A balance between elastic and rigidified chromatin states regulates transcriptional plasticity at human centromeres

Histone variants fine-tune the regulation of transcription, replication, DNA damage repair, and cell division relies on distinct chromatin states. The histone H3 variant CENP-A/CENH3 seeds the kinetochore, creating the physical interface between centromeric chromatin and mitotic spindles. How kinetochore proteins modify CENP-A nucleosome dynamics and how these dynamics affect centromere chromatin states is poorly understood. Using interdisciplinary analyses, we report that CENP-A nucleosomes are intrinsically elastic, but CENP-C binding suppresses this innate elasticity. Shifting the balance between elastic and rigid CENP-A states in vivo results in the suppression of centromeric chromatin plasticity, so that centromeric chromatin becomes less permissive to RNA polymerase 2, thereby diminishing new CENP-A loading. These data suggest a link between innate structural properties possessed by histone variant nucleosomes, and adaptability of chromatin states in vivo, which in turn dictate the transcriptional plasticity of the underlying locus.

nucleosomes, and adaptability of chromatin states in vivo, which in turn dictate the transcriptional plasticity of the underlying locus.

Introduction
The adaptive nature of chromatin states allows a cell to replicate, divide, differentiate, transcriptionally respond to various stimuli, and repair damaged DNA 1,2 . In part, this dynamic chromatin landscape is shaped by removing old and incorporating new nucleosomes 3,4 , by incorporating specific histone variants 5 , and by incorporating covalent modifications [6][7][8] . How different histone variants convey unique biophysical properties of their nucleosomes to the chromatin fiber, and whether such non-canonical nucleosomes modulate chromatin dynamics remains a subject of intense studies.
One of the most striking cellular events is mitosis, when chromosomes condense into rod-shaped structures, temporarily yet dramatically changing the transcriptional landscape 9,10 . Chromosome segregation is a mechanical process, where chromosomes are actively pulled from the metaphase plate towards the poles 11 . This process relies, in part, on the presence of specialized centromeric nucleosomes. Epigenetically the centromere is marked by the enrichment of the histone H3 variant CENP-A/CENH3 [12][13][14][15][16] . Despite lack of sequence conservation at the level of CENP-A or its associated DNA 12,17 , in most species, CENP-A chromatin provides the foundation by recruiting a triad of inner kinetochore proteins: CENP-B, CENP-C, and CENP-N [18][19][20][21] . Deleting either CENP-A or CENP-C results in cell death or induces senescence 22,23 . This happens only after a few cell cycles, suggesting that both CENP-A and CENP-C are likely present in excess over that required to form a functional kinetochore for one cell cycle. While CENP-A and CENP-C are long-lived proteins, guaranteeing faithful chromosome segregation even after their genes have been deleted 24-28 , incorporation of new CENP-A in human cells depends strongly on the transcription of centromeres at the end of mitosis/early G1 27,29,30 . A major paradox is how active transcription, which normally requires accessible chromatin, is accomplished at a time when kinetochore-bound centromeric chromatin is engaged in completion of mitosis.
Consequently, elucidating biophysical features of the inner kinetochore-associated chromatin, composed of CENP-A nucleosomes and its closest bound partners such as CENP-C, remains a fundamental biological question.
To investigate these questions, in this report, we used in silico, in vitro, and in vivo tools to dissect the dynamic nature of CENP-A nucleosomes compared to H3 nucleosomes, either with, or without CENP-C. Using all-atom molecular dynamic simulations, we report that when a CENP-A nucleosome is bound to CENP-C, its local and global flexibilities are severely limited, so that CENP-C fixes specific conformational states of a CENP-A nucleosome. To experimentally test the global changes of nucleosome dynamics, we directly measured the elasticity of CENP-A nucleosomes, finding to our surprise, that they are twice as elastic as H3 nucleosomes. Remarkably, upon CENP-C binding, CENP-A nucleosomes markedly rigidify by three-fold and cause three-dimensional compaction of centromeric chromatin fibers. In vivo, we demonstrate that CENP-A chromatin plasticity is required for recruitment of the transcriptional machinery. Indeed, overexpression of CENP-C led to overcompaction of centromeric CENP-A chromatin, reduced the levels of RNA polymerase 2 (RNAP2) occupancy at centromeres, which we show is concomitant with a reduction of de novo CENP-A loading in early G1. These data support a model in which a balance between elastic and rigidified centromeric nucleosomes regulates the plasticity and fidelity of the centromeric chromatin fiber in vivo.

CENP-A nucleosomes are highly elastic compared to H3 nucleosomes
Previous computational results found that CENP-A nucleosomes are conformationally distortable compared to canonical H3 nucleosomes 31 . One prediction from these experiments was that nucleosome distortability would correlate with nucleosome elasticity; in other words, the more deformable a nucleosome, the more likely it is to be elastic. Elasticity of materials is measured by the ratio of stress (N/m 2 or Pascal) to strain (indentation), known as the Young's modulus. In physics and biology, nano-indentation experiments are a well-accepted means of measuring the elastic properties of biological materials 32-34 .
Despite the longstanding use of nanomechanical force spectroscopy, we were surprised to discover that the elasticity of nucleosomes of any kind, has never been reported. Therefore, we performed in-fluid, single-molecule, nano-indentation force spectroscopy 35,36 of canonical H3 and variant CENP-A nucleosomes within in vitro reconstituted arrays under physiological conditions ( Figure 1A). As reported earlier 37,38 , we found that in vitro reconstituted CENP-A nucleosomes possess dimensions similar to H3 nucleosomes (3.7±0.3 and 3.8±0.3 nm, resp.) (Table S1). Next, we performed in-fluid nano-indentation on these reconstituted nucleosomes ( Figure 1A,B). H3 nucleosomes had a Young's modulus of 11.1±4.5 MPa. Strikingly, CENP-A nucleosomes were nearly twice as elastic (6.2±3.9 MPa, Figure 1B-D, Table S2) as H3 nucleosomes. showing that CENP-A nucleosomes are more elastic than H3 nucleosomes (ANOVA test P<0.0001).

Modeling CENP-A:CENP-C CD nucleosomes predicts a change in conformational flexibility
These elasticity measurements are compatible with previous computational modeling results, which demonstrated that CENP-A nucleosomes have an intrinsically more distortable B.
D. In another series of experiments, a retroviral gene trap insertion of H3f3a created a hypomorphic mutation. The resulting mutant mice were indistinguishable from wild-type mice at birth, but nevertheless 50% died within 24 hours. Surviving mutant mice displayed retarded growth, impaired neuromuscular activity, and reduced fertility [38], pointing to the importance of the H3f3a in maintaining proper cellular activity. The phenotype for the H3f3b knockout was even equally severe, with 50% of H3f3b knock-out embryos dying during the second half of embryogenesis. Most of these embryos exhibited abnormal development indicative of a broad failure of embryonic growth [39]. An even more dramatic phenotype was observed when both H3.3s were knocked-down by morpholinos [40], or with siRNAs [40]. Morpholino disruption of H3.3 in Xenopus resulted in defects in late gastrulation, a phenotype mimicked by knock-down of the H3.3 chaperone HIRA (Table 1) [40]. Knock-down of both H3.3 genes in mouse oocytes resulted in arrest in early blastocyte stage. This phenotype is exclusively dependent on the maternal H3.3 pool to regulate the reactivation of imprinted genes in both the maternal and paternal genome [41], since the paternal genome has not yet been activated. Finally, a role for H3.3 in establishing heterochromatin at endogenous retroviral elements in mouse embryonic stem cells has been shown [42]. Altogether, these targeted gene disruption studies emphasize the importance of H3.3 in regulating various stages of development.
Independent of its importance in development, in slow dividing or non-replicative cells, H3.3 also accumulates at transcribed regions and sites of DNA repair [43]. Not only is H3.3 enriched at these genomic regions, it can also induce senescence together with its cleaved version (1-21 aa), which is incorporated into the chromatin by the chaperone HUCA complex, and subsequently represses the transcription of cell cycle regulators, presumably due to the loss of N-terminal modifications [43]. In another series of experiments, a retroviral gene trap insertion of H3f3a created a hypomorphic mutation. The resulting mutant mice were indistinguishable from wild-type mice at birth, but nevertheless 50% died within 24 hours. Surviving mutant mice displayed retarded growth, impaired neuromuscular activity, and reduced fertility [38], pointing to the importance of the H3f3a in maintaining proper cellular activity. The phenotype for the H3f3b knockout was even equally severe, with 50% of H3f3b knock-out embryos dying during the second half of embryogenesis. Most of these embryos exhibited abnormal development indicative of a broad failure of embryonic growth [39]. An even more dramatic phenotype was observed when both H3.3s were knocked-down by morpholinos [40], or with siRNAs [40]. Morpholino disruption of H3.3 in Xenopus resulted in defects in late gastrulation, a phenotype mimicked by knock-down of the H3.3 chaperone HIRA (Table 1) [40]. Knock-down of both H3.3 genes in mouse oocytes resulted in arrest in early blastocyte stage. This phenotype is exclusively dependent on the maternal H3.3 pool to regulate the reactivation of imprinted genes in both the maternal and paternal genome [41], since the paternal genome has not yet been activated. Finally, a role for H3.3 in establishing heterochromatin at endogenous retroviral elements in mouse embryonic stem cells has been shown [42]. Altogether, these targeted gene disruption studies emphasize the importance of H3.3 in regulating various stages of development.
Independent of its importance in development, in slow dividing or non-replicative cells, H3.3 also accumulates at transcribed regions and sites of DNA repair [43]. Not only is H3.3 enriched at these genomic regions, it can also induce senescence together with its cleaved version (1-21 aa), which is incorporated into the chromatin by the chaperone HUCA complex, and subsequently represses the transcription of cell cycle regulators, presumably due to the loss of N-terminal modifications [43].

CENP-A H3
F nucleosome core, compared to H3 nucleosomes 31 . The corresponding free energy landscape predicts the existence of multiple conformational states of CENP-A. Next, we performed allatom explicit-solvent molecular dynamics simulations with, or without the central domain fragment of CENP-C (CENP-C CD ) [39][40][41] , from which we obtained the free energy landscapes of CENP-A nucleosomes. Strikingly, in the presence of CENP-C CD , the otherwise rugged free energy landscape of CENP-A nucleosomes collapses into just two broad basins (Figure 2A), with a distribution similar to that of H3 31 . This change in the free energy landscape manifested itself in the loss of the bimodal distribution of the movements of the center of mass ( Figure S1A). Furthermore, local structural flexibility was also suppressed upon CENP-C CD binding ( Figure   S1B). Overall, these findings indicate that CENP-C CD limits the conformational distortability and motions of CENP-A nucleosomes.
Next, we computationally examined the effect of doubling the amount of CENP-C per CENP-A nucleosome. Compared to a single CENP-C CD fragment, binding two CENP-C CD fragments globally reduced whole histone motions and local residue fluctuations of CENP-A nucleosomes ( Figure S2A, B). We also assessed DNA gyre sliding and gaping motions within the CENP-A nucleosome. We modeled these motions using the same residues as in previous smFRET experiments 33 . In support of these published experimental data, our high-resolution analysis showed that a single CENP-C CD fragment dampened the CENP-A nucleosome gyre gaping; where DNA slides asymmetrically away from the CENP-C bound-face of CENP-A nucleosomes.
One prediction from these modeling data is that increasing CENP-C concentration should dampen CENP-A nucleosomal plasticity in a dose-dependent manner.

CENP-C CD rigidifies CENP-A nucleosomes
We were curious whether CENP-C CD would suppress CENP-A nucleosomal elasticity. First, we measured the dimensions of CENP-C CD bound to CENP-A nucleosomes, finding that they are slightly taller than CENP-A nucleosomes alone (3.7±0.3 nm vs. 4.1±0.4 nm) (Table S1). Next, we measured the Young's modulus. About half the CENP-A nucleosomes remained highly elastic (5 MPa), whereas the other half lost elasticity by a factor of three (14.5 MPa) (1-way ANOVA P<0.0001; Figure 2C-E, Table S2). This bimodal distribution of the CENP-A+CENP-C CD population most likely represents two distinct CENP-A sub-species: one free (5 MPa), and the other bound to CENP-C CD (14.5 MPa). When we doubled the amount of CENP-C CD , virtually all CENP-A nucleosomes lose elasticity, and become rigidified (17.1±10.6 MPa, Figure   1D,E, Table S2) in a dose-dependent manner.   fertility, the basis of which is not fully understood [37]. H3f3a is expressed ubiquitously during mouse embryonic development until day E13.5, as well as adult heart, kidney, brain, testes, and ovaries [38].
In another series of experiments, a retroviral gene trap insertion of H3f3a created a hypomorphic mutation. The resulting mutant mice were indistinguishable from wild-type mice at birth, but nevertheless 50% died within 24 hours. Surviving mutant mice displayed retarded growth, impaired neuromuscular activity, and reduced fertility [38], pointing to the importance of the H3f3a in maintaining proper cellular activity. The phenotype for the H3f3b knockout was even equally severe, with 50% of H3f3b knock-out embryos dying during the second half of embryogenesis. Most of these embryos exhibited abnormal development indicative of a broad failure of embryonic growth [39]. An even more dramatic phenotype was observed when both H3.3s were knocked-down by morpholinos [40], or with siRNAs [40]. Morpholino disruption of H3.3 in Xenopus resulted in defects in late gastrulation, a phenotype mimicked by knock-down of the H3.3 chaperone HIRA (Table 1) [40]. Knock-down of both H3.3 genes in mouse oocytes resulted in arrest in early blastocyte stage. This phenotype is exclusively dependent on the maternal H3.3 pool to regulate the reactivation of imprinted genes in both the maternal and paternal genome [41], since the paternal genome has not yet been activated. Finally, a role for H3.3 in establishing heterochromatin at endogenous retroviral elements in mouse embryonic stem cells has been shown [42]. Altogether, these targeted gene disruption studies emphasize the importance of H3.3 in regulating various stages of development. Independent of its importance in development, in slow dividing or non-replicative cells, H3.3 also accumulates at transcribed regions and sites of DNA repair [43]. Not only is H3.3 enriched at these genomic regions, it can also induce senescence together with its cleaved version (1-21 aa), which is incorporated into the chromatin by the chaperone HUCA complex, and subsequently represses the transcription of cell cycle regulators, presumably due to the loss of N-terminal modifications [43]. fertility, the basis of which is not fully understood [37]. H3f3a is expressed ubiquitously during mouse embryonic development until day E13.5, as well as adult heart, kidney, brain, testes, and ovaries [38].
In another series of experiments, a retroviral gene trap insertion of H3f3a created a hypomorphic mutation. The resulting mutant mice were indistinguishable from wild-type mice at birth, but nevertheless 50% died within 24 hours. Surviving mutant mice displayed retarded growth, impaired neuromuscular activity, and reduced fertility [38], pointing to the importance of the H3f3a in maintaining proper cellular activity. The phenotype for the H3f3b knockout was even equally severe, with 50% of H3f3b knock-out embryos dying during the second half of embryogenesis. Most of these embryos exhibited abnormal development indicative of a broad failure of embryonic growth [39]. An even more dramatic phenotype was observed when both H3.3s were knocked-down by morpholinos [40], or with siRNAs [40]. Morpholino disruption of H3.3 in Xenopus resulted in defects in late gastrulation, a phenotype mimicked by knock-down of the H3.3 chaperone HIRA (Table 1) [40]. Knock-down of both H3.3 genes in mouse oocytes resulted in arrest in early blastocyte stage. This phenotype is exclusively dependent on the maternal H3.3 pool to regulate the reactivation of imprinted genes in both the maternal and paternal genome [41], since the paternal genome has not yet been activated. Finally, a role for H3.3 in establishing heterochromatin at endogenous retroviral elements in mouse embryonic stem cells has been shown [42]. Altogether, these targeted gene disruption studies emphasize the importance of H3.3 in regulating various stages of development. Independent of its importance in development, in slow dividing or non-replicative cells, H3.3 also accumulates at transcribed regions and sites of DNA repair [43]. Not only is H3.3 enriched at these genomic regions, it can also induce senescence together with its cleaved version (1-21 aa), which is incorporated into the chromatin by the chaperone HUCA complex, and subsequently represses the transcription of cell cycle regulators, presumably due to the loss of N-terminal modifications [43]. fertility, the basis of which is not fully understood [37]. H3f3a is expressed ubiquitously during mouse embryonic development until day E13.5, as well as adult heart, kidney, brain, testes, and ovaries [38].
In another series of experiments, a retroviral gene trap insertion of H3f3a created a hypomorphic mutation. The resulting mutant mice were indistinguishable from wild-type mice at birth, but nevertheless 50% died within 24 hours. Surviving mutant mice displayed retarded growth, impaired neuromuscular activity, and reduced fertility [38], pointing to the importance of the H3f3a in maintaining proper cellular activity. The phenotype for the H3f3b knockout was even equally severe, with 50% of H3f3b knock-out embryos dying during the second half of embryogenesis. Most of these embryos exhibited abnormal development indicative of a broad failure of embryonic growth [39]. An even more dramatic phenotype was observed when both H3.3s were knocked-down by morpholinos [40], or with siRNAs [40]. Morpholino disruption of H3.3 in Xenopus resulted in defects in late gastrulation, a phenotype mimicked by knock-down of the H3.3 chaperone HIRA (Table 1) [40]. Knock-down of both H3.3 genes in mouse oocytes resulted in arrest in early blastocyte stage. This phenotype is exclusively dependent on the maternal H3.3 pool to regulate the reactivation of imprinted genes in both the maternal and paternal genome [41], since the paternal genome has not yet been activated. Finally, a role for H3.3 in establishing heterochromatin at endogenous retroviral elements in mouse embryonic stem cells has been shown [42]. Altogether, these targeted gene disruption studies emphasize the importance of H3.3 in regulating various stages of development. Independent of its importance in development, in slow dividing or non-replicative cells, H3.3 also accumulates at transcribed regions and sites of DNA repair [43]. Not only is H3.3 enriched at these genomic regions, it can also induce senescence together with its cleaved version (1-21 aa), which is incorporated into the chromatin by the chaperone HUCA complex, and subsequently represses the transcription of cell cycle regulators, presumably due to the loss of N-terminal modifications [43].  Indeed, in vitro, we noticed a qualitative increase in clustering of reconstituted CENP-A nucleosome when exposed to CENP-C. This phenomenon occurred in a dose-dependent manner ( Figure S5A). We were curious to know whether we could induce CENP-A chromatin compaction simply by adding recombinant CENP-C CD fragment to kinetochore-depleted CENP-A chromatin fraction. We purified bulk CENP-A chromatin from human cells and incubated these samples with our CENP-C for 30 minutes, followed by analysis of clustering. We observed a ~30% increase in chromatin compaction upon the addition of CENP-C CD ( Figure 3B, Table   S3). One logical outcome from these results is that excess CENP-C would encode a centromeric fiber that is less permissive to chromatin binding factors, such as the transcriptional machinery.
We tested this hypothesis by overexpressing CENP-C in vivo for three days, after which we purified kinetochore-associated CENP-A chromatin and kinetochore-depleted CENP-A chromatin by serial N-ChIP ( Figure 3C). First, we assessed whether overexpression of CENP-C would also induce CENP-A chromatin compaction in vivo. We purified kinetochore-depleted CENP-A chromatin (second ACA N-ChIP) and measured compaction as above. We observed a doubling of compacted chromatin states relative to controls ( Figure 3D, Table S3). Next, we measured the nucleosomal dimensions of the kinetochore-associated CENP-A nucleosomes, and of kinetochore-depleted CENP-A nucleosomes. In cells overexpressing CENP-C, CENP-A nucleosomes displayed a marked increase in particle height, whereas kinetochore-associated CENP-A nucleosomes did not display a change in particle height (2.0±0.5 nm vs 3.5±0.8 nm, and 2.7±1.0 nm vs 2.7±1.2 nm resp. Figure S5B, Table S4). These data suggest that in the CENP-C overexpression background, there is not only enhanced clustering and compaction of CENP-A chromatin, but also a general shift towards suppression of plasticity of individual CENP-A nucleosomes.

CENP-C overexpression suppresses RNA polymerase 2 occupancy and de novo CENP-A loading
During late mitosis and early G1, RNAP2 is present at the centromere 43 . We wondered whether overexpression of CENP-C, and thus the induction of more compacted CENP-A chromatin, would lead to reduced accessibility for RNAP2. Indeed, by western blot analysis, when CENP-C is overexpressed we observed a significant reduction in RNAP2 at both CENP-A domains (3and 2-fold reduction, resp.; t-test p<0.05; Figure 3D, Table S5). Work from several labs have recently suggested that transcription of centromeric DNA is required for de novo CENP-A loading 43 . In this scenario, overexpression of CENP-C, which suppressed RNAP2 at centromeres (Figure 3E), should also lead to reduced de novo CENP-A loading. An initial clue leading to this possibility was already gleaned from our initial western  blot analysis, in which overexpression of CENP-C led to a significant reduction in the kinetochore-depleted CENP-A population (t-test p<0.05; Figure 3E, Table S5). We wanted test the idea that this reduction in CENP-A levels might have arisen from reduced de novo CENP-A loading. Therefore, we turned to the well-established SNAP-tagged CENP-A system combined with quench pulse-chase immunofluorescence 27 to track de novo integrated CENP-A. Strikingly, in the CENP-C overexpression background, we observed a 2.3-fold reduction of de novo incorporation of CENP-A (t-test p<0.01; Figure 4, Table S6).
Taken together, these data suggest elastic CENP-A nucleosomes create an intrinsically open chromatin environment, which is rigidified by CENP-C.

Discussion
Previous computational modeling experiments suggested that nucleosomes containing the CENP-A can intrinsically sample altered conformations and are structurally "frustrated" compared to canonical H3 nucleosomes 31 . In a plasticine vs. rock model, we predicted that a structurally "frustrated" nucleosome would manifest itself in distinctly different elasticity.
Indeed, by novel single molecule nano-indentation force spectroscopy, CENP-A nucleosomes are twice as elastic as canonical nucleosomes (Figure 1).
FRET experiments of in vitro reconstituted CENP-A mononucleosomes showed restricted DNA gyre gapping and sliding 33 . These data suggested that kinetochore components could choose and fix one or a few specific conformational states. Indeed, when we modeled CENP-A nucleosomes alone, vs. those bound to CENP-C CD , we observed both, a marked diminution of motion, and free energy minima, representing lost conformational flexibility (Figure 2A,B, S1, S2). The diminution of conformational flexibility correlates with a loss of elasticity of CENP-A nucleosomes upon CENP-C CD binding ( Figure 2D,E). Our hypothesis is that elasticity of CENP-A contributes to the accessibility of the centromeric chromatin fiber, potentially by allowing nucleosomes to deform or slide more easily. Indeed, in support of this idea, overexpression of CENP-C resulted in centromere chromatin compaction and decreased localization of RNAP2 at CENP-A chromatin, which correlates with the loss of de novo CENP-A loading (Figure 3, 4). In a working model based on these data, a balance between kinetochore bound rigid CENP-A nucleosomes, and unbound elastic CENP-A nucleosomes Not all nucleosomes are identical, as many contain histone variants, giving them distinct functions. In this report, we demonstrate how a single histone variant can alter the intrinsic biophysical properties of a nucleosome, which can be over-ridden by its cognate protein partners, thereby impacting the structural and functional state of the resulting chromatin fiber. An ongoing extension of this work is to test whether partnering with alternative histone variants, such as that reported in cancer cells 41,[43][44][45][46] , impacts CENP-A nucleosomal elasticity. In this context, testing whether suppressing such cancer-specific partnerships changes the outcome at ectopically occupied loci, most significantly, at neocentromere like domains 41,46 is an outstanding avenue of future investigation.
We note that centromeric DNA and centromeric protein genes are rapidly evolving [12][13][14][15][16][17]47 . Not all species share all kinetochore components: centromeric genes are lost, duplicated, and sometimes invented 48-50 . Despite these evolutionary changes, the distinctive chromatin structure of centromeres must be maintained, to accomplish its conserved function during mitosis.
Investigating whether CENP-A structures and their elasticities are conserved, or co-evolve with specific kinetochore proteins, will also provide critical clues into what drives the evolution of centromere chromatin, in turn serving as an excellent model for studying the evolution of epigenetic systems in the genome.

Experimental Model and Subject Detail
HeLa cells (female cells derived from cervical adenocarcinoma) were obtained from ATCC CCL-2 and grown at 37ºC and 5% CO2 in T-175 tissue culture flasks from Sarstedt (Cat. #83.3912.002).

All-atom computational modeling
Two nucleosomal systems were built for simulation: the CENP-A nucleosome as described previously 1 and the CENP-A nucleosome with CENP-C fragment bound from PDB ID: 4X23 2 .
The CENP-C CD fragments were docked onto the CENP-A interface using the CE algorithm 3

AFM and image analysis
Imaging of CENP-C and CENP-A N-ChIP and bulk chromatin was performed as described 6,7 with the following modifications. Imaging was performed by using standard AFM equipment diluted CENP-A chromatin or 1,000× diluted bulk chromatin was deposited on APS-mica. APSmica was prepared as previously described 6,7 . The samples were incubated for 10 min, rinsed gently to remove salts, and dried under vacuum before imaging. Images were acquired at high resolution and preprocessed on the NanoScope instrument software.
For the compaction study, we added 1 ng CENP-C CD to purified ACA samples and incubated them for 30 minutes prior to deposition on APS-mica and subsequent imaging. To determine the compaction frequency, we manually counted compacted chromatin clusters based on their size being at least twice as wide as an individual nucleosome, but with an identifiable entry and exit DNA strand.
Automated image analysis was performed as described in 6  Young's modulus of the sample and radius of the tip respectively. The radius of the tip was confirmed by SEM and found to be about 10 nm in width.

Immunostaining of mitotic chromosomes
HeLa cells were synchronized to mitosis with double thymidine block. Primary antibodies CENP-C and CENP-A were used at dilution 1:1000. Alexa secondary (488, and 568) were used at dilution of 1:1000. Images were obtained using DeltaVision RT system fitted with a CoolSnap charged-coupled device camera and mounted on an Olympus IX70. Deconvolved IF images were processed using ImageJ. Mitotic defects (lagging chromosomes and/or multipolar spindles) were counted for 83 and 76 cells (mock, GFP-CENP-C, respectively). Images were collected using a DeltaVision RT system fitted with a CoolSnap charged-coupled device camera and mounted on an Olympus IX70. Deconvolved IF images were processed using ImageJ and the macro CRaQ 11 .

Quantification and Statistical Analyses
Significant differences for nucleosome height measurement from AFM analyses and significant differences for immunostaining quantification, and chromatin compaction quantification, were performed using the t-test as described in the figure legends and main text. Significant differences for the Young's modulus of in vitro reconstituted H3, CENP-A, and CENP-A + CENP-C CD were determined using 1-way ANOVA test. Significance was determined at p <0.05.     Supplemental Figure S1 Supplemental Figure S1 A.

B.
A.

B.
A.

COMs
RMSF of all three system DNA Gaping + Sliding