequent studies\nmay further explore experimental measurements of binding ener-\ngetics to complement and validate the computational predictions\npresented here.\nAuthor contributions\nAbhik Ghosh Moulick: Writing – review and editing, Writing –\noriginal draft, Visualization, Validation, Software, Methodology ,\nInvestigation, Formal analysis, Data curation, Conceptualization.\nSylvia Erhardt: Writing – review and editing, Validation, Funding\nacquisition. Wolfgang Wenzel: Writing – review and editing, Re-\nsources, Funding acquisition. Mariana Kozlowska: Writing – re-\nview and editing, Supervision, Resources, Project administration,\nMethodology , Funding acquisition, Formal analysis, Conceptual-\nization.\nConflicts of interest\nThe authors declare no conflicts of interest.\nData availability\nData generated is available from the corresponding author upon\nreasonable request.\nAcknowledgements\nThis research was made possible by funding from the Carl-Zeiss-\nStiftung and Center SynGen. The authors gratefully acknowl-\nedge the computing time provided on the high-performance com-\nputer HoreKa by the National High-Performance Computing Cen-\nter at KIT (NHR@KIT). This center is jointly supported by the\nFederal Ministry of Education and Research and the Ministry of\nScience, Research and the Arts of Baden-Württemberg, as part of\nthe National High-Performance Computing (NHR) joint funding\nprogram (https://www.nhr-verein.de/en/our-partners). HoreKa\nis partly funded by the German Research Foundation (DFG).\nReferences\n1 R. D. Kornberg, Science, 1974, 184, 868–871.\n2 C. L. Woodcock, A. I. Skoultchi and Y. Fan, Chromosome Re-\nsearch, 2006, 14, 17–25.\n3 E. Segal, Y. Fondufe-Mittendorf, L. Chen, A. Thåström,\nY. Field, I. K. Moore, J.-P. Z. Wang and J. Widom, Nature,\n2006, 442, 772–778.\n4 K. Luger, A. W. Mäder, R. K. Richmond, D. F. Sargent and\nT. J. Richmond, Nature, 1997, 389, 251–260.\n5 G. Arya and T. Schlick, Proceedings of the National Academy\nof Sciences, 2006, 103, 16236–16241.\n6 J. Huertas and V. Cojocaru, Journal of molecular biology,\n2021, 433, 166744.\n7 G. Li and D. Reinberg, Current opinion in genetics & develop-\nment, 2011, 21, 175–186.\n8 S. Liu, A. Athreya, Z. Lao and B. Zhang, Annual review of\nbiophysics, 2024, 53, year.\n9 B. D. Strahl and C. D. Allis, Nature, 2000, 403, 41–45.\n10 M. S. Cosgrove, J. D. Boeke and C. Wolberger, Nature struc-\ntural & molecular biology, 2004, 11, 1037–1043.\n11 K. Luger, M. L. Dechassa and D. J. Tremethick,Nature reviews\nMolecular cell biology, 2012, 13, 436–447.\n12 G. Liu, H. Zhang and Y. Peng, Biocell, 2025, 49, 519.\n13 B. Brinkley and E. Stubblefield, Chromosoma, 1966, 19, 28–\n43.\n14 H. Nagpal and T. Fukagawa, Chromosoma, 2016, 125, 645–\n659.\n15 S. Nurk, S. Koren, A. Rhie, M. Rautiainen, A. V. Bzikadze,\nA. Mikheenko, M. R. Vollger, N. Altemose, L. Uralsky , A. Ger-\nshman et al., Science, 2022, 376, 44–53.\n16 N. Altemose, G. A. Logsdon, A. V. Bzikadze, P. Sidhwani,\nS. A. Langley , G. V. Caldas, S. J. Hoyt, L. Uralsky , F. D.\nRyabov, C. J. Shew et al., Science, 2022, 376, eabl4178.\n17 P. E. Warburton, C. A. Cooke, S. Bourassa, O. Vafa, B. A. Sul-\nlivan, G. Stetten, G. Gimelli, D. Warburton, C. Tyler-Smith,\nK. F. Sullivan et al., Current Biology, 1997, 7, 901–904.\n18 B. E. Black and D. W. Cleveland, Cell, 2011, 144, 471–479.\n19 K. L. McKinley and I. M. Cheeseman, Nature reviews Molecu-\nlar cell biology, 2016, 17, 16–29.\n20 J. S. Verdaasdonk and K. Bloom, Nature reviews Molecular\ncell biology, 2011, 12, 320–332.\n21 S. Yatskevich, K. W. Muir, D. Bellini, Z. Zhang, J. Yang, T. Tis-\ncher, M. Predin, T. Dendooven, S. H. McLaughlin and D. Bar-\nford, Science, 2022, 376, 844–852.\n22 M. E. Pesenti, T. Raisch, D. Conti, K. Walstein, I. Hoffmann,\nD. Vogt, D. Prumbaum, I. R. Vetter, S. Raunser and A. Musac-\nchio, Molecular cell, 2022, 82, 2113–2131.\n23 C. W. Carroll, K. J. Milks and A. F. Straight, Journal of Cell\nBiology, 2010, 189, 1143–1155.\n24 C. W. Carroll, M. C. Silva, K. M. Godek, L. E. Jansen and A. F.\nStraight, Nature cell biology, 2009, 11, 896–902.\n25 S. Chittori, J. Hong, H. Saunders, H. Feng, R. Ghirlando,\nA. E. Kelly , Y. Bai and S. Subramaniam, Science, 2018, 359,\n339–343.\n26 P. K. Allu, J. M. Dawicki-McKenna, T. Van Eeuwen,\nM. Slavin, M. Braitbard, C. Xu, N. Kalisman, K. Murakami\nand B. E. Black, Current Biology, 2019, 29, 2625–2639.\n27 K. Klare, J. R. Weir, F. Basilico, T. Zimniak, L. Massimiliano,\nN. Ludwigs, F. Herzog and A. Musacchio, Journal of Cell Bi-\nology, 2015, 210, 11–22.\n28 D. P. Melters, M. Bui, T. Rakshit, S. A. Grigoryev, D. Sturgill\nand Y. Dalal, Life Science Alliance, 2025, 8, year.\n29 K. Zhou, M. Gebala, D. Woods, K. Sundararajan, G. Edwards,\nD. Krzizike, J. Wereszczynski, A. F. Straight and K. Luger,\nNature structural & molecular biology, 2022, 29, 403–413.\n30 L. Y. Guo, P. K. Allu, L. Zandarashvili, K. L. McKinley ,\nN. Sekulic, J. M. Dawicki-McKenna, D. Fachinetti, G. A. Logs-\ndon, R. M. Jamiolkowski, D. W. Clevelandet al., Nature com-\nmunications, 2017, 8, 15775.\n31 H. Nagpal, A. Ali-Ahmad, Y. Hirano, W. Cai, M. Halic, T. Fuk-\nagawa, N. Sekuli ´c and B. Fierz, Nature Communications,\n2023, 14, 8227.\n32 S. Cao, K. Zhou, Z. Zhang, K. Luger and A. F. Straight,Molec-\nular biology of the cell, 2018, 29, 751–762.\n14 | 1–16\n+ P V S O B M \u0001 / B N F \r \u0001 < Z F B S > \r \u0001 < W P M \u000f > \r\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 10, 2025. ; https://doi.org/10.1101/2025.11.10.686994doi: bioRxiv preprint \n\n33 K. Chakraborty , M. Kang and S. M. Loverde, The Journal of\nPhysical Chemistry B, 2018, 122, 11827–11840.\n34 P. Khatua, P. Tang, A. G. Moulick, R. Patel, A. Manandhar\nand S. M. Loverde, Biophysical Journal, 2024, 123, 137a.\n35 D. Winogradoff and A. Aksimentiev, Journal of molecular bi-\nology, 2019, 431, 323–335.\n36 G. A. Armeev, A. S. Kniazeva, G. A. Komarova, M. P. Kirpich-\nnikov and A. K. Shaytan, Nature communications, 2021, 12,\n2387.\n37 J. Lequieu, D. C. Schwartz and J. J. de Pablo, Proceedings of\nthe National Academy of Sciences, 2017, 114, E9197–E9205.\n38 H. Kohestani and J. Wereszczynski, Biophysical Journal,\n2021, 120, 1498–1509.\n39 S. Li, T. Wei and A. Panchenko, Biophysical Journal, 2023,\n122, 218a.\n40 A. S. Fedulova, G. A. Armeev, T. A. Romanova, L. Singh-\nPalchevskaia, N. A. Kosarim, N. A. Motorin, G. A. Komarova\nand A. K. Shaytan, Wiley Interdisciplinary Reviews: Computa-\ntional Molecular Science, 2024, 14, e1728.\n41 G. N. Rychkov, A. V. Ilatovskiy , I. B. Nazarov, A. V. Shvetsov,\nD. V. Lebedev, A. Y. Konev, V. V. Isaev-Ivanov and A. V.\nOnufriev, Biophysical journal, 2017, 112, 460–472.\n42 X. Ding, X. Lin and B. Zhang, Nature communications, 2021,\n12, 1091.\n43 S. E. Farr, E. J. Woods, J. A. Joseph, A. Garaizar and\nR. Collepardo-Guevara, Nature communications, 2021, 12,\n2883.\n44 H. Kono, S. Sakuraba and H. Ishida, Biophysics and physico-\nbiology, 2019, 16, 337–343.\n45 H. Ishida and H. Kono, Proceedings of the National Academy\nof Sciences, 2021, 118, e2020452118.\n46 S. Bowerman and J. Wereszczynski, Biophysical journal,\n2016, 110, 327–337.\n47 J. Huertas, E. J. Woods and R. Collepardo-Guevara, Current\nOpinion in Cell Biology, 2022, 75, 102067.\n48 S. Portillo-Ledesma, Z. Li and T. Schlick, Current opinion in\nstructural biology, 2023, 78, 102506.\n49 M. Baaden and S. J. Marrink, Current opinion in structural\nbiology, 2013, 23, 878–886.\n50 T. Sun, V. Minhas, N. Korolev, A. Mirzoev, A. P. Lyubart-\nsev and L. Nordenskiöld, Frontiers in Molecular Biosciences,\n2021, 8, 645527.\n51 C. Hyeon and D. Thirumalai, Nature Communications, 2011,\n2, 487.\n52 G. Reddy and D. Thirumalai, Nucleic Acids Research, 2021,\n49, 4907–4918.\n53 D. Chakraborty , B. Mondal and D. Thirumalai, Journal of\nChemical Theory and Computation, 2024, 20, 1398–1413.\n54 U. Kapoor, Y. C. Kim and J. Mittal,Journal of Chemical Theory\nand Computation, 2024, 20, 1717–1731.\n55 T. Sun, V. Minhas, A. Mirzoev, N. Korolev, A. P. Lyubartsev\nand L. Nordenskiöld, Journal of Chemical Theory and Com-\nputation, 2022, 18, 3948–3960.\n56 D. A. Beard and T. Schlick, Structure, 2001, 9, 105–114.\n57 Z. Li, S. Portillo-Ledesma and T. Schlick, Biophysical journal,\n2023, 122, 2884–2897.\n58 J. Lequieu, A. Córdoba, J. Moller and J. J. De Pablo, The\nJournal of chemical physics, 2019, 150, year.\n59 M. R. Machado, E. E. Barrera, F. Klein, M. Sóñora, S. Silva\nand S. Pantano, Journal of chemical theory and computation,\n2019, 15, 2719–2733.\n60 L. Darré, M. R. Machado, A. F. Brandner, H. C. González,\nS. Ferreira and S. Pantano, Journal of chemical theory and\ncomputation, 2015, 11, 723–739.\n61 P. C. Souza, R. Alessandri, J. Barnoud, S. Thallmair,\nI. Faustino, F. Grünewald, I. Patmanidis, H. Abdizadeh, B. M.\nBruininks, T. A. Wassenaar et al., Nature methods, 2021, 18,\n382–388.\n62 J. S. Patel and F. M. Ytreberg, Journal of Chemical Theory\nand Computation, 2018, 14, 991–997.\n63 S. Pusara, W. Wenzel and M. Kozlowska, International Jour-\nnal of Biological Macromolecules, 2024, 263, 130365.\n64 M. R. Machado and S. Pantano, Journal of chemical theory\nand computation, 2015, 11, 5012–5023.\n65 A. Brandner, A. Schüller, F. Melo and S. Pantano,Biochemical\nand Biophysical Research Communications, 2018, 498, 319–\n326.\n66 A. Ghosh Moulick, R. Patel, A. Onyema and S. M. Loverde,\nThe Journal of Chemical Physics, 2025, 162, 0021–9606.\n67 E. E. Barrera, M. R. Machado and S. Pantano, Journal of\nChemical Theory and Computation, 2019, 15, 5674–5688.\n68 M. R. Machado, H. C. González and S. Pantano, Journal of\nChemical Theory and Computation, 2017, 13, 5106–5116.\n69 M. R. Machado, A. Zeida, L. Darré and S. Pantano, Interface\nFocus, 2019, 9, 20180085.\n70 P. D. Dans, A. Zeida, M. R. Machado and S. Pantano, Journal\nof chemical theory and computation, 2010, 6, 1711–1725.\n71 J. Abramson, J. Adler, J. Dunger, R. Evans, T. Green,\nA. Pritzel, O. Ronneberger, L. Willmore, A. J. Ballard, J. Bam-\nbrick et al., Nature, 2024, 630, 493–500.\n72 K. Lindorff-Larsen, S. Piana, K. Palmo, P. Maragakis, J. L.\nKlepeis, R. O. Dror and D. E. Shaw,Proteins: Structure, Func-\ntion, and Bioinformatics, 2010, 78, 1950–1958.\n73 P. Mark and L. Nilsson, The Journal of Physical Chemistry A,\n2001, 105, 9954–9960.\n74 D. Van Der Spoel, E. Lindahl, B. Hess, G. Groenhof, A. E.\nMark and H. J. Berendsen, Journal of computational chem-\nistry, 2005, 26, 1701–1718.\n75 T. J. Dolinsky , J. E. Nielsen, J. A. McCammon and N. A.\nBaker, Nucleic acids research, 2004, 32, W665–W667.\n76 L. Darré, M. R. Machado, P. D. Dans, F. E. Herrera and\nS. Pantano, Journal of Chemical Theory and Computation,\n2010, 6, 3793–3807.\n77 G. Bussi, D. Donadio and M. Parrinello, The Journal of chem-\nical physics, 2007, 126, 0021–9606.\n78 M. Parrinello and A. Rahman, Journal of Applied physics,\n1981, 52, 7182–7190.\n79 M. R. Machado and S. Pantano, Bioinformatics, 2016, 32,\n+ P V S O B M \u0001 / B N F \r \u0001 < Z F B S > \r \u0001 < W P M \u000f > \r\n1–16 | 15\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 10, 2025. ; https://doi.org/10.1101/2025.11.10.686994doi: bioRxiv preprint \n\n1568–1570.\n80 J. Parsons, J. B. Holmes, J. M. Rojas, J. Tsai and C. E. Strauss,\nJournal of computational chemistry, 2005, 26, 1063–1068.\n81 J. A. Maier, C. Martinez, K. Kasavajhala, L. Wickstrom, K. E.\nHauser and C. Simmerling, Journal of chemical theory and\ncomputation, 2015, 11, 3696–3713.\n82 D. A. Case, H. M. Aktulga, K. Belfon, D. S. Cerutti, G. A.\nCisneros, V. W. D. Cruzeiro, N. Forouzesh, T. J. Giese, A. W.\nGotz, H. Gohlke et al., Journal of chemical information and\nmodeling, 2023, 63, 6183–6191.\n83 P. Smardz, M. M. Anila, P. Rogowski, M. S. Li, B. Ró˙zycki and\nP. Krupa, International Journal of Molecular Sciences, 2024,\n25, 6698.\n84 N. Michaud-Agrawal, E. J. Denning, T. B. Woolf and O. Beck-\nstein, Journal of computational chemistry, 2011, 32, 2319–\n2327.\n85 R. J. Gowers, M. Linke, J. Barnoud, T. J. E. Reddy , M. N.\nMelo, S. L. Seyler, J. Domanski, D. L. Dotson, S. Buchoux,\nI. M. Kenney et al., MDAnalysis: a Python package for the\nrapid analysis of molecular dynamics simulations, Los alamos\nnational laboratory (lanl), los alamos, nm (united states)\ntechnical report, 2019.\n86 J. Kästner, Wiley Interdisciplinary Reviews: Computational\nMolecular Science, 2011, 1, 932–942.\n87 S. Kumar, J. M. Rosenberg, D. Bouzida, R. H. Swendsen and\nP. A. Kollman, Journal of computational chemistry, 1992, 13,\n1011–1021.\n88 J. S. Hub, B. L. De Groot and D. Van Der Spoel, Journal of\nchemical theory and computation, 2010, 6, 3713–3720.\n89 J. Taylor, Introduction to error analysis, the study of uncer-\ntainties in physical measurements , University Science Books,\n1997.\n90 S. Pentakota, K. Zhou, C. Smith, S. Maffini, A. Petrovic, G. P.\nMorgan, J. R. Weir, I. R. Vetter, A. Musacchio and K. Luger,\nElife, 2017, 6, e33442.\n91 W. Humphrey , A. Dalke and K. Schulten,Journal of molecular\ngraphics, 1996, 14, 33–38.\n92 H. Nagpal and B. Fierz, Journal of molecular biology, 2021,\n433, 166676.\n93 K. M. Sathyan, D. Fachinetti and D. R. Foltz, Nature Commu-\nnications, 2017, 8, 14678.\n94 D. Fachinetti, H. Diego Folco, Y. Nechemia-Arbely , L. P. Va-\nlente, K. Nguyen, A. J. Wong, Q. Zhu, A. J. Holland, A. Desai,\nL. E. Jansen et al., Nature cell biology, 2013, 15, 1056–1066.\n95 D. Fachinetti, J. S. Han, M. A. McMahon, P. Ly , A. Abdullah,\nA. J. Wong and D. W. Cleveland, Developmental cell, 2015,\n33, 314–327.\n96 G. A. Logsdon, E. J. Barrey , E. A. Bassett, J. E. DeNizio, L. Y.\nGuo, T. Panchenko, J. M. Dawicki-McKenna, P. Heun and\nB. E. Black, Journal of Cell Biology, 2015, 208, 521–531.\n97 H. D. Folco, C. S. Campbell, K. M. May , C. A. Espinoza,\nK. Oegema, K. G. Hardwick, S. I. Grewal and A. Desai, Cur-\nrent Biology, 2015, 25, 348–356.\n98 C. A. Musselman, M.-E. Lalonde, J. Côté and T. G. Ku-\ntateladze, Nature structural & molecular biology, 2012, 19,\n1218–1227.\n99 A. O. Bailey , T. Panchenko, J. Shabanowitz, S. M. Lehman,\nD. L. Bai, D. F. Hunt, B. E. Black and D. R. Foltz,Molecular &\nCellular Proteomics, 2016, 15, 918–931.\n100 N. Kunitoku, T. Sasayama, T. Marumoto, D. Zhang,\nS. Honda, O. Kobayashi, K. Hatakeyama, Y. Ushio, H. Saya\nand T. Hirota, Developmental cell, 2003, 5, 853–864.\n101 A. O. Bailey , T. Panchenko, K. M. Sathyan, J. J. Petkowski, P.-\nJ. Pai, D. L. Bai, D. H. Russell, I. G. Macara, J. Shabanowitz,\nD. F. Hunt et al., Proceedings of the National Academy of Sci-\nences, 2013, 110, 11827–11832.\n16 | 1–16\n+ P V S O B M \u0001 / B N F \r \u0001 < Z F B S > \r \u0001 < W P M \u000f > \r\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted November 10, 2025. ; https://doi.org/10.1101/2025.11.10.686994doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}