Nuclear deformability and telomere dynamics areregulated by cell geometric constraintsEkta Makhijaa, D. S. Jokhuna, and G. V. Shivashankara,b,c,1

aMechanobiology Institute, National University of Singapore, Singapore 117411; bDepartment of Biological Sciences, National University of Singapore,Singapore 117543; and cInstitute of Molecular Oncology, Italian Foundation for Cancer Research, 20139 Milan, Italy

Edited by Dennis E. Discher, University of Pennsylvania, Philadelphia, PA, and accepted by the Editorial Board November 20, 2015 (received for review July5, 2015)

Forces generated by the cytoskeleton can be transmitted to thenucleus and chromatin via physical links on the nuclear envelopeand the lamin meshwork. Although the role of these active forcesin modulating prestressed nuclear morphology has been wellstudied, the effect on nuclear and chromatin dynamics remainsto be explored. To understand the regulation of nuclear deformabilityby these active forces, we created different cytoskeletal states inmouse fibroblasts using micropatterned substrates. We observedthat constrained and isotropic cells, which lack long actin stressfibers, have more deformable nuclei than elongated and polarizedcells. This nuclear deformability altered in response to actin, myosin,formin perturbations, or a transcriptional down-regulation of laminA/C levels in the constrained and isotropic geometry. Furthermore, toprobe the effect of active cytoskeletal forces on chromatin dynamics,we tracked the spatiotemporal dynamics of heterochromatin fociand telomeres. We observed increased dynamics and decreasedcorrelation of the heterochromatin foci and telomere trajectories inconstrained and isotropic cell geometry. The observed enhanceddynamics upon treatment with actin depolymerizing reagents inelongated and polarized geometry were regained once the reagentwas washed off, suggesting an inherent structural memory inchromatin organization. We conclude that active forces from thecytoskeleton and rigidity from lamin A/C nucleoskeleton can togetherregulate nuclear and chromatin dynamics. Because chromatinremodeling is a necessary step in transcription control and itsmemory, genome integrity, and cellular deformability duringmigration, our results highlight the importance of cell geometricconstraints as critical regulators in cell behavior.

mechanotransduction | cell geometry | actomyosin contractility |chromatin dynamics | telomere dynamics

Physical properties of the nucleus, such as its morphology anddeformability, have been associated with important cellular

functions like gene expression, genome integrity, and cell be-havior (1–3). The major cellular components that regulate thesephysical properties are the cytoskeleton to nuclear links and thenuclear lamina (4–8). Lineage-specific physical properties of thenucleus emerge during cellular differentiation; although stem cellnuclei are highly deformable (9, 10) and have a dynamic chromatinwith hyperdynamic chromatin proteins (11), with differentiation,nuclei lose their deformability and become less deformable (12).The nucleus in a differentiated cell is physically coupled to thecytoskeleton via lamins and the linker of nucleoskeleton andcytoskeleton (LINC) complex, which comprises transmembraneSad1p, UNC-84 (SUN) and Klarsicht, ANC-1, Syne Homology(KASH) domain proteins (13–18). Any perturbation to thesecomponents is linked to changes in nuclear morphology anddeformability (19). Therefore, the meshwork of actin stress fibersand lamin A/C serves as a critical physical intermediate in themaintenance of nuclear functional homeostasis. The physicalproperties of the nucleus govern the spatiotemporal packaging ofchromatin, which regulates lineage-specific gene expression pro-grams (20–22). In addition, modulations in cytoskeletal to nuclearlinks have been implicated in DNA damage and genome integrity

(23, 24). The maintenance of nuclear physical properties is alsoessential in cell migration during developmental programs (25) aswell as in wound healing (26, 27). Defects in nuclear morphologyand its deformability have also been shown to be important inmetastatic potential and cancer cell invasion (28). Further, anumber of diseases have been associated with loss of the me-chanical integrity of the nuclear lamina (29–31). However, theregulation of the mechanical integrity of the cell nucleus by theactive cytoskeletal network is not well understood.Recent studies have revealed that cytoskeletal organization

and nuclear morphology are regulated by extracellular mechan-ical signals, such as substrate stiffness and geometry (32–41).With the cytoskeleton physically linked to the nucleoskeleton,these extracellular mechanical signals can therefore be used tomediate changes in chromatin structure. Active cytoskeletalforces can mediate the mechanotransduction to the nucleus, toremodel 3D chromosome organization as well as permissivity tochromatin structure by regulatory molecules (42, 43). Cytoskel-etal to nuclear links are also essential to the maintenance ofpoised euchromatin and more repressive condensed chromatin,i.e., heterochromatin assembly (44). Heterochromatin is stabi-lized by links between the actin cytoskeleton and the nuclearmembrane (45–47). In addition, telomeres, the ends of chro-mosomes, are necessary for genome stability because they pro-tect chromosomes from DNA damage response machinery (48).In this context, the dynamic control of nuclear deformability by

Significance

Physical properties of the cell nucleus are important for variouscellular functions. However, the role of cell geometry and ac-tive cytoskeletal forces in regulating nuclear dynamics andchromatin dynamics is not well understood. Our results showcells with reduced matrix constraints have short actomyosinstructures. These dynamic structures together with lower laminA/C levels, resulting in softer nuclei, may provide the drivingforce for nuclear fluctuations. Furthermore, we observed in-creased dynamics of heterochromatin and telomere structuresunder such reduced cell–matrix interactions. We conclude thatextracellular matrix signals alter cytoskeletal organization andlamin A/C expression levels, which together lead to nuclearand chromatin dynamics. These results highlight the impor-tance of matrix constraints in regulating gene expression andmaintaining genome integrity.

Author contributions: E.M., D.S.J., and G.V.S. designed research; E.M. and D.S.J. per-formed research; E.M., D.S.J., and G.V.S. analyzed data; and E.M., D.S.J., and G.V.S. wrotethe paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. D.E.D. is a guest editor invited by the EditorialBoard.

Freely available online through the PNAS open access option.1To whom correspondence should be addressed. Email: shiva.gvs@gmail.com.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1513189113/-/DCSupplemental.

E32–E40 | PNAS | Published online December 22, 2015 www.pnas.org/cgi/doi/10.1073/pnas.1513189113

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 16

, 202

0

actomyosin contractility and its effect on heterochromatin andtelomere dynamics are still unclear.To understand the mechanism underlying the cytoskeleton me-

diated alterations in nuclear and chromatin dynamics, we modu-lated cytoskeletal organization using cell geometric constraints andmeasured nuclear deformability and heterochromatin and telo-mere dynamics. We found that in cells with constrained and iso-tropic geometry, the nucleus is more deformable than cells withelongated and polarized geometry. We further showed that this canbe attributed to differential force generating actomyosin structuresand differential lamin A/C expression levels in the two geometries.These active cytoskeletal forces were also found to regulate thedynamics of subnuclear heterochromatin and telomere structures.Our observations suggest that active forces from the cytoskeletonregulate nuclear and chromatin dynamics, which could in turn af-fect the spatiotemporal regulation of genomic processes and thuscell behavior.

ResultsReduced Matrix Constraints Enhance Nuclear Deformability. Variousstudies have shown that the geometry of the cell regulates cyto-skeletal organization. Hence, to probe the effect of cytoskeletalorganization on nuclear dynamics, NIH 3T3 fibroblast cells werecultured on fibronectin micropatterns of two extreme geometries;namely, large polarized (1,800 μm2 1:5 rectangle) or constrainedisotropic (500 μm2 circle). Cells cultured on these geometriesexhibited distinct cytoskeletal organization and nuclear mor-phologies. Consistent with previous studies (32), large polarized(LP) cells were flat, their actin was organized as long apical stressfibers, and the nucleus was also flat and elongated. On the otherhand, constrained isotropic (CI) cells were taller, their actin was

organized as short filaments or patches, and the nucleus wasrounded (Fig. 1A and Fig. S1 A and B). Despite the nuclear heightbeing greater in CI cells, the projected nuclear area, surface area,and volume were less compared with LP cells (Fig. S1C).Next, to study the dynamics of nuclear morphology as a function

of the two extreme cytoskeletal organizations, time lapse imagingwas performed using fibroblasts stably expressing H2B-EGFP andcultured on LP or CI fibronectin micropatterns (Movie S1). Thetime lapse images were thresholded to obtain the nuclear pe-riphery before the time series was converted to a z stack (Fig. S1D)and reconstructed in Imaris to form a surface (Fig. 1B). Thesekymographs revealed increased nuclear periphery fluctuations inCI cells. Also, superimposition of nuclear peripheries at differenttime points (Fig. 1C) revealed that in LP cells, the nuclear pe-riphery does not undergo a significant alteration with time asopposed to CI cells, which show significant fluctuations of thenuclear periphery within 10 min. To quantify the fluctuations ofthe nuclear envelope, projected nuclear area was plotted as afunction of time. This was obtained by thresholding either thewidefield images of the nucleus or the maximum intensity pro-jection of all confocal z-slices of the nucleus. Such projected nu-clear area vs. time curves were then fitted with a third-orderpolynomial, and the residuals were normalized by the value of thepolynomial fit at each time point. These normalized residualfluctuations were then plotted as a function of time. Typical timetraces of projected nuclear area fluctuations (PNAFs) revealed arelatively constant projected nuclear area in LP cells over a periodof 20 min and up to 10% fluctuations in CI cells (Fig. 1D). Thisincreased nuclear deformability of CI cells was consistently dif-ferent from LP cells, as observed over multiple cells (Fig. 1E).Typical time traces of absolute residual area (in μm2) also showedhigher fluctuations in CI cells compared with LP cells (Fig. S1E).However, the nuclear surface area and volume in CI cells did notshow such large fluctuations (Fig. S1 F and G). To quantify theamplitude of fluctuations, PNAF data from all cells and timepoints were combined to obtain a distribution. SD (σ) of thisdistribution (or the full width at half maximum of its Gaussian fit,which equals 2.355σ) was used to compare the amplitude of areafluctuations (Fig. 1F). The σ of PNAF was 5.3% in CI cells,compared with only 1.7% in LP cells (Fig. 1F, Inset). To un-derstand this cell geometry-mediated threefold change in nucleardeformability, we next probed the role of cytoskeletal forces andnuclear stiffness.

Actin, Myosin, and Formin Regulate Matrix-Assisted Nuclear Deformability.To study the role of cytoskeletal forces in PNAF, actin organizationwas perturbed in both LP and CI cells by treating them with actindepolymerizing and actin stabilizing agents cytochalasin-D andjasplakinolide, respectively. In each case, time lapse imaging wasfirst performed on control cells (n > 15). These same cells werethen treated with pharmacological agents and reimaged. Peripherykymographs for all treatments are shown in Fig. 2A. Here thePNAF showed a nonmonotonic dependence on the state of actinpolymerization. Depolymerization of F-actin in LP cells usingcytochalasin-D decreased the projected nuclear area exponentiallywith time. In this case, the PNAF was calculated by normalizing theresiduals of an exponential fit with the value of the fit at each timepoint. Cytochalasin-D treatment in LP cells enhanced PNAF from1.7% to 5.3% (Fig. 2 A and B and Movie S2), and actin stabili-zation (using jasplakinolide) in CI cells reduced PNAF from 5.3%to 1.6% (Fig. 2 A and B and Movie S3). Surprisingly, further actindepolymerization in CI cells using cytochalasin-D also reduced thePNAF from 5.3% to 2.2% (Fig. 2 A and B and Movie S4). Con-sistent PNAF were obtained upon actin perturbation in multiplecells (Fig. S2 A–C). Such nonmonotonic dependence of PNAF onactin polymerization (Fig. 2B and Fig. S3) suggests that only cellswith intermediate state of actin polymerization exhibit fluctuationsin the projected nuclear area.

Fig. 1. Reduced matrix constraints enhance nuclear deformability. (A) Maxi-mum intensity projection for confocal images of typical large polarized (LP)and constrained isotropic (CI) cell stained with Phalloidin (green) andHoechst (blue). (Scale bar, 10 μm.) See also Fig. S1. (B) Surface rendering ofnuclear periphery kymographs for a typical LP and CI cell time series. (Scalebar, 10 μm.) (C) Widefield epifluorescence images of nuclei in LP and CI cellsexpressing H2B-EGFP. Colored outlines mark the periphery of these nucleiat various time points. (D) Typical PNAFs of LP and CI cells as a function oftime. (E) PNAF vs. time plot for multiple cells. CI, 27 cells; LP, 83 cells. (F) Grayand light red curves represent a normalized histogram of combined PNAFs forall cells and all time points in LP and CI patterns, respectively. Black and redcurves represent the Gaussian fittings. Inset shows SDs of the two distributions.Performing two-sample F-test for variance on the two distributions yields **P <0.001 for nLP = 3,204 from 83 cells and nCI = 1,803 from 27 cells.

Makhija et al. PNAS | Published online December 22, 2015 | E33

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 16

, 202

0

To further explore the origin of such nuclear fluctuationsmediated by intermediate state of actin polymerization, the myosinactivity was perturbed using blebbistatin in cells with enhancedPNAF, i.e., LP cells treated with cytochalasin-D and CI cells. Ineach case the PNAF decreased to half (Fig. 2 C–F, Fig. S2 D and E,and Movies S5 and S6), suggesting that along with small polymerized

units of actin, myosin is also necessary to induce the observednuclear fluctuations. The blebbistatin mediated decrease in PNAFwas highly reproducible (Fig. S2E). Consistent with this, ATPdepletion also significantly decreased the PNAF (Fig. S4).To assess if nucleators of actin polymerization, such as formin,

could be generating these active forces, we inhibited formin ac-tivity using SMIFH2 in cells with enhanced PNAF, i.e., CI cellsand cytochalasin-D–treated LP cells. The PNAF in both caseswere reduced by half in all cells (Fig. 2 G and H, Fig. S2F, andMovie S7), confirming that formin indeed plays a role in nuclearfluctuations. Next, we assessed if the physical links between thecytoskeleton and the cell nucleus are required to generate theobserved projected area fluctuations.

LINC Complex and Microtubules Affect Amplitude of Nuclear AreaFluctuations. Actin is physically linked to the nucleus via nesprin,which is a component of the LINC complex (49). To understandwhether this physical link is necessary for nuclear deformability,cells were transfected with dominant negative (DN)-KASH fusedwith GFP, which displaces the endogenous nesprin from the nu-clear envelope to the endoplasmic reticulum, thereby disruptingthe LINC complex (50). These cells were then cultured on CI fi-bronectin micropatterns, and time lapse imaging of H2B-mRFPlabeled nuclei was performed in cells expressing DN-KASH GFP(Fig. 3A and Movie S8). Periphery kymographs of these nuclei(Fig. 3B) showed similar fluctuations as control CI cells (Fig. 1B).Typical time traces of the normalized PNAF (Fig. 3C) and the σ oftheir distribution (Fig. 3C, Inset) revealed that nesprin perturbedcells still exhibit nuclear fluctuations and the fluctuation amplitudereduced only 1.3-fold compared with control CI cells. Because thisfluctuation amplitude is still higher than control LP cells (whichshow threefold reduction compared with control CI cells), LINC

Fig. 2. Actin, myosin, and formin regulate matrix assisted nuclear deform-ability. (A) Surface rendering of nuclear periphery kymographs for typicalcontrol and treated LP and CI cells. (Scale bar, 10 μm.) (B) SDs of PNAF distri-butions of multiple control and treated LP and CI cells. Left to right representsincreasing actin polymerization. nCI+CytoD = 625 from 16 cells, nCI = 1,803 from27 cells, nLP+CytoD = 1,177 from 43 cells, nLP+CytoD+Washoff = 1,288 from 22 cells,nLP = 3,204 from 83 cells, and nCI+Jas = 1,067 from 13 cells. Performing two-sample F-test for variance on the various distributions yields **P < 0.001 forall conditions. (C) A typical PNAF trace for a LP cell sequentially treated withcytochalasin-D and blebbistatin. Inset represents normalized SDs of PNAF dis-tributions obtained by combining all cells and time points. Performing two-sample F-test for variance on the two distributions yields **P < 0.001 fornLP+CytoD = 1,177 from 43 cells and nLP+CytoD+Blebb = 1,196 from 20 cells.(D) Merge of nuclear periphery outlines at 15-min interval for a typical LP cellin untreated, cytochalasin-D, and cytochalasin-D + blebbistatin conditions.(E) PNAF vs. time plot for multiple control and blebbistatin-treated CI cells.Inset represents normalized SDs of PNAF distributions obtained by combin-ing all cells and time points. Performing two-sample F-test for variance onthe two distributions yields **P < 0.001 for nCI = 800 and nCI+Blebb = 805 from25 cells. (F) Merge of nuclear periphery outlines at 15-min interval for atypical CI cell in untreated and blebbistatin conditions. (G) Typical PNAF tracefor a LP cell sequentially treated with cytochalasin-D and SMIFH2. Insetrepresents normalized SDs of PNAF distributions obtained by combining allcells and time points. Performing two-sample F-test for variance on the twodistributions yields **P < 0.001 for nLP+CytoD = nLP+CytoD+SMIFH2 = 144 from fourcells. (H) PNAF vs. time plot for multiple control and SMIFH2-treated CI cells.Inset represents normalized SDs of PNAF distributions obtained by combin-ing all cells and time points. Performing two-sample F-test for variance onthe two distributions yields **P < 0.001 for nCI = 1,222 and nCI+SMIFH2 = 1,194from 26 cells. See also Fig. S2.

Fig. 3. Role of LINC complex and microtubules in nuclear deformability.(A) Widefield epifluorescence image of a typical CI cell coexpressing H2B-mRFP (green) and Nesprin DN-KASH EGFP (red). H2B-mRFP is shown in greento maintain consistency across the figures. (B) Surface rendering of nuclearperiphery kymograph for a typical Nesprin mutant CI cell. (C) PNAF vs. timeplot for multiple Nesprin mutant CI cells. Inset represents normalized SDs ofPNAF distributions obtained by combining all cells and time points forcontrol and DNKASH CI conditions. Performing two-sample F-test for vari-ance on the two distributions yields **P < 0.001 for nCI = 1,803 from 27 cellsand nCI+DNKASH = 530 from 5 cells. (D) Typical PNAF trace for a CI cell treatedwith nocodazole. Inset represents normalized SDs of PNAF distributionsobtained by combining all cells and time points. Performing two-sampleF-test for variance on the two distributions yields **P < 0.001 for nCI = 820and nCI+Noco = 640 from 26 cells.

E34 | www.pnas.org/cgi/doi/10.1073/pnas.1513189113 Makhija et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 16

, 202

0

complex may only partially be responsible for actomyosin medi-ated nuclear fluctuations.We next explored whether microtubules serve as additional

force transducers between the actomyosin structures and thenuclear periphery. Microtubules are physically coupled to thenucleus via LINC complex (51) and to actin via microtubule–actin crosslinking factors (MACFs) (52). To probe whether themicrotubule network that surrounds the nucleus could transmitforce for nuclear fluctuations, microtubules were depoly-merized using nocodazole in cells on CI patterns. Typical timetraces of the normalized PNAF (Fig. 3D) and the σ of theirdistribution (Fig. 3D, Inset) revealed 1.5-fold increase comparedwith control CI cells. This can be explained by either loss ofperinuclear microtubule cage and/or enhanced actomyosin con-tractile forces following microtubule depolymerization (53).Taken together, these results show that perturbations of thephysical links only partially affect the amplitude of PNAF, sug-gesting that there are additional mechanisms that drivenuclear fluctuations.

Lamin A/C Levels Inversely Regulate Nuclear Deformability. Struc-tural lamin proteins in the nuclear envelope regulate nuclearstiffness: whereas lamin A/C confers rigidity to the nucleus, itsabsence increases nuclear deformability (54, 55). To under-stand the relation between lamin-mediated nuclear stiffnessand cytoskeletal-mediated nuclear fluctuations we overexpressedlamin A/C using transient transfection in fibroblasts before cul-turing them on CI patterns. Time lapse imaging was then per-formed for cells expressing both lamin A/C-RFP and H2B-EGFP(Fig. 4A and Movie S9). Typical time traces of normalized PNAFin these cells (Fig. 4B) showed significantly lower amplitude thanin control CI cells (Fig. 1E). The σ of the distribution of the areafluctuations was reduced 1.5-fold compared with control CI cells(Fig. 4B, Inset). The nuclear periphery kymograph (Fig. 4D) alsoshowed a significant decrease in periphery fluctuations comparedwith control CI cells (Fig. 1B).This result suggested that geometric constraints placed on a

cell might regulate lamin A/C expression and thus alter PNAF.We therefore checked whether the endogenous lamin A/C levelswere altered in wild-type cells on CI compared with LP patterns.To achieve this, lamin A/C mRNA levels were measured usingquantitative (q)RT-PCR in both geometries. CI cells showed80% reduction in lamin A/C mRNA levels compared with LPcells (Fig. 4C). Decreases in lamin A/C protein in poorly spreadcells on soft substrates have been reported (56).To further investigate the role of lamin A/C in the inhibition of

PNAF, lamin A/C knockout and control (stably transfected withempty vector) MEFs (57) transfected with H2B-EGFP were cul-tured on LP micropatterns (Fig. 4 E and G). Time lapse imagingof the nuclei in these cells (Fig. 4 G and H and Movie S10)revealed increased PNAF in the knockout cells (Fig. 4 F and H).The σ of distribution of area fluctuations showed 3.4-fold increasein the knockout cells compared with control MEFs (Fig. 4H,Inset). The nuclear periphery kymograph (Fig. 4D) also showedincreased fluctuations in knockout MEFs. Taken together, theseresults suggest that lamin A/C-mediated nuclear stiffness inverselyregulates the cytoskeletal-mediated nuclear deformability.

Deformable Nuclei Have Increased Heterochromatin Dynamics. Tounderstand whether PNAF have an effect on chromatin dy-namics, we followed the trajectories of heterochromatin foci,visible as bright spots in H2B-EGFP labeled nuclei. XY tra-jectories (Fig. 5 A and B) and line kymographs across the nu-cleus (Fig. S5A) were obtained from maximum intensity projectedtime series of cells on CI and LP micropatterns. These trajectoriesand kymographs showed an increase in heterochromatin dynamicsin CI compared with LP cells. Such dynamics in CI cells wereabolished by blebbistatin treatment. In contrast, heterochromatin

foci became more dynamic in LP cells following cytochalasin-Dtreatment. Mean squared displacement (MSD) vs. time curvesshow that although the foci are usually confined in LP cells, theybecome more diffusive in CI cells (Fig. 5C). Further, the MSD vs.time curves for cytochalasin-D and blebbistatin treatments suggestthat such confined and diffusive dynamics of the heterochroma-tin foci in LP and CI patterns, respectively, are regulated byactomyosin contractility.We then sought to answer whether the correlation between

heterochromatin foci trajectories changes as a function of PNAF.For this, 3D trajectories of individual foci were obtained usingImaris (Movie S11), and their pairwise vector Pearson correla-tion coefficient was calculated (Materials and Methods and Fig.S5 B and C). In a typical CI cell, most pairs of foci trajectorieswere uncorrelated (Fig. 5 D and F). On the other hand, in atypical LP cell, the foci trajectories were highly correlated (Fig. 5E and G). Additionally, the foci pairs in LP cells could be dis-tinguished into two groups, based on their z-position. The foci inthe apical region were correlated with other foci in apical region,whereas those in the basal region were correlated with other foci inbasal region, but apical and basal foci were uncorrelated with eachother (Fig. 5 E and G and Fig. S5 D and E). The foci trajectorycorrelations increased drastically in CI cells following blebbistatin

Fig. 4. Role of lamin A/C in nuclear deformability. (A) Widefieldepifluorescence image of a typical CI cell coexpressing H2B-EGFP (green) andlamin A/C-mRFP (red). (B) PNAF vs. time plot for multiple lamin A/C over-expressing CI cells. Inset represents normalized SDs of PNAF distributionsobtained by combining all cells and time points for control and lamin A/Coverexpression CI conditions. Performing two-sample F-test for variance onthe two distributions yields **P < 0.001 for nCI = 1,803 from 27 cells andnCI+laminA=C = 210 from 4 cells. (C) mRNA levels obtained by qRT-PCR for CIcells normalized with respect to LP cells (n = 3 samples). Error bars representSE. **P < 0.001. (D) Surface rendering of a nuclear periphery kymograph fortypical lamin A/C overexpressing CI cell, control MEF cell, and lamin−/− MEFcell. (E) Widefield epifluorescence image of a typical MEF cell expressingH2B-EGFP. (F) PNAF vs. time plot for multiple MEF cells cultured on LP pat-terns (10 cells). (G) Widefield epifluorescence image of a typical lamin−/− MEFcell expressing H2B-EGFP. (H) PNAF vs. time plot for multiple lamin−/− MEFcells cultured on LP patterns. Inset represents normalized SDs of PNAF dis-tributions obtained by combining all cells and time points for MEFs andlamin−/− MEFs. Performing two-sample F-test for variance on the two dis-tributions yields **P < 0.001 for nMEF = 340 from 10 cells and nlamin�=� MEF =495 from 15 cells.

Makhija et al. PNAS | Published online December 22, 2015 | E35

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 16

, 202

0

treatment (Fig. 5H), whereas they were reduced for most foci pairsin LP cells upon cytochalasin-D treatment (Fig. 5I, Fig. S5F, andMovie S12). These results suggest that active cytoskeletal forcesmodulate correlated dynamics of chromatin domains.To understand whether such cytoskeletal mediated chromatin

perturbation was reversible, LP cells treated with cytochalasin-Dfor 30 min were washed, and time lapse imaging was performedusing the same cells. Surprisingly, the spatial map of the chro-matin images, which was altered upon cytochalasin-D treatment,was fully restored after the agent was washed off (Fig. 5J).Photobleached regions in the nucleus were also restored afterthe drug was washed off in LP cells, whereas in CI cells thebleach patterns were lost even before the drug treatment (Fig. S6A and B). The correlation between H2B-EGFP intensity histo-grams (Fig. 5K) and polarization anisotropy histograms (Fig.S6C) of drug-perturbed and washed nuclei with control nucleirevealed that an hour after the drug was washed off, the nucleihad almost returned to their initial configuration. Additionally,the pairwise foci trajectory correlations were completely restoredan hour after the drug was washed off, and apical and basalgroupings similar to control nuclei were established (Fig. 5L).The XYZ trajectories of some heterochromatin foci during thenuclear deformation phase immediately upon cytochalasin-Dtreatment also overlapped with their trajectory during the resto-ration phase immediately following the washout (Fig. S6D). These

results indicate the reversible nature of cytoskeletal-mediatedchromatin dynamics.

Telomere Dynamics Are Regulated by Active Cytoskeletal Forces.Next, we assessed the effect of PNAF on the dynamics of spe-cific functional units inside the nucleus: the telomeres. To probethe effect of PNAF on telomeres, H2B-EGFP fibroblasts weretransfected with telomeric repeat binding factor 1 (TRF1) taggedwith dsRed. These cells were cultured on LP and on CI micro-patterns, and time lapse confocal imaging was performed. Telo-meres were visible as distinct bright punctae of TRF1 dsRed in thenucleus. An overlap of z-projected TRF1 dsRed images with atime difference of 3 min showed almost perfect merge in case ofLP cells, whereas there was a significant shift in telomere positionsin case of CI cells (Fig. 6 A and B and Movies S13 and S14). Toprobe long-timescale dynamics of telomeres, line kymographsacross the nucleus in the z-projected TRF1 dsRed time series wereplotted for 1 h, which showed higher dynamics of telomeres in CIcells compared with LP cells (Fig. S7 A and B). Typical 3D tra-jectories of telomeres revealed that in LP cells, telomeres weremore confined compared with CI cells, where they explored avolume of ∼0.15 μm3 compared with 0.01 μm3 in LP cells (Fig.6C). FWHM of the histogram plotted for distance from meanposition by combining data from all time points and all telomeresshowed almost double radial spread of the telomere trajectory inCI cells compared with LP cells (Fig. 6D). Because CI cellsshowed a reduction in lamin A/C levels and the associated en-hanced telomere dynamics, we next assessed its role in regulatingthe observed dynamics. For this, lamin A/C−/− cells were plated onLP patterns, and telomere dynamics were measured (Movie S15).The radial spread in lamin A/C-deficient cells on LP patterns wasdoubled compared with WT LP cells, suggesting that lamin A/Ccould be an essential intermediate in stabilizing telomere dynamics.Because actomyosin contractility was a critical determinant ofnuclear and chromatin dynamics, we measured telomere dynamicsin blebbistatin-treated CI cells (Movie S16) and cytochalasin-D–treated LP cells (Movie S17). The radial spread of the trajectorydecreased from 0.66 μm to 0.22 μm upon blebbistatin treatment inCI cells, whereas it did not show drastic change in LP cells uponcytochalasin-D treatment (Fig. 6E). In addition, telomere speedsshowed a similar trend to their radial spreads. Telomeres in CI cellsand lamin-deficient cells on LP patterns moved significantly fasterthan control LP cells. Further, blebbistatin treatment significantlyslowed down the telomeres in CI cells, whereas cytochalasin-Dtreatment did not alter the speed of telomeres in LP cells (Fig. S7C and D).We characterized the telomere motion further by quantifying

the MSD of individual telomeres in each case. MSD vs. timeplots showed much more diffusive telomeres in CI cells com-pared with LP cells (Fig. 6F) with the exponent α, which definesthe type of motion, varying from 0.75 in CI to 0.55 in LP cells(Fig. 6G). The exponent α was also higher in lamin-deficient cellson LP patterns and cytochalasin-D–treated LP cells comparedwith control LP cells. Blebbistatin treatment of CI cells reducedα significantly compared with control CI cells (Fig. 6H).Next, to probe whether the correlation between telomere tra-

jectories change as a function of PNAF, pairwise vector Pearsoncorrelation coefficient was calculated for each pair of telomeres indifferent conditions. In a typical LP cell, most telomere trajecto-ries were correlated (mean correlation ∼0.65), whereas in CI cells,much fewer pairs were correlated (mean correlation ∼0.15) (Fig. 6I and J). Cytochalasin-D treatment altered the correlation map oftelomere trajectories in LP cells making them more uncorrelated(mean correlation ∼0.1), whereas blebbistatin treatment in CI cellsincreased the trajectory correlations (mean correlation ∼0.4).Lamin A/C-deficient cells on LP patterns also showed much fewercorrelated telomere pairs (mean correlation ∼0.1) compared withcontrol LP cells (Fig. 6K and Fig. S7 E–G). These results suggest

Fig. 5. Actomyosin contractility regulates matrix-associated heterochro-matin dynamics. (A and B) XY trajectories of multiple heterochromatin focibefore and after blebbistatin and cytochalasin-D perturbations in CI and LPcells, respectively. (C) Mean squared displacement vs. time plots for het-erochromatin foci before and after blebbistatin and cytochalasin-D pertur-bations in CI and LP cells, respectively. Error bars represent SE. (D and E)Maximum intensity z-projected images of H2B-EGFP nucleus in typical CI andLP cells. Heterochromatin foci have been numbered and color-coded basedon their z-position. Blue represents basal plane, and yellow represents apicalplane. (F and G) Pearson correlation coefficient calculated between 3D tra-jectories of all heterochromatin foci pairs labeled in D and E. See also Fig. S5.(H and I) Pearson correlation coefficient calculated between 3D trajectoriesof the same heterochromatin foci pairs upon blebbistatin and cytochalasin-Dtreatments in CI and LP cells, respectively. (J) Widefield epifluorescenceimages of H2B-EGFP nucleus in typical LP cell in untreated, cytochalasin-D,and washoff conditions. See also Fig. S6. (K ) Pearson correlation co-efficient for H2B-EGFP intensity histograms in control, cytochalasin-D-treated, and washoff conditions. Error bars represent SE. *P = 0.05, n = 7cells. (L) Pearson correlation coefficient calculated between 3D trajectoriesof same heterochromatin foci pairs (same as in E, G, and I) upon washoff ofcytochalasIn-D.

E36 | www.pnas.org/cgi/doi/10.1073/pnas.1513189113 Makhija et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 16

, 202

0

that active cytoskeletal forces as well as lamin A/C together regu-late the correlated dynamics of telomeres.

DiscussionA critical step in the alteration of genome function is the reg-ulation of nuclear and chromatin dynamics. Chromatin dy-namics can be induced by nuclear ATPases such as polymerases,topoisomerases, and chromatin remodeling complexes at milli-second timescales (58–60). However, recent findings suggest that

chromatin dynamics can also result from matrix remodeling (42,43). In addition, perturbation of cytoskeletal filaments via RNAinterference affects histone diffusion timescales (7). Further-more, altering cell geometric constraints affects gene expressionvia redistribution of chromatin remodeling enzymes (61). In thiscontext, to understand how physical forces from the cytoskeletonmight regulate genome programs, we performed quantitativeanalysis of nuclear deformations and chromatin dynamics arisingfrom such active cytoskeletal forces. Centromeres and telomeresare constitutive heterochromatin structures associated with generegulation and chromosome stability. Hence, we also probed theeffect of active cytoskeletal forces on dynamics of these functionalchromatin units. We observed an unusual nonmonotonic de-pendence of actomyosin contractility in the regulation of nucleardynamics. By altering cell geometric constraints and introducingpharmacological reagents, both actomyosin contractility and mi-crotubule organization could be modulated to investigate thisphenomenon. The nonmonotonic dependence on actomyosincontractility suggests that small actin networks cause nuclear fluc-tuations. Small punctated actin structures were indeed observed inCI and cytochalasin-D–treated LP cells, which show higher nuclearfluctuations (Fig. S8 A and B). These actin structures, whichcolocalize with myosin (Fig. S8C), have been previously reported asdynamic clusters (62) and could be the key intermediates in themechanotransduction pathway driving nuclear dynamics. To assesswhether forces generated by actomyosin contractility were applieddirectly on the nucleus, we overexpressed a dominant negativeKASH plasmid, which disrupts the link between actin and thenuclear envelope. However, cells with perturbed LINC complex onCI pattern did not show significant change in PNAF compared withcontrol CI cells. Because major changes in microtubule organiza-tion were observed when cell geometry was constrained, weinduced microtubule depolymerization to assess whether thesefilaments were involved in the mechanotransduction of force to thenucleus. However, depolymerization of microtubules did notabolish nuclear deformability, suggesting that the small actinnetworks were exerting forces on the nucleus directly. Becausethe actin cytoskeleton is physically linked to the lamin meshworkvia the LINC complex, we tested whether nuclear stiffness wasmodulated by altered cell geometry and whether this enhancedthe nuclear dynamics. We found that lamin A/C was down-regulated as matrix constraints were reduced; an effect similarto the down-regulation of lamin A/C with matrix stiffness (56).This finding prompted us to transiently transfect cells with laminA/C. Indeed, overexpression of lamin A/C abolished the nuclearfluctuations, and this was in contrast to lamin A/C knockdown,which increased nuclear fluctuations. Taken together, theseexperiments highlighted an important transcription-dependentmechanoregulatory pathway involving actomyosin contractilitythat couples matrix properties to nuclear dynamics. A recent studyalso revealed that Rac-1–mediated nuclear actin also causes nu-clear envelope deformations (63).Next, we wanted to test whether the dynamics of chromatin

remodeling were affected by the enhanced nuclear deformability.Heterochromatin structures have been shown to be stabilized incells that are strongly adhered to the extracellular matrix (64). Wehypothesized that a reduction in the number of physical links tothe extracellular matrix may disrupt heterochromatin integrity,thus making chromatin more permissive. Consistent with this,heterochromatin dynamics increased and correlation betweenheterochromatin foci trajectories decreased in cells on constrainedisotropic geometry. The dynamic correlations of heterochromatinin cells of elongated polarized geometry were position dependent.Here the apical and basal foci were highly correlated with otherapical and basal foci, respectively. Also, the motion of telomeresnearer to the nuclear membrane was more correlated to envelopefluctuations than that of a more interiorly positioned telomere(Fig. S9), suggesting that an elaborate structural network inside

Fig. 6. Actomyosin contractility regulates matrix-associated telomere dy-namics. (A and B) Overlap of nucleus periphery and maximum intensityz-projected images of a TRF1-dsRed nucleus at 3-min interval in typical LP andCI cells. Bright spots represent telomeres. (Scale bar, 4 μm.) (C) XYZ trajectoriesof typical telomeres in LP and CI cells. (D) Normalized histogram of displace-ment from mean position, combining data for n = 89 telomere trajectoriesfrom n = 3 LP cells and n = 68 telomere trajectories from n = 3 CI cells. Blackand red curves represent the Gaussian fittings. (E) Full width at half maxima(FWHM) of Gaussian fitting of histograms for displacement from mean posi-tion in LP cells, cytochalasin-D–treated LP cells (2 cells, 62 telomeres), CI cells,blebbistatin-treated CI cells (3 cells, 141 telomeres), and laminA/C-deficientCI cells (2 cells, 58 telomeres). (F ) Mean squared displacement vs. time plotsfor multiple telomeres in control LP and CI cells. (G) Box plot for the ex-ponent α in LP and CI cells obtained by fitting MSD ‹r2(τ)› vs. time lag τplots to the equation ‹r2(τ)› = Dτα. The bottom and top of the box repre-sent the first and third quartiles, whereas the line and dot inside thebox represent the median and mean, respectively. The ends of the whis-kers correspond to the lowest/highest data point of the distribution. nLP =62 telomeres from two cells, and nCI = 141 telomeres from three cells.(H) Mean α values for LP cells, cytochalasin-D–treated LP cells, CI cells,blebbistatin-treated CI cells, and laminA/C-deficient CI cells. Error barsrepresent SE, and the means are significantly different with **P < 0.01.nLP = 68 telomeres from three cells, nLP+CytoD = 62 telomeres from two cells,nCI = 68 telomeres from three cells, nCI+Blebb. = 141 telomeres from threecells, and nlamin�=� MEFs = 58 telomeres from two cells. (I and J) Pearsoncorrelation coefficient calculated between 3D trajectories of 15 telomerepairs (chosen at random) in typical LP and CI cells, respectively. The telo-meres are sorted (top to bottom and left to right) in ascending z-position,i.e., basal to apical. (K ) Mean value of the correlation coefficient for LPcells, cytochalasin-D–treated LP cells (3 cells, 89 telomeres), CI cells (3 cells,68 telomeres), blebbistatin-treated CI cells (3 cells, 141 telomeres), andlaminA/C-deficient CI cells (2 cells, 58 telomeres). Error bars represent SE,and the means are significantly different with **P < 0.01. See also Fig. S7.

Makhija et al. PNAS | Published online December 22, 2015 | E37

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 16

, 202

0

the nucleus may be modulated by changes in actomyosin contrac-tility. The pharmacological inhibitor washout experiments revealedthat dynamic correlation in heterochromatin organization andits apical–basal grouping, which was reduced upon actin de-polymerization, was restored after the actin depolymerizing agentwas washed out. This suggests that a structural memory in thespatial organization of heterochromatin exists and highlights theimportance of mechanical homeostatic balance for higher-orderchromatin organization in living cells. Our studies also revealed animportant role for actomyosin contractility in maintaining telo-mere positioning and dynamics. The enhanced dynamics of telo-meres with actin depolymerization suggests that the telomere endshave to be protected during various cellular processes including itsmigration. Because these ends are susceptible to DNA damage,the cytoskeletal control of telomere stability is important togenomic integrity. In conclusion, our data systematically revealan important link between cytoskeletal components and nuclearand chromatin dynamics (Fig. 7). We suggest that the mechanicalchanges within the extracellular matrix tune the epigenetic statesof chromatin dynamics via actomyosin contractility to regulatecellular functions including transcription, genome integrity, mi-gration, and cellular homeostasis.

Materials and MethodsCell Culture, Pharmacological Perturbations, and Plasmid Transfections. Wild-type NIH 3T3 fibroblasts and NIH 3T3 fibroblasts stably expressing H2B-EGFPwere cultured in low-glucose Dulbecco’s Modified Eagle Medium (Gibco; LifeTechnologies) supplemented with 10% (vol/vol) FBS (Gibco; Life Technolo-gies) and 1% penicillin-streptomycin (Gibco; Life Technologies) at 37 °C and5% CO2 in humid conditions. Cells were trypsinized (Gibco; Life Technolo-gies) and seeded on fibronectin (Sigma) micropatterned dishes for3 h before imaging.

Cytochalasin-D (Sigma) was used at 500 nMworking concentration, and cellswere imaged either immediately or 30 min after treatment. Jasplakinolide(Gene Ethics) was used at concentration of 200 nM, and cells were imaged

20 min after treatment. Blebbistatin (Merck) was used at concentration of25 μM, and cells were imaged 30 min after treatment. SMIFH2 (ChemBridgeCorporation) was used at a concentration of 20 μM, and cells were imaged anhour after treatment. Nocodazole (Sigma) was used at a concentration of10 μg/mL, and cells were imaged an hour after treatment.

All transfections were carried out using jetPRIME (Polyplus transfection).

Real-Time PCR. Real-time PCR was used to quantify the fold change in laminA/C expression in CI and LP cells. mRNA was extracted using RNeasy Mini kit(Qiagen). Five hundred nanograms of total mRNA extracted was subjected tocDNA synthesis using iScript cDNA Synthesis kit (Bio-Rad). qPCR was per-formed using SsoFast qPCR (Bio-Rad) for 40 cycles in a Bio-Rad CFX96. Therelative fold change in levels of lamin A/C was then measured with respect toGAPDH levels.

Primer sequences used were

LaminA (fwd-GTACAACCTGCGCTCACGCACCGT;

rev-CACTGCGGAAGCTTCGAGTGACT); and

GAPDH (fwd-GACCAGGTTGTCTCCTGCGACTT;rev-CCATGAGGTCCACCACCCTGTT).

Preparation of PDMS Stamps, Microcontact Printing, and Cell Seeding onPatterns. To make stamps, PDMS (Sylgard 184; Dow Corning) precursorand curing agent were mixed homogeneously in 10:1 ratio and poured overthe silicon wafer which had large polarized (LP) or constrained isotropic (CI)micropatterned wells. After degassing in the desiccator for 30 min to removeair bubbles from the PDMS mixture, the silicon wafer with the PDMS mixturewas cured in the oven at 80 °C for 2 h. Solidified PDMS was then peeled fromthe wafer and cut into ∼1 cm × 1 cm stamps. These stamps were oxidizedusing plasma for 4 min, and then 15 μL of 100 μg/mL fibronectin solution(mixed with Alexa Fluor 647 dye; Sigma) was poured over each stamp.Extra solution was wiped with a tissue, and the stamp was allowed to dryfor 10 min. The stamp was then checked under the microscope for com-plete drying between the micropatterned structures, after which it wasinverted carefully onto the surface of an uncoated hydrophobic 35-mmdish (Ibidi). The stamp was gently removed after 2 min, and the stampingon the dish was checked by visualizing Alexa 647 fluorescence in the far-red channel in the epifluorescence microscope. To passivate the non-patterned surface of the dish, it was then treated with 2 mg/mL pluronicF-127 for 5 min and washed twice with PBS and cell culture medium beforeseeding single cells.

Imaging, Image Processing, and 3D Rendering. All imaging was carried outusing a 100× objective on a NikonA1R Confocal microscope. Time lapse im-aging was done in either widefield or confocal mode with 30-s, 60-s, or 90-stime intervals for up to 60 min in each condition. The z-depth for confocalimaging was 0.5 μm.

H2B-EGFP images were thresholded, projected nuclear area was calcu-lated, and time lapse images of the nuclear periphery were generated usingcustom written code in MATLAB. Merged images of the nuclear periphery atdifferent time points were generated in ImageJ.

To generate edge kymographs, time series of nuclear periphery imageswere analyzed in IMARIS as a z-stack (Fig. S1D) and then fitted with a surface.Three-dimensional rendering of actin, microtubules, and the nucleus wasalso done using IMARIS. Line kymographs across the nucleus for visualizingheterochromatin foci dynamics were generated in ImageJ.

Data Analysis and Statistical Tests. Absolute projected nuclear area (in μm2)was first measured by thresholding either the widefield nucleus images ormaximum intensity projection of confocal z-slices of the nucleus. This pro-jected area was then plotted as a function of time and fitted with third-order polynomial (or exponential for LP+CytoD condition) curves in ORIGIN.The residual values were divided by the value of the polynomial (or expo-nential) at each time point and multiplied by 100 to obtain the percentagenormalized residual area fluctuations or the PNAF.

Such PNAFs were combined from multiple cells and time points for eachcondition to obtain a normal distribution. SD (σ) of such distribution indicatesthe amplitude (in percentage) of area fluctuations. To determine whetherthe difference in σ of PNAF distribution for various conditions is statisticallysignificant, two-sample F-test for variance was performed in ORIGIN. P val-ues were less than 0.001 in all cases.

Heterochromatin Foci and Telomere Trajectory Correlation Analysis. Time lapseconfocal stacks of H2B-EGFP nuclei were opened in IMARIS. Nucleus

Fig. 7. A model summarizing cytoskeletal and nucleoskeletal regulation ofnuclear and chromatin dynamics. The actin is shown in purple, the micro-tubules are shown in green, the lamin A/C is shown as blue (time t) and red(time t + Δt) outline of the nucleus, and heterochromatin foci are shown asblue (time t) and red (time t + Δt) structures of DNA. Briefly, in LP cells thelong apical actin stress fibers press on the nucleus, making it flat and elon-gated. The lamin A/C expression levels are higher. The nuclear periphery, aswell as heterochromatin foci, is less dynamic. In contrast, in CI cells, longactin stress fibers are absent, and actin exists as meshwork of short filamentsand punctae. The zoom-in of actin structures shows actin, myosin, and for-min asters as reported earlier (62). Lamin A/C expression levels are lower. Wespeculate that active forces from the dynamic actin–myosin–formin asters, aswell as decreased rigidity because of lower lamin A/C expression levels, makethe nuclear periphery and heterochromatin foci more dynamic in CI cells. Seealso Fig. S8.

E38 | www.pnas.org/cgi/doi/10.1073/pnas.1513189113 Makhija et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 16

, 202

0

trajectory was corrected for translation and rotation shift. Bright spotscorresponding to heterochromatin foci/telomere were picked using sur-face thresholding, and their centroid xyz trajectories were obtained. Tocalculate correlation between the xyz trajectories of heterochromatinfoci (or telomeres), Pearson correlation coefficient was used, and thescalar variables A(t) and B(t) were replaced by vectors AðtÞ��!

and BðtÞ��!as

shown below:

vector  correlation  coefficient=

P�AðtÞ��!

−~Amean

�  ·  

�BðtÞ��!

−~Bmean

�ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi�~AðtÞ−~Amean

�2r ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi�

~BðtÞ−~Bmean

�2r .

Various trajectories were simulated to verify that the vector correlationcoefficient yields high correlation values between trajectories of particlesmoving in same direction and low correlation values between those moving

in opposite directions (Fig. S5B). Vector correlation coefficients of the sim-ulated trajectories (Fig. S5C) were calculated using custom written codes inMATLAB. The vector correlation is highest (+1) when the angle betweentrajectories is 0°, is lowest (−1) when the angle between trajectories is 180°,and has values between −1 and +1 for angles between 0° and 180°. Usingthis approach, vector correlation coefficients were then calculated betweenpairs of heterochromatin foci (and telomeres), and their correlation matriceswere generated using custom written codes in MATLAB.

ACKNOWLEDGMENTS.We thankMarco Foiani, Jacques Prost, andMadan Raofor useful discussions. We thank Steven Wolf for critical reading of the man-uscript, Zhang Bo for preparing the summary cartoon, and Mallika Nagrajanfor helping with TRF transfections. Lamin A/C knockdown MEFs and NesprinDN-KASH plasmid were kind gifts from the Colin Stewart and Brian Burkelaboratories. We thank the Mechanobiology Institute, National Universityof Singapore, Singapore, and Ministry of Education Tier-3 Grant Programfor funding.

1. Isermann P, Lammerding J (2013) Nuclear mechanics and mechanotransduction inhealth and disease. Curr Biol 23(24):R1113–R1121.

2. Woringer M, Darzacq X, Izeddin I (2014) Geometry of the nucleus: A perspective ongene expression regulation. Curr Opin Chem Biol 20:112–119.

3. Burke B, Stewart CL (2014) Functional architecture of the cell’s nucleus in develop-ment, aging, and disease. Curr Top Dev Biol 109:1–52.

4. Mazumder A, Shivashankar GV (2010) Emergence of a prestressed eukaryotic nucleusduring cellular differentiation and development. J R Soc Interface 7(Suppl 3):S321–S330.

5. Broers JL, et al. (2004) Decreased mechanical stiffness in LMNA-/- cells is caused bydefective nucleo-cytoskeletal integrity: Implications for the development of lam-inopathies. Hum Mol Genet 13(21):2567–2580.

6. Shumaker DK, Kuczmarski ER, Goldman RD (2003) The nucleoskeleton: Lamins andactin are major players in essential nuclear functions. Curr Opin Cell Biol 15(3):358–366.

7. Ramdas NM, Shivashankar GV (2015) Cytoskeletal control of nuclear morphology andchromatin organization. J Mol Biol 427(3):695–706.

8. Guilluy C, et al. (2014) Isolated nuclei adapt to force and reveal a mechano-transduction pathway in the nucleus. Nat Cell Biol 16(4):376–381.

9. Pajerowski JD, Dahl KN, Zhong FL, Sammak PJ, Discher DE (2007) Physical plasticity of thenucleus in stem cell differentiation. Proc Natl Acad Sci USA 104(40):15619–15624.

10. Talwar S, Kumar A, Rao M, Menon GI, Shivashankar GV (2013) Correlated spatio-temporal fluctuations in chromatin compaction states characterize stem cells. BiophysJ 104(3):553–564.

11. Meshorer E, et al. (2006) Hyperdynamic plasticity of chromatin proteins in pluripotentembryonic stem cells. Dev Cell 10(1):105–116.

12. Bhattacharya D, Talwar S, Mazumder A, Shivashankar GV (2009) Spatio-temporalplasticity in chromatin organization in mouse cell differentiation and during Dro-sophila embryogenesis. Biophys J 96(9):3832–3839.

13. Wang N, Tytell JD, Ingber DE (2009) Mechanotransduction at a distance: Mechanicallycoupling the extracellular matrix with the nucleus. Nat Rev Mol Cell Biol 10(1):75–82.

14. Versaevel M, et al. (2014) Super-resolution microscopy reveals LINC complex re-cruitment at nuclear indentation sites. Sci Rep 4:7362.

15. Maniotis AJ, Chen CS, Ingber DE (1997) Demonstration of mechanical connectionsbetween integrins, cytoskeletal filaments, and nucleoplasm that stabilize nuclearstructure. Proc Natl Acad Sci USA 94(3):849–854.

16. Osmanagic-Myers S, Dechat T, Foisner R (2015) Lamins at the crossroads of mecha-nosignaling. Genes Dev 29(3):225–237.

17. Swift J, Discher DE (2014) The nuclear lamina is mechano-responsive to ECM elasticityin mature tissue. J Cell Sci 127(Pt 14):3005–3015.

18. Chambliss AB, et al. (2013) The LINC-anchored actin cap connects the extracellularmilieu to the nucleus for ultrafast mechanotransduction. Sci Rep 3:1087.

19. Sims JR, Karp S, Ingber DE (1992) Altering the cellular mechanical force balance resultsin integrated changes in cell, cytoskeletal and nuclear shape. J Cell Sci 103(Pt 4):1215–1222.

20. Lanctôt C, Cheutin T, Cremer M, Cavalli G, Cremer T (2007) Dynamic genome archi-tecture in the nuclear space: Regulation of gene expression in three dimensions. NatRev Genet 8(2):104–115.

21. Talwar S, Jain N, Shivashankar GV (2014) The regulation of gene expression duringonset of differentiation by nuclear mechanical heterogeneity. Biomaterials 35(8):2411–2419.

22. Shivashankar GV (2011) Mechanosignaling to the cell nucleus and gene regulation.Annu Rev Biophys 40:361–378.

23. Wang L, et al. (2013) Actin polymerization negatively regulates p53 function by im-pairing its nuclear import in response to DNA damage. PLoS One 8(4):e60179.

24. Zuchero JB, Belin B, Mullins RD (2012) Actin binding to WH2 domains regulatesnuclear import of the multifunctional actin regulator JMY. Mol Biol Cell 23(5):853–863.

25. Kumar A, Shivashankar GV (2012) Mechanical force alters morphogenetic movementsand segmental gene expression patterns during Drosophila embryogenesis. PLoS One7(3):e33089.

26. Gundersen GG, Worman HJ (2013) Nuclear positioning. Cell 152(6):1376–1389.27. Harada T, et al. (2014) Nuclear lamin stiffness is a barrier to 3D migration, but softness

can limit survival. J Cell Biol 204(5):669–682.

28. Ribeiro AJ, Khanna P, Sukumar A, Dong C, Dahl KN (2014) Nuclear stiffening inhibitsmigration of invasive melanoma cells. Cell Mol Bioeng 7(4):544–551.

29. Worman HJ, Ostlund C, Wang Y (2010) Diseases of the nuclear envelope. Cold SpringHarb Perspect Biol 2(2):a000760.

30. Folker ES, Ostlund C, Luxton GW, Worman HJ, Gundersen GG (2011) Lamin A variantsthat cause striated muscle disease are defective in anchoring transmembraneactin-associated nuclear lines for nuclear movement. Proc Natl Acad Sci USA108(1):131–136.

31. Lee JS, et al. (2007) Nuclear lamin A/C deficiency induces defects in cell mechanics,polarization, and migration. Biophys J 93(7):2542–2552.

32. Li Q, Kumar A, Makhija E, Shivashankar GV (2014) The regulation of dynamic me-chanical coupling between actin cytoskeleton and nucleus by matrix geometry.Biomaterials 35(3):961–969.

33. Solon J, Levental I, Sengupta K, Georges PC, Janmey PA (2007) Fibroblast adaptationand stiffness matching to soft elastic substrates. Biophys J 93(12):4453–4461.

34. Zemel A, Rehfeldt F, Brown AE, Discher DE, Safran SA (2010) Optimal matrix rigidityfor stress fiber polarization in stem cells. Nat Phys 6(6):468–473.

35. Chen CS (2008) Mechanotransduction - A field pulling together? J Cell Sci 121(Pt 20):3285–3292.

36. Engler AJ, Sen S, Sweeney HL, Discher DE (2006) Matrix elasticity directs stem celllineage specification. Cell 126(4):677–689.

37. Kilian KA, Bugarija B, Lahn BT, Mrksich M (2010) Geometric cues for directingthe differentiation of mesenchymal stem cells. Proc Natl Acad Sci USA 107(11):4872–4877.

38. Roca-Cusachs P, et al. (2008) Micropatterning of single endothelial cell shape reveals atight coupling between nuclear volume in G1 and proliferation. Biophys J 94(12):4984–4995.

39. Wang D, et al. (2014) Tissue-specific mechanical and geometrical control of cell via-bility and actin cytoskeleton alignment. Sci Rep 4:6160.

40. Khatau SB, et al. (2009) A perinuclear actin cap regulates nuclear shape. Proc NatlAcad Sci USA 106(45):19017–19022.

41. Kim DH, Wirtz D (2015) Cytoskeletal tension induces the polarized architecture of thenucleus. Biomaterials 48:161–172.

42. Iyer KV, Pulford S, Mogilner A, Shivashankar GV (2012)Mechanical activation of cells induceschromatin remodeling preceding MKL nuclear transport. Biophys J 103(7):1416–1428.

43. Poh YC, et al. (2012) Dynamic force-induced direct dissociation of protein complexesin a nuclear body in living cells. Nat Commun 3:866.

44. Mazumder A, Shivashankar GV (2007) Gold-nanoparticle-assisted laser perturbationof chromatin assembly reveals unusual aspects of nuclear architecture within livingcells. Biophys J 93(6):2209–2216.

45. Dahl KN, Engler AJ, Pajerowski JD, Discher DE (2005) Power-law rheology of iso-lated nuclei with deformation mapping of nuclear substructures. Biophys J 89(4):2855–2864.

46. Makatsori D, et al. (2004) The inner nuclear membrane protein lamin B receptor formsdistinct microdomains and links epigenetically marked chromatin to the nuclear en-velope. J Biol Chem 279(24):25567–25573.

47. Raz V, et al. (2006) Changes in lamina structure are followed by spatial reorganizationof heterochromatic regions in caspase-8-activated human mesenchymal stem cells.J Cell Sci 119(Pt 20):4247–4256.

48. Blackburn EH (2001) Switching and signaling at the telomere. Cell 106(6):661–673.49. Zhang Q, Ragnauth C, Greener MJ, Shanahan CM, Roberts RG (2002) The nesprins are

giant actin-binding proteins, orthologous to Drosophila melanogaster muscle proteinMSP-300. Genomics 80(5):473–481.

50. Lombardi ML, et al. (2011) The interaction between nesprins and sun proteins at thenuclear envelope is critical for force transmission between the nucleus and cyto-skeleton. J Biol Chem 286(30):26743–26753.

51. Roux KJ, et al. (2009) Nesprin 4 is an outer nuclear membrane protein that can inducekinesin-mediated cell polarization. Proc Natl Acad Sci USA 106(7):2194–2199.

52. Sun D, Leung CL, Liem RK (2001) Characterization of the microtubule binding domainof microtubule actin crosslinking factor (MACF): Identification of a novel group ofmicrotubule associated proteins. J Cell Sci 114(Pt 1):161–172.

53. Waterman-Storer C, et al. (2000) Microtubules remodel actomyosin networks in Xenopusegg extracts via two mechanisms of F-actin transport. J Cell Biol 150(2):361–376.

Makhija et al. PNAS | Published online December 22, 2015 | E39

BIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 16

, 202

0

54. De Vos WH, et al. (2010) Increased plasticity of the nuclear envelope and hypermobilityof telomeres due to the loss of A-type lamins. Biochim Biophys Acta 1800(4):448–458.

55. Lammerding J, et al. (2006) Lamins A and C but not lamin B1 regulate nuclear me-chanics. J Biol Chem 281(35):25768–25780.

56. Swift J, et al. (2013) Nuclear lamin-A scales with tissue stiffness and enhances matrix-directed differentiation. Science 341(6149):1240104.

57. Sullivan T, et al. (1999) Loss of A-type lamin expression compromises nuclear envelopeintegrity leading to muscular dystrophy. J Cell Biol 147(5):913–920.

58. Zidovska A, Weitz DA, Mitchison TJ (2013) Micron-scale coherence in interphasechromatin dynamics. Proc Natl Acad Sci USA 110(39):15555–15560.

59. Hameed FM, Rao M, Shivashankar GV (2012) Dynamics of passive and active particlesin the cell nucleus. PLoS One 7(10):e45843.

60. Sinha DK, Banerjee B, Maharana S, Shivashankar GV (2008) Probing the dynamicorganization of transcription compartments and gene loci within the nucleus of livingcells. Biophys J 95(11):5432–5438.

61. Jain N, Iyer KV, Kumar A, Shivashankar GV (2013) Cell geometric constraintsinduce modular gene-expression patterns via redistribution of HDAC3regulated by actomyosin contractility. Proc Natl Acad Sci USA 110(28):11349–11354.

62. Luo W, et al. (2013) Analysis of the local organization and dynamics of cellular actinnetworks. J Cell Biol 202(7):1057–1073.

63. Navarro-Lérida I, et al. (2015) Rac1 nucleocytoplasmic shuttling drives nuclear shapechanges and tumor invasion. Dev Cell 32(3):318–334.

64. Brown SW (1966) Heterochromatin. Science 151(3709):417–425.

E40 | www.pnas.org/cgi/doi/10.1073/pnas.1513189113 Makhija et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 16

, 202

0