CELL GEOMETRIC CONSTRAINTS INDUCE ACTOMYOSIN CONTRACTILITY DEPENDENT CHANGES IN GENE EXPRESSION AND CHROMOSOME POSITIONING

In chapter 2 of this thesis, using fibronectin coated micro-fabricated patterns of varying geometries along with whole transcriptome micro-array complemented with confocal imaging and qu

Chapter1: Introduction

Recent work in the field of mechanobiology have suggested that cells respond to physical signals like shear stress, substrate rigidity, stretching and even geometrical cues available

in their microenvironment (1) The relevance of physical signals has been implicated in cell differentiation (2), tissue morphogenesis (3) and development (4) The transmission

of these signals from plasma membrane to the nucleus involves cytoskeletal and nuclear envelope proteins (5, 6) The nuclear morphology and the nucleoskeletal organization changes upon receiving physical signals, which also correlates with alterations in chromatin dynamics leading to changes in gene expression (7-9) The induction of modular gene expression programs eventually correlates with the non-random spatial organization of chromosomes

In the introduction chapter, I shall first summarize the current understanding of (i) the transmission of physical signals from focal adhesion to the nucleus and the critical molecules involved in this process (ii) how these signals regulate gene expression In the end, I shall state the aim of my thesis

Briefly, I study the regulation of nuclear morphology by geometrical cues and how it correlates with chromatin compaction and gene expression The critical molecular intermediates involved in the geometry regulated chromatin compaction and gene expression are further explored Lastly, using stem cell differentiation, I have tried to establish a link between changes in global gene expression, total chromosome activity and chromosomal organization

1.1 Sensing and transmission of physical signals: Cells respond to the alterations in their microenvironment This has been shown using various approaches including

changes in matrix elasticity (10), cell geometry (11, 12), matrix topography (13), application of shear stress (14, 15) and substrate stretching (16) Physical sensing of extra cellular environment is initiated by focal adhesions (17-20)complex proteins and they have been identified as mechano-sensors (21, 22) Some of the well-studied mechano-sensitive proteins of focal adhesion complex includes talin (23), paxillin (24), zyxin (25, 26) and p130Cas (27), which undergo post-translational modification like phosphorylation upon stretching Application of force ~20pN, using Atomic force microscopy (AFM), resulted in extension of talin rod domain by 100nm Again using AFM, less contractile response by zyxin null cells has been shown, which is largely due

to acto-myosin complex The stretching and extension events of the focal adhesion proteins has been proposed to be required for force sensing and presumably to further transmit them to downstream cytoskeletal proteins majorly actin Although the connections between focal adhesions and cytoskeletal networks are very dynamic, seconds to minutes turn over-time scale, but it can efficiently propagate physical signals (28-31)

Mechanical forces transduced from focal adhesions to the actin cytoskeleton (32) results

in actin reorganization, which has been studied using fluid shear stress (33)and patterned islands (34) Actin cytoskeleton orientation changes based on the direction of fluid shear stress and the shape of the underlying pattern By applying nanonewton force

micro-on cells using AFM, actin reorganizatimicro-on has been shown to be a two-step process involving short term local deformation and long term remodelling (35) Physical signals

not only alters actin organization (36) but also results in actin filament regeneration which is mediated by formins (37, 38)which also regulate its activity by modulating free pool of monomeric actin

Actin interacts with myosin to form acto-myosin complex, which are contractile in nature

(39, 40) and has been described as “tension sensor” (41) Acto-myosin complex also provides a path for propagation of physical signals (42, 43) within the cell and has been implicated as the major regulator of force mediated alterations in developmental programs Actin in the cells, with respect to nucleus, has been shown to be present on top

of the nucleus and at the bottom of the nucleus known as apical actin (44) and basal actin respectively Thick apical actin fibers directly associates with focal adhesion, are contractile in nature (45) and are highly mechano-sensitive Within 5 minutes of application of low shear stress apical actin get reformed and reorganized contrary to basal actin which requires 50-fold higher levels of shear stress (46) Recently, various isoforms

of myosin have also been shown to display mechanosensitive behaviour (47-51) In a recent work, using optical tweezers, the strain sensing ability of myosin I have been characterized (52) Micro-pipette experiments have also revealed mechano-sensitivity of myosin II, whereby it has been elucidated that myosin II gets accumulated at the point of mechanical perturbation (53) Contractile nature of acto-myosin complex has been shown

to applies tensile load on the nucleus (54) largely via apical actin Decrease in nuclear area and increase in height has been shown by using low dose of Lantrunculin-A, which only depolyemerize apical actin (44) Contrary to actin, microtubules exert compressive load on the nucleus (54) whereas role of vimentin, an intermediate filaments, in nuclear dynamics are still largely unknown

Acto-myosin complex directly interacts with proteins on nuclear envelope and hence establish a physical connection between plasma membrane and nucleus Physical signals transmitted via acto-myosin contractility impinges on nucleus through connections between actin cytoskeleton and nuclear envelope (55), which are highly dynamic in nature There are various structural elements in the nuclear envelope (56, 57) major being inner and outer nuclear membrane (INM and ONM respectively) INM and ONM is separated by a gap size of ~50nm and harbour LINC (Linker of nucleus and cytoskeleton) complex, group of proteins bridging cytoskeleton and nucleus (58, 59) Transmission of physical signals between the cytoskeleton and the nucleus is largely mediated by LINC complex since impaired propagation of intracellular forces upon LINC complex perturbation has been shown (60, 61) Using RNAi studies, the role of nesprin, a LINC complex protein on ONM, in nuclear deformation upon force application has been studied (62) Actin and other cytoskeletal proteins including microtubules and intermediate filaments directly interacts with LINC complex proteins (63)

1.2 Response upon receiving physical signals: Physical signals transmitted via

cytoskeleton and nuclear envelope proteins impinge on nuclear architecture and dynamics (8, 9, 64, 65) including nuclear rotation, movement and positioning (46, 66-72) Imposing differential geometrical constraints on the cells resulted in changes in nuclear volume, height, shape and positionins (64, 65, 73) Similarly, changes in substrate rigidity from 0.4KPa to 300KPa resulted in significant reduction in nuclear height (74) further showing effect of physical signals on nuclear architecture Externally applied forces also changes nuclear stiffness, which ranges from 0.1KPa to 10KPa depending on the cell type (75-78) and has been quantified by micropipette aspiration of cells Physical signals regulated

acto-myosin contractility dependant changes in morphology, stiffness and temporal dynamics of nucleus has been suggested to correlate with chromatin compaction and gene expression (81) and are hallmark of stem cell differentiation (79) (80)

spatio-Several studies have shown that stretching of human osteoblasts (82, 83), fibroblasts (84) and other cell types (85, 86), application of fluid shear stress on human mesenchymal stem cells (87) and culturing stem cells on varying geometrical patterns (43)affects gene expression profile Application of geometrical cues on smooth muscle cells, directly impinges on nuclear volume, DNA synthesis and gene expression (88) Culturing stem cells on micro and nanoscale patterns modulates their self-renewal and differentiation potential (2, 89, 90) By applying varying geometric cues on mesenchymal stem cells (MSCs), lineage specification was found to get altered which was largely sensitive to the curvature of the underlying patterns (43) The contribution of various physical signals to human embryonic stem cell lineage commitment further strengthens the role of physical signals in modulating genetic profiles (89) However, the transmission of physical signals

to the nucleus are not sufficient for inducing changes in gene expression This requires interaction of chromatin with specific transcription factors and co-factors as described later

1.3 Chemical intermediates involved in physical signals transduction: Changes in

gene expression largely requires specific activity of transcription factors (TFs) and factors (91) NF-κB (92), FosB (93), JNK-AP1 (94), Egr1 (95), BMP2 (96) are some of the TFs which has been suggested to be mechano-sensitive These TFs are compartmentalized in the nucleus and the cytoplasm Depending upon the physical signal their shuttling dynamics between two compartments alters resulting in modular changes

co-in gene expression One of the important factor which translate varied physical signals into differential gene expression is YAP\TAZ (97) Nuclear to cytoplasmic redistribution

of YAP\TAZ occurs upon culturing cells on patterns of varying sizes or using substrates and pillars of different stiffness and rigidity As the cell size and the substrate stiffness decreases YAP\TAZ translocates to the nuclues which inturn influences the differentiation potential of MSCs and the survival of endothelial cells (97, 98) Another important mechano-sensitive transcription co-factor is MRTF-A (99) MRTF-A is known

to form complex with Serum response factor (SRF) and regulates a large number of genes majorly related to actin cytoskeleton and focal adhesion By providing varying geometrical cues to keratinocytes, MRTF-A localization, activity and hence differential effect on differentiation was determined (100) Application of a static tensile load of 0.65pN/µm2 on cells using collagen coated beads also resulted in nuclear translocation of MRTF-A further showing mechano-senstive behavior of MRTF-A Nuclear to cytoplasmic redistribution of majority of these factors and co-factors is largely attributed

to acto-myosin contractility and cytoskeletal tension Maintenance of nuclear YAP/TAZ

in MSCs is reported to be Rho GTPase and stress fibers dependent whereby treatment with Latrunculin-A resulted in lower nuclear levels of YAP\TAZ Tensile load mediated nuclear translocation of MRTF-A also depends on activation of RhoA phosphorylation which results in actin polymerization (101, 102) Higher levels of polymerized actin results in MRTF-A nuclear translocation since monomeric actin (G-actin) sequesters MRTF-A in the cytoplasm Hence, these factors have been characterized as the chemical link between the physical signals and changes in gene expression which are regulated by altered cytoskeleton dynamics and tension

Once these mechano-sensitive TFs translocates to the nucleus, they need to converge on their downstream target genes residing on various chromosomes This points towards the argument that there must be a well defined spatial organization of chromosomes, which is physiologically and energetically more favourable than random positioning, so that TFs can access their targets efficiently In the next section, a detailed description of chromosome organization and gene clustering is discussed

1.4 Non random organization of chromosomes territories: Packaging of DNA inside

the nucleus is highly ordered with different levels of organization DNA packaging starts with its interaction with an octamer of core histone proteins (two copies each of H2A, H2B, H3 and H4) called the 10nm chromatin structure Negatively charged DNA interacts with positively charged histone with around 147 bps wrapped around each histone octamer (103, 104) collectively known as nucleosome By interacting with linker histones these structures undergo condensation to form 30nm chromatin fiber and further interaction with non histone proteins condenses chromatin into coiled coil 300nm higher order structures (105-109) These structures further get remodelled to form compact heterochromatin (transcriptionally silent) and loosely wound euchromatin (transcriptionally

active) regions which finally results in the formation of “chromosome territories” (110-113)

Large scale reorganization also results in highly compacted form of chromatin which is inaccessible for transcription regulators Various chromatin modifications including acetylation, phosphorylation, sumolyation, methylation and ubiquitylation, maintain a dynamic balance between condensed and open chromatin (114) Currently the mechanism by which TFs find their gene targets in this highly complex chromatin arrangement is understood to be sliding and hopping mechanism along the DNA This non specific binding with DNA has been postulated to help in searching for the correct binding sequence (115)

Such a hit and trial method along with the complexity in chromatin organization argues alternative or specific approaches taken by the cell to organize genes and chromosome so as

to provide specificity to gene expression and easy accessibility for TFs

Using fluorescence in-situ hybridization (FISH) (116), the spatial arrangement of chromosomes in various cell types has been characterized Chromosome size has been elucidated as the first level of chromosome organization By separately labelling micro and macro chromosomes in chicken nuclei, it was observed that macro chromosomes largely position themselves at the periphery compared to micro chromosomes which were positioned more centrally (117) Even in human cells, size based positioning of chromosomes has been observed (118) Along with the size, the radial positioning of chromosomes also depends on gene density as observed in chicken (117), human (110, 119) and primates cells (120) Recently, correlation between chromosomal transcriptional activity and spatial organization has been modelled and experimentally shown (121) Using stem cells, it has been demonstrated that positions of chromosomes are developmentally regulated Large scale spatial repositioning of chromosomes was observed during adipocyte differentiation (122) Further, tissue specific chromosomal organization has been documented (123, 124) Various structural proteins also contribute to chromosome repositioning Nuclear lamina which includes lamin and lamin binding proteins mediates chromosomal positioning (125, 126) Along with chromosome positioning, gene clustering has also been considered to be an alternative approach for efficient and energetically favourable transcription regulation Genes sharing common transcription factor have been shown to cluster together in yeast (125) Active globin genes have been shown to co-localize in sites of active transcription (127)

Similar clustering has been observed for Hox genes(128) and Myc and Igh in erythroid cells (127) Furthermore, using a variety of techniques including 3C and 4C complemented with

FISH and high resolution quantitative imaging, association between mouse globin genes and numerous other genes has been elucidated in transcription factories (129) Inspite of all these developments in understanding the spatial arrangements of chromosomes, it is still largely unclear how physical signals mediated changes in nuclear dyamics and gene expression impinges on chromosomes arrangement and positioning

1.5 Objectives of the thesis:

As explained above, emerging evidences in mechano-biology have characterized the transmission of physical signals in the cells and have shown their impact on gene expression Most of the studies have probed the long term modulation in genetic profiles but an extensive genetic analysis during the early time points is unexplored This is highly required in the field of cell mechanics since it will help in characterizing the early mechano-sensitive genes and the critical biochemical pathways involved Further, the impact of physical signals on changing chromatin compaction, a pre-requisite for altering gene expression, is yet to be probed How these externally applied signals alters chromatin compaction and what are the crucial chromatin remodellers involved in this alteration are largely unclear Further, a link between physcial signals regulated gene expression, chromatin compaction and nuclear dynamics needs to be characterized In continuation, how these links impinges on the spatial reorganization of chromosomes needs to be characterized to understand the non-random organization of chromosomes In this thesis, I have used approaches like application of geometrical constraints on differentiated cells and differentiation of stem cells to answer these above mentioned questions

In chapter 2 of this thesis, using fibronectin coated micro-fabricated patterns of varying geometries along with whole transcriptome micro-array complemented with confocal imaging and quantitative analysis, I have tried to understand the early events involved in cell geometry (size, shape and aspect ratio) mediated alterations in chromatin compaction and its correlation with nuclear volume and gene expression Further, the geometry sensitive histone

deacetylases and transcription factors and co-factors which alters chromatin compaction and modulate early gene expressions respectively has been probed

In chapter 3, using stem cell differentiation and quantitative 3D Fluorescence in-situ hybridization, I have tried to link alterations in nuclear architecture and large scale modular changes in gene expression with chromosome activity Further, I have explored how the chromosome activity changes during the early onset of differentiation and couple with emergence of chromosomal positioning

Chapter 2: Cell geometric constraints induce modular gene expression pattern via redistribution of HDAC3 regulated by acto- myosin contractility

2.1 Introduction:

Cellular geometry impinges on both structural and functional properties of the cells which include nuclear deformation, cytoskeleton re-organization, chromatin compaction, gene expression, growth, apoptosis and cell division (130-134) Other physical cues like substrate stretching, fluid flow, substrate rigidity, cellular topography have also been shown to alter cellular morphology, nuclear architecture and gene expression (10, 135, 136) Recent studies have also shown that alterations in cellular geometry influence the fate of mesenchymal stem cells and human embryonic stem cells (43, 137, 138) Depending on the geometry of underlying substrate, these cells differentiate to multiple lineages including mesoderm and endoderm Perliminary studies has largely attributed these changes in differentiation potential to the changes in acto-myosin contractility generated by differential cellular geometry (43, 139, 140) Also it has been shown that acto-myosin contractility directly impinges on nuclear architecture (54, 141) Nucleus is a non-random entity and three dimensional organization of nucleus regulates the developmental programs Alteration in nucleus architecture modulates chromatin compaction state, which a prerequisite for changes in gene expression Changes in compaction state of chromatin is largely brought about by post-translational modifications of histone tails, which alter higher order chromatin assembly and hence the accessibility of gene regulatory sites by transcriptional machinery (142, 143) Changes in chromatin compaction is followed by its interaction with transcription factors and co-factors, key signaling intermediates, which shuttle between cytoplasm and nucleus and render specificity to the gene expression programs A crucial transcription co-factor which responds to mechanical cues is MRTF-A whose activity has recently been

implicated in nuclear mechanotransduction (99, 100) However, the molecular

mechanisms underlying control of chromatin compaction and gene expression by

geometric via modulation of cytoplasmic to nuclear redistribution of regulatory

molecules are still unclear

This chapter is focused on understanding the coupling between cellular geometry and

gene expression via differential shuttling of transcription regulators Micro-contact

printing was used to generate fibronectin coated patterns of different shape, size and

aspect ratio to study the nuclear mechanotransduction Whole transcriptome analysis of

cells cultured on these patterns revealed geometry dependent alterations in gene

expression profile The differentially regulated genes belong to the category of actin

related genes since promoter analysis of these genes showed that serum response factor

(SRF) was an essential regulatory factor sensitive to geometric cues Genomic programs

were found to be largely sensitive to changes in cell size Matrix related genes were

found to be upregulated in cells of larger area whereas reduced cell-substrate contact

resulted in up-regulation of genes involved in cellular homeostasis Changes in geometric

cues also resulted in differential modulation of nuclear morphology, acto-myosin

contractility and histone acetylation Simultaneously, it was also found that HDAC3, a

histone deacetylase which shuttle between cytoplasm and nucleus, modulate levels of

histone acetylation in an acto-myosin dependent manner Further, we show that

geometric control of serum response element (SRE) promoter activity was dependent on

transduction of its co-factor MRTF-A Lastly, we have shown that nuclear accumulation

of MRTF-A negatively regulates NF-κΒ activity Taken together, our work provides

mechanistic insights underlying the regulation of gene expression by cellular geometry

2.2 Materials and Methods:

2.2.1 Micro-patterning: Polydimethylsiloxane (PDMS) elastomer (SYLGARD 184,

DOW Corning, USA) was used in a 1:10 ratio of curative to precursor as per manufacturer’s protocol, to prepare stamps by moulding it in micro-fabricated silicon wafers Micro-patterned PDMS stamps were oxidized and sterilized under high power in Plasma Cleaner (Model PDC-002, Harrick Scientific Corp) 50µg/ml fibronectin solution was allowed to adsorb onto the surface of each PDMS stamp under sterile condition The PDMS stamp was then deposited onto the surface of hydrophobic dishes (ibidi, Germany)

to allow transferring of the micro-features Surface was then treated with 2mg/ml Pluronic F-127 (Sigma, USA) for 30 minutes to passivate non-fibronectin coated regions

2.2.2 Cell culture and drug treatment: NIH3T3 fibroblast cells were cultured in Low

Glucose DMEM (Life Technologies, USA) supplemented with 10% Fetal Bovine Serum (GIBCO, Life Technologies, USA) and 1% Antibiotic-Antimycotic (Life Technologies, USA), at 37°C in 5% CO2 Prior to seeding the cells on patterned substrate, Pluronic F127 was removed and dishes were rinsed twice with 1X PBS 65,000 cells were seeded for 45 minutes Unadhered cells were removed and the remaining cells were washed once with 1X PBS and 1 ml of DMEM supplemented with 10% FBS and incubated for 3 hours Actin filaments were depolymerized with either 1mM Cytochalasin-D (Sigma, USA) for 40 minutes or 500nM Latrunculin A (Sigma, USA) for 40 minutes Myosin-II motor activity was inhibited by treating cells with 50µM Blebbistatin (Sigma, USA) for

40 minutes For reporter assay, cells were treated with 500mM Latrunculin A for 90 minutes or 25µM Blebbistatin for 90 minutes ROCK activity was inhibited by treating cells with 50µM of Y-27632 (Sigma, USA) for 1 hour Histone Deacetylase activity was

inhibited by treatment with 200ng/ml Trichostatin A (Sigma, USA) for 18-24 hours For some experiments different concentration and time of treatment of the above mentioned drugs was used which are mentioned in the results

2.2.3 Immunostaining: Cells were rinsed twice with 1X PBS followed by fixation using

4% paraformaldehyde (Sigma, USA) in PHEM Buffer (500 ml of 2X PHEM buffer contain 18.14gm PIPES, 6.5gm HEPES, 3.8gm EGTA, 0.99gm MgSO4, pH=7 adjusted with KOH) for 20 minutes Cells were washed and permeabilized with 0.3% Triton-X (Sigma, USA) in 1X PBS for 10 minutes After washing twice with 1X PBS, the cells were treated with 5% BSA (blocking solution) for 1 hour This was followed by incubation with required primary antibodies (in blocking buffer) The primary antibodies used were: p-MLC (Cell Signaling 1:75), AcH3K9 (Abcam 1:400), MRTF-A (Santa Cruz 1:150), p65 (Abcam 1:400), HDAC3 (Santa Cruz 1:150), which were used to stain myosin light chain phosphorylated at serine 19, Histone H3 acetylated at lysine 9, MRTF-A, total p65 and HDAC3 respectively Cells were washed with blocking solution and incubated with corresponding secondary antibody along with Hoescht-33342 (1mg/ml 1:1000) for 45 minutes F-Actin was labeled using Phalloidin Alexa Fluor® 488

or 568 (Life Technologies, USA, 1:100) and G-actin was labeled using deoxyribonuclease-I, Alexa Fluor® 488 conjugate (Life Technologies, USA, 1:1000)

2.2.4 Imaging and analysis: Images of fully adhered single cells were taken using 710

Meta and Nikon A1R using either 40X, 0.7 NA air objective, 63X, 1.25 NA oil objective

or 100X, 1.4 NA oil objective Imaging conditions were kept similar in all the experiments To estimate protein levels, fluorescence images were captured on confocal microscope using 63X objective with z-step of 0.8µm For nuclear volume analysis,

images were captured using 100X objective, 3X zoom and z-step of 0.15µm For SRE reporter assay, the wide field fluorescence images were captured using 40X objective on Andor BV897 EMCCD camera (Andor, USA) The data was normalized and is presented

as Mean±S.E Among cells on different aspect ratios, values were normalized with mean

of AR 1:1 cells where as among cells of different shapes, values were normalized with mean of circular cells For cells of different sizes, values were normalized with mean of cells of area 500µm2 All the quantifications were done using MATLAB (Math works) or Image J (public-domain software developed at National Institutes of Health, Bethesda, MD)

2.2.5 Fluorescence correlation spectroscopy (FCS) sample preparation and measurement: PDMS elastomer was used in 1:4 ratio of curative to precursor as per

manufacturer’s protocol A thin uniform layer (8µm thickness) of PDMS was deposited

on glass bottom dish using spin coater (5000 rpm for 2 minutes followed by 1500 rpm for

30 seconds) PDMS was cured at 80oC for 2 hours The micro-patterning, as described above, was carried out on these dishes after UV treating them for 5 minutes Cells were cultured on these dishes for 3 hours before FCS experiments were carried out on Zeiss LSM 710 confocal microscope equipped with confocor3 set up using 40X, 1.2 NA, water immersion objective

The intensity time series I(t) values were collected for a period of 10sec intervals at an optimum experimental setting to avoid artefacts arising due to photo bleaching and to ensure high counts/particle The pinhole size was kept at 1A.U for 488-nm laser line excitation Pinhole was aligned using 10nM Atto 488 free dye solution in water The corresponding confocal diameter was ∼267nm, as determined from calibration using Atto

488 free dye solution in water (using 29.8µsec (experimentally determined) and 4x10-6

cm2/sec for 3D free τ and Diffusion constant respectively) 488nm line of Argon ion laser

was used for excitation From the intensity time series, the autocorrelation function,

𝐺 𝜏 = ! !!! ×! !! ! ! was calculated, where τ is the correlation time Using Zen2009

software (Zeiss, Germany), individual correlation curve were plotted and fit to one components 3D anomalous diffusion model to determine the diffusion timescale (τ) and anomaly parameter (α) Initial 9.8 µsecs of the correlation curves was not taken into fit procedure to account for triplet state timescale FCS measurement on serum starved cells was carried out on cells of random geometry cultured on non fibronectin coated glass bottom dishes to avoid induction of MRTF-A activity by fibronectin

2.2.6 Microarray sample preparation and analysis: Cells cultured on different

geometrical patterns were harvested for extracting the RNA to analyze gene expression patterns Cells cultured on unpatterned fibronectin were used as control for the microarray Microarray data analysis was performed using custom written algorithm in MATLAB For the microarray across different shapes and sizes, the RMA value for each gene was obtained in triplicates and their p-values were computed by student’s t-test Genes with p-value < 0.05 were considered to be significantly expressing and a cut-off of 1.2 fold was used to identify differentially expressing genes

The microarray data for circular and triangular cells were obtained in duplicates The replicate samples for each microarray were compared and a difference in their gene activities was estimated The mean (µ) and standard deviation (σ) of the difference was calculated and similarly expressing genes were selected such that the difference between

the replicates is within one σ from the mean of the difference The activity of each selected gene was estimated from the mean of the two replicates and this was used for all further analysis Logarithmic ratios were estimated for each gene of circle and triangle with respect to the unpatterned fibronectin control Differentially expressing genes were identified such that the -1 < logarithmic ratio < 1 and the activity of the upregulated gene

is >500 For a direct comparison between circles and triangles, logarithmic ratio for triangle was estimated with respect to circle and the criteria for selection differentially expressed genes was similar to the criteria described above

2.2.7 Transcription factor binding site analysis: Analysis of transcription factor

binding sites, upstream of the promoter regions of the differential genes was performed using the search engine, MAPPER-2 (144) Initial run of the genes, using a region 5000bp upstream of the promoter to 500bp downstream provided scores and E-values which were statistical measures of binding efficiency of the TF to the promoter sites Score is a measure of the strength of binding and E-value provides the false discovery rate involved in the estimation of the score Any TF binding prediction was considered significant, if the scores were large and the corresponding E values were small We obtained the distribution of scores and E-values for all the identified TFs Then, mean and standard deviation was calculated for the distribution of scores (µs,σs) and E-values (µe,σe) Significant TFs were selected as:

𝑆𝑐𝑜𝑟𝑒 > µ!+ 1.75𝜎!

𝐸𝑉𝑎𝑙𝑢𝑒 < µ!− 1.75𝜎!

In order to estimate the statistical significance of the TF binding sites, 11 genes were selected randomly from the genome and the TF binding sites were identified by using the

same thresholds for scores and E-value as used for the 11 differentially regulated genes in triangle compared to circles TF binding sites were identified for 86 such random selections and a false discovery rate (FDR) was obtained The TFs with FDR < 0.05 were considered to be statistically significant Similar analysis was performed to identify the significant TF binding sites in genes upregulated in circles compared with triangles

2.2.8 Functional analysis of genes: The functional analysis of upregulated and

downregulated genes was performed using DAVID (145, 146) Based on Gene ontology annotations genes were organized and plotted as pie charts

2.2.9 Real time analysis: cDNA was prepared using SuperScriptIII First strand synthesis

system (Life Technologies, USA) Real time experiment was performed using SsoFast EvaGreen Supermix (Bio-Rad #172-5203) in CFX96 Real time machine (Bio-Rad)

2.2.10 Reporter assay Serum response elements (SRE) or NF-kB reporter assay was

performed using Cignal reporter assay kit (Qiagen, USA) Reporter plasmid has inducible transcription factor-responsive EGFP reporter, encoding green fluorescent protein under the control of a basal promoter element (TATA box) joined to tandem repeats of SRE or NF-κB Constitutively expressing EGFP gene under CMV promoter was used as positive control For reporter activity measurements, reporter or positive control plasmids were transfected in cells and plated on cells of different geometries Reporter activity in cell was normalized with corresponding CMV reporter activity for all the geometries

2.2.11 Cell proliferation assay Percent cells in the S phase were calculated for cells on

triangular and circular patterns (1800µm2), cultured for 9 hours, by quantifying the

incorporation of 5-bromo-2’-deoxyuridine (BrdU) into cellular DNA using an in situ cell proliferation kit (Roche Applied Science)

2.3 Results:

2.3.1 Cell geometry impose modular changes in gene expression: To understand the

changes in gene expression imposed by cellular geometry, whole transcriptome analysis using microarray was performed on cells grown on un-patterned or on different geometries As shown in Fig 2.1A and B, NIH3T3 cells acquire a normal spreading area

of around 850+30µm2 when plated on plastic substrate whereas on fibronectin substrate, the mean spreading area is around 1300+30µm2 As can be seen in the area distribution plot (Fig 2.1A), cells can spread greater than 2000µm2 on fibronectin However, when cells were cultured on geometric patterns, cells acquire the shape and size of the underlying fibronectin coated patterns To ensure that the single cells are spread well, we choose the geometric patterns of area 1800µm2, larger than the mean spreading area of 1300µm2 Geometric constraints were imposed on NIH3T3 cells by culturing them on plastic substrates with fibronectin coated micropatterns These micropatterns were either rectangular (aspect ratios 1:1 to 1:5) or of different shapes of equal area (1800µm2) or same shape with varying areas (500-2000µm2) Fig 2.1C shows NIH3T3 cells cultured

on these different patterns

In order to avoid heterogeneity and uncertainty in gene expression arising due to occupancy, we standardized the cell seeding density for all the patterns of different geometries Multi-occupancy of cells on the pattern results in differential tension between cells; cells at the edges have higher acto-myosin contractility while cells in the centre have reduced acto-myosin contractility Such variability in mechanical tension results in the variability in gene expression and therefore we ensured single cell attachment per pattern which was then quantified Fig 2.2A showed fibronectin (conjugated with Alexa

multi-Flour 488) coated pattern with NIH3T3 cells cultured on them and stained with Phalloidin As can be seen in Fig 2.2B, more than 80% of micropatterns had single cells with a small fraction of patterns (15%) harbouring multiple cells Fig 2.3A and B shows the field images of cells of different shapes (circle and triangle respectively) but of equal area 500µm2 After 3 hours of cell plating, more than 90% of the patterns were found to have singlet cells spreading on them Single cell on individual pattern is marked with red boundary Similarly, Fig 2.3C shows field view of cells of AR 1:5 (area 1800µm2) Again the numbers of singlets (marked with red boundary) obtained were around 80% after 3 hours of cell plating (Fig 2.3D)

Figure 2.1: Sculpting single cell geometry: (A) Representative images of NIH3T3 cells on

plastic substrate and on fibronectin substrate (B) Normalized distribution of spreading area of NIH3T3 cells on plastic substrate (n=150) and fibronectin substrate (n=200) (C) DIC images of NIH3T3 cells of different aspect ratio or different shapes but equal cell area 1800µm2 or NIH3T3 cells of same shape (triangle) but increasing spreading area (500 to 2000µm2) Nucleus

is stained with Hoechst Scale bar =10µm

Figure 2.2: Standardization of cell seeding density: (A) Cells grown on triangular patterns

of area 1800µm2 coated with Alexa Fluor 488 conjugated fibronectin and stained with phalloidin 65,000 cells were seeded on 35 mm untreated tissue culture dishes and incubated for 45 minutes before washing them (B) The protocol used for seeding cells resulted in 80% efficiency of obtaining singlet in culture Data is given as Mean±S.E

Figure 2.3: Standardization of cell seeding density: (A) Cells grown on triangular pattern

of area 500µm2 (B) Cells grown on circular pattern of area 500µm2 (C) Cells grown on AR 1:5 of area 1800µm2 65,000 cells were seeded on 35 mm untreated tissue culture dishes and incubated for 45 minutes before washing them (D) The protocol used for seeding cells resulted in 80% efficiency of obtaining singlet in culture Data is given as Mean±S.E Scale Bar= 400µm

2D matrix was plotted to see the number of genes differentially regulated between cells

of different geometries As can be seen in Fig 2.4A, there is a distinct separation in the number of differentially regulated genes depending on cellular geometry Cell size was found to impinge more on the number of differentially regulated genes as compare to cells of different shapes or aspect ratio Expression of more than 250 genes was altered upon change in cell size where as 70-150 genes were altered between cells of different shape or aspect ratio (of equal sizes) Gene ontology (GO) analysis was than done to get

an insight into the molecular consequences of these differentially expressed genes The

GO analysis cluster the genes based on cellular component, biological process and molecular function Since NIH3T3 cells are polarized in shape, under physiological condition, GO analysis between cells plated on small circle were compared with AR 1:5

of larger size (Note# Throughout this chapter small size refers to patterns of area 500µm2

and larger size refers to patterns of area 1800µm2) GO groups were assigned to regulated and down-regulated genes separately Genes involved in regulation of cell division, apoptosis and programmed cell death were found to be upregulated in cells of smaller circle (Fig 2.4B) Contrary to this, the main GO groups for the genes up-regulated in cells of AR 1:5 (area 1800µm2) falls into the category of cell migration, cell-substrate adhesion and actin-cytoskeleton

up-In contrast, comparison between small triangle with larger ones belong to regulation of gene expression, negative regulation of transcription and chromatin DNA binding (Fig 2.4C) The main GO groups for the genes up-regulated in bigger sizes falls into the category of cell motility, cell motion, regulation of cell proliferation, differentiation and communication and cell-substrate junction

Figure 2.4: Cell geometry imposes modular changes in gene expression: NIH3T3

fibroblast cells cultured on fibronectin coated patterns of different sizes, shapes or aspect ratio (Aspect ratio is defined as cell adhering to micropatterns of similar area but different ratio of short axis to long axis) (A) 2D matrix showing total number of differentially regulated genes among cells of different geometries (B) Gene ontology (GO) analysis of 290 differentially regulated genes in small circle as compared to larger cells of AR 1:5 The pie-chart includes significantly (p<0.05) represented gene clusters (25 up-regulated and 20 down-regulated genes) (C) Gene ontology (GO) analysis of 200 differentially regulated genes in small triangle as compared to large triangle The pie-chart includes significantly (p<0.05) represented gene clusters (42 up-regulated and 34 down-regulated genes)

2.3.2 Actin and related genes are largely sensitive to changes in cellular geometry: A

direct comparison was than performed to get an insight into the genes differentially regulated among cells of different geometries Top 15 up-regulated and down-regulated genes between small triangle and big triangle, square (AR 1:1) and rectangle (AR 1:5) of equal area (1800µm2) and triangle and circle of equal area (1800µm2) respectively are shown in Fig 2.5A, B and C Gene expression profiles between cells of varying sizes revealed actin and actin related genes to be significantly altered For example, actinin, an actin cross linker, was found to be upregulated in cells of larger area (Fig 2.6A) Simultaneously, genes directly involved in actin polymerization like Rho and formin were also found to be significantly upregulated upon increase in cellular spreading area (Fig 2.6A)

An important transcription co-factor, known to be involved in direct regulation of a number of actin related genes and whose activity is dependent on actin dynamics, is MRTF-A MRTF-A, is a nuclear to cytoplasmic shuttling transcription co-factor which interact with serum response factor (SRF) to bring about changes in gene expression

(147-150) Consistent with this, MRTF-A regulated genes like Zyxin, Actin-gamma 3,

fibronectin were found to be differentially regulated based on cell size (Fig 2.6A) As

already shown in 2D matrix (Fig 2.4A) that the cellular shape had less substantial impact

on gene expression, the analysis of annotated promoter sequences of these differentially expressed genes also revealed binding sites for SRF (Fig 2.6B) In addition, promoter analysis revealed Peroxisome Proliferator Activated Receptor alpha (PPAR alpha) and Hepatocyte Nuclear Factor-3 Homologue 1 (HFH-1) binding sites were significantly enriched in genes differentially regulated in triangles as compared to circles of equal

sizes (Fig 2.6B and C) The significance was estimated by false discovery rate measurement performed on random selection of annotated genes from the genome (Fig 2.6D and E) The list of genes regulated by the identified transcription factors (TFs) is shown in Fig 2.6F The whole transcriptome analysis concluded that the changes in cell size have maximum bearing on gene expression as compared to changes in cell shape or cellular aspect ratio Simultaneously, it also coupled actin and its related genes with changes in cellular geometry Next, correlative changes between nuclear architecture and histone acetylation levels were carried out to understand the signaling intermediates regulating these alterations in gene expression profiles

Figure 2.5: Cell geometry based differential gene expression: Color maps showing

differentially regulated genes between cells of different geometries (A) size (B) aspect ratio (of equal area) and (C) shape (of equal area)

Figure 2.6: Geometric regulation of actin related genes and transcription factors:

(A) Actin related genes differentially expressed between cells cultured on smaller and larger triangular patterns (B) Transcription factor (TF) showing significant binding sites in differentially regulated genes between circular and triangular cells of equal area (1800µm2) (C) When a particular gene (Acta1 in this case) was mined for transcription

factor binding sites, binding score and its E-Value was obtained Scatter plot of score vs

E value shows the threshold used to select the significant binding sites Each point in the plot denotes a TF The points in red are the TFs with significant binding efficiency and the rest of the TFs are shown in black (D) Distribution of binding sites of SRF for 86 random selections of 11 genes to estimate the false discovery rates Black arrow indicates the actual number of binding sites (E) False discovery estimation for binding sites of the transcription factor HFH1 (F) Table showing the genes regulated by the identified TFs and the number of binding sites on the gene

2.3.3 Cellular geometry impinges on nuclear architecture: In order to probe the

changes in nuclear geometry upon changes in cellular geometry, nuclear height was characterized by imaging cells cultured on different micropatterns, labelled with hoescht (a nuclear stain) Orthogonal views and quantification of nuclear height clearly shows that nuclei of circular cells have higher nuclear height and lower cross-sectional area as compared to nuclei of triangular cells of similar area (Fig 2.7A and B) Similarly, cells of

AR 1:1 showed higher nuclear height and lower nuclear cross-sectional area as compared

to cells of AR 1:5 of similar area (Fig 2.7C and D) Nuclear projected area (Npa) was further characterized which also shows large scale changes between nuclei of cells of different geometries Quantification of Npa in cells of different geometries (circle, pentagon, hexagon, square and triangle) (equal area 1800µm2) revealed a significant increase of 25±5% in Npa between cells cultured on circular and triangular fibronectin patterns (Fig 2.7E) Similarly, comparison across cells of varying aspect ratios (equal area 1800µm2) also showed an increasing trend whereby cells of AR 1:1 showed lower

Npa as compared to cells of AR 1:5 (Fig 2.7F) For cells of same shape but varying spreading area, a monotonic increase in Npa was observed as the cell spreading increases (Fig 2.7G) consistent with a previous study in yeast cells (151) Cell of spreading area 2000µm2 showed twice Npa as compared to cell of spreading area 500µm2 Next, we plotted the variations in nuclear height and Npa across different geometries

Figure 2.7: Cell geometry impinges on nuclear height and projected nuclear area: (A & B)

Orthogonal views of nuclei of circular and triangular cells of equal area show higher nuclear height and lower cross-sectional area in circular cells as compared to triangular cells (C & D) Similarly, higher nuclear height and lower nuclear cross-sectional area were observed in cells of AR 1:1 as compared to cells of AR 1:5 of equal area (E) Quantification of projected nuclear area (Npa) in cells of different geometries (equal area 1800µm2) revealed a significant increase of 25±5% in Npa between cells cultured on circular and triangular fibronectin patterns (F) Similar changes in Npa were observed between cells of AR 1:1 and 1:5 of equal area, where cells of AR 1:1 showed slightly smaller Npa as compared to cells of AR 1:5 (G) Npa was seen to monotonically increase with an increase in cell area for a given shape Data is given as Mean±S.E with 50<n<100 Scale bar=10µm * represent p-value: * p<0.05, ** p<0.01, *** p<0.001

As evident from Fig 2.8A, nuclei of triangular cells showed higher Npa and smaller nuclear height as compared to nuclei of circular cells of equal area 1800µm2 Fig 2.8B shows the same variation of Npa and nuclear height between cells of different aspect ratio

of equal area 1800µm2 Further, circularity and aspect ratio of the nuclei were quantified

Circularity was quantified as 4 2

Perimeter

Area

π As shown in Fig 2.8C, increase in cellular

aspect ratio, keeping the cell area same, resulted in an increase in nuclear aspect ratio whereas the circularity of the nucleus decreases However, both these parameters remained unchanged with changes in cellular shape or size Fig 2.8C and D showing that nuclear shape is largely sensitive to cellular aspect ratio

Next, we characterized nuclear volume It was observed that as the cell size increases, nuclear volume also increases Cells cultured on smaller patterns showed significantly smaller nuclear volume as compared to cells cultured on larger patterns (Fig 2.9D) Contrary to this, we observed insignificant changes in nuclear volume with changes in cell shape, in accordance with previous studies in endothelial cells (65), (Fig 2.9B) or

AR (Fig 2.9C) This suggests that the nuclear volume is not sensitive to changes in cell shape or aspect ratio but responds to changes in cell spreading area Since cell geometry (area, shape and aspect ratio) impinges on nuclear morphology, we next characterize whether these changes correlate with global changes in histone acetylation, a marker of decondensed chromatin assembly enabling access to gene regulatory sites

Figure 2.8: Cell geometry impinges on nuclear size and shape: Variation in nuclear area

as a function of Z depth between cells of different shape of equal area (A) or aspect ratios

of equal area (B) (C) Quantitation of nuclear circularity and aspect ratio as a function of cell aspect ratio and shape and (D) cell spreading area Data is given as Mean±S.E with 50<n<100 Scale bar=10µm * represent p-value: * p<0.05, ** p<0.01, *** p<0.001

2.3.4 Correlation between nuclear volume and histone acetylation: To directly probe

the effect of cell geometry on histone acetylation, we measured the levels of histone H3 acetylated at lysine 9 (AcH3K9) AcH3K9 has been characterized as marker for chromatin decompaction Concomitant with increase in nuclear volume, as increase in AcH3K9 levels was observed with increase in cell area (Fig 2.9D) But for a fixed cell area, the global acetylation levels were independent of cellular shape or aspect ratio where the nuclear volume remains constant (Fig 2.9B and C) Apparent chromatin spatial density, a measure of average chromatin compaction ratio (152), remained constant between cells of different shape of equal area (Fig 2.9E and F), but decreases with an increase in cell area (Fig 2.9G) suggesting lower chromatin compaction in cells

of larger spreading area

To probe the coupling between nuclear volume and histone acetylation, cells on larger triangular pattern were treated with Trichostatin A, a drug which hyper-acetylates core histones by inhibiting HDACs TSA treatment results in increased total histone acetylation levels and was used as a positive control (153) TSA treated cells showed an increase in AcH3K9 levels which correlated with a significant nuclear volume increase (Fig 2.10A and B) Further, to ascertain the interdependence between nuclear volume and AcH3K9 levels, cells were treated with Cytochalasin-D (Cyto-D) Cyto-D is a well characterized pharmacological inhibitor which depolymerizes actin and hence decreases nuclear volume, as has been previously characterized (54) Cyto-D treatment of cells cultured on larger triangular pattern showed significant decrease in both nuclear volume and AcH3K9 levels (Fig 2.10A and B) A log vs log plot between normalized nuclear volume and histone acetylation levels shows linear interdependence between them with r

= 0.75 (Fig 2.10C) The above results suggest that cellular geometry regulates both nuclear volume and histone acetylation which is largely sensitive to changes in cell size

We next assess the molecular intermediates involved in this geometry dependent regulation of global histone acetylation levels

2.3.5 Acto-myosin contractility governs histone acetylation by regulating spatial redistribution of HDAC3: In order to probe the effect of cell geometry in modulating

histone acetylation levels, we analyzed HDAC3 HDAC3 is a class I histone deacetylase, majorly involved in altering chromatin structure, which has histone and non-histone binding substrate (154, 155) HDAC3 has been shown to be present both in nucleus and cytoplasm (156-160) To check whether HDAC3 is involved in modulating levels of AcH3K9, cells were transfected with HDAC3-EGFP and levels of AcH3K9 were quantified As shown in Fig 2.11A and B, cells transfected with HDAC3-EGFP showed marked reduction in the levels of AcH3K9 Intensity line profile of two cells (Fig 2.11A), one transfected with HDAC3-EGFP and other non-transfected, clearly shows the difference in the AcH3K9 levels between them Further, we characterized the nuclear translocation dynamics of HDAC3 In accord with previous studies, TNF-alpha treatment which degrades IκB-alpha complex resulted in significant nuclear translocation of HDAC3 (Fig 2.11C) The kinetics follows a sigmoidal dynamics with time constant 49±24 minutes (Fig 2.11D)