Chester Clinic

 

Effects of substrate stiffness on cell morphology, cytoskeletal structure, and adhesion

Cell Motility and the Cytoskeleton 60:24 –34 (2005)
Effects of Substrate Stiffness on Cell Morphology, Cytoskeletal Structure, Tony Yeung,1 Penelope C. Georges,1 Lisa A. Flanagan,2 Beatrice Marg,2 Miguelina Ortiz,1 Makoto Funaki,1 Nastaran Zahir,1 Wenyu Ming,1 Valerie Weaver,1 and Paul A. Janmey1,2* 1Institute for Medicine and Engineering, University of Pennsylvania, Philadelphia 2Hematology Division, Brigham and Women's Hospital, Boston, Massachusetts The morphology and cytoskeletal structure of fibroblasts, endothelial cells, andneutrophils are documented for cells cultured on surfaces with stiffness rangingfrom 2 to 55,000 Pa that have been laminated with fibronectin or collagen asadhesive ligand. When grown in sparse culture with no cell-cell contacts, fibro-blasts and endothelial cells show an abrupt change in spread area that occurs at astiffness range around 3,000 Pa. No actin stress fibers are seen in fibroblasts onsoft surfaces, and the appearance of stress fibers is abrupt and complete at astiffness range coincident with that at which they spread. Upregulation of ␣5integrin also occurs in the same stiffness range, but exogenous expression of ␣5integrin is not sufficient to cause cell spreading on soft surfaces. Neutrophils, incontrast, show no dependence of either resting shape or ability to spread afteractivation when cultured on surfaces as soft as 2 Pa compared to glass. The shapeand cytoskeletal differences evident in single cells on soft compared to hardsubstrates are eliminated when fibroblasts or endothelial cells make cell-cellcontact. These results support the hypothesis that mechanical factors impactdifferent cell types in fundamentally different ways, and can trigger specificchanges similar to those stimulated by soluble ligands. Cell Motil. Cytoskeleton60:24 –34, 2005.
2004 Wiley-Liss, Inc.
Key words: substrate stiffness; cell morphology; fibroblasts; mechanosensing; cell-matrix interaction;
actin cytoskeleton; integrin expression
transcriptional changes mediating cells' responses tostiffness are beginning to be characterized.
Most cells in multicellular organisms are at- Previous studies using fibroblasts have shown that tached to much softer materials than the glass and cells generate more traction force and develop a broader plastic surfaces on which nearly all studies are done in and flatter morphology on stiff substrates than they do on vitro. The most common attachment site for a mam-malian cell is another similar cell or the extracellularmatrix, and these materials have elastic moduli on the order of 10 to 10,000 Pa [Bao and Suresh, 2003; Correspondence to: Paul Janmey, Institute for Medicine and Engineering (IME), University of Pennsylvania, 1010 Vagelos Laboratories, 3340 Wakatsuki et al., 2000]. Forces generated by cytoskel- Smith Walk, Philadelphia, PA 19104.
etal motors applied to membrane attachment sites can deform materials with this range of stiffness but can-not move an attachment site on a rigid surface. Con- Received 22 April 2004; accepted 27 August 2004 sequently, cell morphology and functions can depend Published online in Wiley InterScience (www.interscience.wiley.
strongly on substrate stiffness under conditions where chemical signals are constant. The acute signal and DOI: 10.1002/cm.20041 2004 Wiley-Liss, Inc.
Cell Morphology and Substrate Stiffness
soft but equally adhesive surfaces and that cells will Preparation of Polyacrylamide Gel
preferentially migrate from a soft to a hard surface [Lo et A 30% w/v acrylamide stock solution is prepared al., 2000]. Cell growth and survival also depend on by mixing 15 g of acrylamide powder (FisherBiotech substrate stiffness, as gelatin-coated materials softer than CAS no. 79061) with 35 mL of deionized H O. A 100 Pa do not support survival of non-transformed cells 1% w/v bis-acrylamide stock solution is prepared by but do support growth of transformed cell types [Wang et mixing 500 mg bis-acrylamide powder (FisherBiotech al., 2000a]. Large changes in the extent of protein ty- CAS no. 110269) with 49.5 mL H O. Polyacrylamide gel rosine phosphorylation and cytoskeletal structure corre- solutions are prepared with acrylamide at final concen- late with stiffness, and are presumably part of the stiff- trations of 3, 5, 7.5, or 12% w/v and bis-acrylamide from ness-sensing apparatus [Pelham and Wang, 1997].
0.05 to 0.6% w/v. To polymerize the solution, 1.5 ␮L Responses to mechanical stimuli may be cell-type spe- TEMED (FisherBiotech CAS no. 110189) and 5 ␮L of cific. Motor neurons derived from embryonic mouse 10% ammonium persulfate are added with the appropri- spinal cord extend neurites with extensive branches on ate amount of H O to yield a final volume of 1,000 ␮L.
soft but not hard surfaces [Flanagan et al., 2002]. In A fixed volume of 45 ␮L of the polyacrylamide solution contrast, smooth muscle cells, like fibroblasts, extend is immediately pipetted onto the center of the 25-mm- processes more avidly on hard surfaces and are rounded diameter glass cover slip. The 18-mm-diameter cover on soft materials [Engler et al., 2004].
slip is then carefully placed on top of the polyacrylamide In this study, we extend the range of cell types solution. The polymerization is completed in about examined on materials with well-controlled stiffness by 10 min and the top coverslip is slowly peeled off. The employing protein-laminated polyacrylamide gels to bottom cover slip with the attached polyacrylamide gel is study GFP-actin and GFP-␣5 integrin-expressing NIH immersed in a multi-well plate (6 Well Cell Culture 3T3 fibroblasts, bovine aorta endothelial cells, and hu- Cluser, Costar 3516, Corning Inc., Corning, NY) with man neutrophils. The results show that the morphologies 3 mL PBS. Polyacrylamide gels with a wide range of of different cell types differ both quantitatively and qual- shear storage moduli, G⬘, can be prepared using the itatively with substrate stiffness, that the stiffness re- concentration scheme described above. The most flexible sponse depends on the nature of the adhesion ligand gel with 3% w/v acrylamide and 0.05% w/v bis-acryl- bound to the surface, and that changes in stiffness can amide has a shear elastic modulus (G⬘) of 2 Pa while the initiate specific transcription events leading to upregula- stiffest gel with 12% w/v acrylamide and 0.60% w/v tion of adhesion receptors.
bis-acrylamide yields a G⬘ of 55,000 Pa.
MATERIALS AND METHODS
Crosslinking of Adhesion Proteins
Preparation of Glass Cover Slips
A heterobifunctional crosslinker, sulfo-SANPAH The substrate is prepared by allowing polyacryl- amide solutions to polymerize between two chemically anoate, Pierce no. 22589), is used to crosslink extracel- modified glass cover slips. Briefly, 200 ␮L of a 0.1 N lular matrix molecules onto the surface of the gel. A NaOH solution is pipetted to cover the surface of a small amount (about 1 mg/ml) of sulfo-SANPAH is 25-mm-diameter glass cover slip (Fisherbrand, catalog dissolved in H O, and 200 ␮L of this solution is pipetted no. 12-545-102; Fisher Scientific, Pittsburgh, PA) for onto the gel surface. The polyacrylamide gel is then 5 min. The NaOH solution is then aspirated, and 200 ␮L placed 6 inches under an ultraviolet lamp and irradiated of 3-APTMS (3-Aminopropyltrimethoxysilane, Sigma for 10 min. It is then washed three times each with 3 mL no. 28-1778, Sigma, St. Louis, MO) is applied for 3 min.
of 200 mM HEPES at pH 8.6. After the last HEPES The glass cover slip is thoroughly rinsed with de-ionized solution is aspirated, 200 ␮L of a 0.14 mg/ml fish fi- water to wash away any remaining 3-APTMS solution.
bronectin solution (Sea Run Holdings, South Freeport, Then, 200 ␮L of 0.5%v glutaraldehyde (Sigma no.
ME) [Wang et al., 2000b] or 0.14 mg/ml type I collagen G7651) in H O is added onto the cover slip for 20 min.
is pipetted on top of the polyacrylamide gel. The multi- The glass cover slip is subsequently rinsed with water.
well plate housing the gels is then incubated at 5°C for at An 18-mm-diameter glass cover slip is placed on top of a piece of parafilm inside a tissue culture dish. A small amount of a 10% by volume Surfasil solution (Pierce no.
42800, Pierce, Rockford, IL) in chloroform is pipetted The viscoelastic properties of polyacrylamide gels onto the parafilm near the cover slip. The tissue culture were quantified by measuring the dynamic shear moduli dish with a half-closed lid is placed inside a vacuum using an RFS II fluids spectrometer (Rheometrics Inc., desiccator for 10 min.
Piscataway, NJ). A 500-␮L sample was polymerized Yeung et al.
between two 25-mm stainless steel parallel plates with a chamber using a Bio-Rad Econo-Column pump (Bio- corresponding sample thickness of approximately 1 mm.
Rad, Richmond, CA). Microscopy began within 3 min The shear storage modulus G⬘, corresponding to the after flow was stopped. 3T3 fibroblasts were allowed to elastic resistance of the gels, was determined from the grow on either a 180- or 55,000-Pa gel. Cells on the shear stress in phase with an oscillatory (1 rad/s) shear 55,000-Pa gel were incubated with CO strain of 2% maximal amplitude, by standard techniques.
medium (GibcoBRL Cat No. 18045-088 Lot No.
Similar measurements were conducted using cone and 1099754) with 10% bovine calf serum to maintain phys- plate geometries where the thickness of the sample was iological pH in the absence of a rich CO environment.
between 50 –200 ␮m, similar to that of gels on which Cells on the 180-Pa gel were incubated with 50% CO2 cells are grown, and nearly identical G⬘ were recorded independent medium, 40% DMEM, and 10% bovine calf (data not shown).
serum. The rate of cell spreading was analyzed by takingimages of sparsely distributed cells at 30-sec intervals.
Cell Culture
GFP-actin expressing NIH-3T3 mouse fibroblasts Quantification of Integrin Expression
prepared as previously described [Cunningham et al., NIH 3T3 GFP-actin expressing fibroblasts grown 2001] were incubated in Dulbecco's Modified Eagle Me- on polyacrylamide gels with varying stiffnesses for 24 h dium (Cellgro no. 10-013CV) with high glucose, L- were used for the quantification of ␣5 integrin expression glutamine, no sodium pyruvate, phenol red, and 10% by Western blotting. Polyacrylamide gels on top of the bovine calf serum (Hyclone Cat No. SH30072.03 Lot 25-mm glass cover slips were gently scraped off in one No. AJH10711 Bottle No. 1007, Hyclone, Logan, UT) at piece by a razor blade. The gels were immersed in a a 5% CO environment.
Laemmli lysis buffer solution (50 mM Tris-HCL, pH 6.8, Bovine aortic endothelial cells were kindly pro- 5 mM EDTA, 2% SDS containing 1 mM sodium fluo- vided by Peter Davies, and incubated in DMEM with ride, 1 mM sodium orthovanadate, and a cocktail of high glucose, L-glutamine, no sodium pyruvate, phenol protease inhibitors) in an Eppendorf tube for 15 min. The red, and 10% bovine calf serum at a 5% CO environ- cell lysate solution with the addition of non-reducing buffer was then run on an 8% SDS-PAGE gel. The Human blood neutrophils were isolated from whole electrophoresis gel was then transferred and blocked blood from healthy volunteers. Collected blood samples using standard protocols. The primary antibody used was were immediately transferred to sodium-heparin-contain- ␣5 integrin, rabbit sera (Chemicon International, Te- ing polypropylene tubes. Neutrophils were isolated by mecula, CA), and the secondary used was a horseradish centrifuging 3 mL whole blood through 3 mL Mono-Poly peroxidase-linked anti-rabbit IgG polyclonal antibody resolving media (ICN Biomedical, Irvine, CA) to resolve (Amersham Pharmacia Biotech, Arlington Heights, IL).
a distinct neutrophil layer. Neutrophils were washed Protein was detected with an ECL-Plus system (Amer- two times with Dulbecco's phosphate-buffered saline sham Pharmacia Biotech). In order to normalize the (DPBS) (Gibco, Gaithersburg, MD), and contaminating Western blot result by the number of cells per sample, erythrocytes lysed by 40-sec incubation in sterile dis- microscopy was performed to estimate the number of tilled water. Final neutrophil pellets were resuspended in cells grown on each gel substrate before the Western blot RPMI 1640 medium (Gibco, Gaithersburg, MD) and procedure. The intensity value of each band after sub- kept on ice prior to plating on substrates.
traction of the background was then normalized by thenumber of cells estimated in each sample.
Phase contrast and fluorescence microscopy were conducted using a Leica DM IRBE microscope and a The ␣5 integrin EGFP fusion (generous gift of Rick Hamamatsu C-4742 digital camera. The multi-well plate Horowitz, University of Virginia) [Laukaitis et al., 2001] was briefly removed from the tissue culture incubator for was excised as a XhoI-NotI fragment from pEGFP-N3 microscopy work. Area and circumference measure- (Clontech, Palo Alto, CA) and subcloned into the retro- ments were obtained by tracing cell boundaries manually viral vector Hermes HRS puro GUS (generous gift of using NIH Image software.
Helen Blau, Stanford University, CA) [Rossi et al., 1998]replacing the GUS cDNA between SalI and NotI sites using standard techniques and bringing the fusion under control of a tetracycline (tet) regulated promoter. To FCS2 Chamber) was used for time-lapse movies to main- produce retrovirus, 3 ⫻ 106 HEK 293 cells were seeded tain temperature at 37°C. Two milliliters of a 10,000 cell/ in gelatin-coated 60-mm dishes, and 24 h later the ml solution were injected into the temperature-controlled Hermes HRS puro ␣5 integrin EGFP construct was co- Cell Morphology and Substrate Stiffness
transfected with pVSVG and pCgp (generous gift ofAlan Kingsman, Oxford University, UK), which expressthe VSVG protein and gag-pol genes, respectively, usingcalcium phosphate. Transfected cells were incubated 8 hat 37°C, media was replaced and returned to 37°C for afurther 16 h, after which they were transferred to a 32°Chumidified incubator. After 24 h at 32°C media wascollected, made 8 ␮g/ml in polybrene (Sigma), and cen-trifuged at 3,000g for 5 min to pellet cells and debris.
NIH-3T3 cells in a 6-cm dish were incubated with 1.5 mlof retrovirus containing supernatant at 32°C for 8 h, afterwhich the supernatant was replaced with 3 ml of regularmedia and returned to a 37°C humidified incubator for36 – 48 h. Retrovirally transduced cells were selected inmedia supplemented with 1 ␮g/ml puromycin until a Mechanical properties of polyacrylamide substrates. The resistant population grew out. Polyclonality was judged shear modulus of polyacrylamide gels with a range of acrylamide visually by the number of puromycin-resistant colonies (indicated as percents near data lines) to bis-acrylamide (indicated as and speed of outgrowth, and only pooled populations crosslinker) proportions was measured. The shear modulus (G⬘), ex- deemed sufficiently polyclonal were used in experiments.
pressed in Pascal, increases at constant polymer mass with increasing Cells from the pooled population were subsequently in- crosslinker. Increasing the concentration of acrylamide from 3 to 12%also creates a large stiffness range from 10 to 50,000 Pa. The solid line fected as above with a high titer MFG-based retrovirus denotes the theoretical stiffness of a rubberlike network if every that expresses the tet repressor fused to the HPV16 crosslink was elastically effective.
activation domain (a derivative of the tet off transactiva-tor construct provided by Helen Blau [Rossi et al., 1998],generated in the laboratory of V. Weaver, unpublished is varied. Gels with a concentration above 7.5 % in partic- data, which has wild type DNA binding activity), pre- ular show a remarkably linear dependence of elastic mod- pared by transfection in 293GPG cells (generous gift of ulus on crosslinker concentration spanning nearly two or- Richard Mulligan, Whitehead Institute, MIT, Cambridge, ders of magnitude in stiffness. For crosslinked rubberlike MA) as described in Ory et al. [1996]. Transduced cells networks like these acrylamide gels, the elasticity is theo- were maintained in 1 ␮g/ml tetracycline to keep ␣5 retically predicted to be related to the concentration of integrin EGFP in the uninduced state. Expression of crosslinks by the relation ␣5 integrin EGFP was induced in cells by culturing in the absence of tetracycline in medium containing Tet SystemApproved FBS (Clontech) for 48 h prior to experiments.
where n is the number of crosslinks per volume V, and Ris the gas constant [Flory, 1953]. This theoretical limit, EGFP Expression Analysis
shown by the solid line in Figure 1, is very close to the Cells were directly fixed using 2% paraformalde- experimental values measured at high total [polyacryl- hyde and visualized using a scanning confocal laser amide] where each bisacrylamide subunit is most likely (model 2000-MP; Bio-Rad Laboratories) attached to a to make an elastically effective crosslink.
fluorescence microscope (Nikon Eclipse TE-300). Con- Figure 2 shows the large dependence of fibroblast focal images were recorded at 120⫻.
morphology on FN-coated gel stiffness. Previous studieshave shown that the density of adhesion protein bound tothe gel surface is independent of gel stiffness [Flanagan et al., 2002], and imaging of the gel surface with fluo- The large range of physiologically relevant stiffness rescently labeled FN showed a constant amount of pro- that can be produced using polyacrylamide gels is shown in tein bound to both the softest and stiffest gels and a Figure 1. The softest gels that can reproducibly be formed uniformity of coverage that was at least as good as that and handled for cell studies are made with 3% acrylamide.
on plastic (data not shown). The shapes of NIH3T3 At low crosslinker concentrations, gels can be produced fibroblasts after one day in culture range from round, with elastic moduli below 10 Pa, the consistency of mucus.
nearly spherical cells with a few irregular protrusions The upper limit to 3% gels appears to be near 600 Pa.
seen on gels of 180 Pa stiffness to large spread cells on Higher moduli require greater concentrations of acrylamide, gels stiffer than 16,000 Pa, which were indistinguishable and as seen in Figure 1, 5.5 and 7.5% gels allow large from those grown on glass or the plastic surfaces sur- variations of elastic moduli as the crosslinker concentration rounding the gel in the culture dish. As visualized by

Yeung et al.
Effect of substrate mechanical properties on fibroblast actin cytoskeleton. a–f: NIH 3T3
fibroblasts expressing EGFP-actin were plated on
polyacrylamide gels with rigidities ranging from
180 Pa (a) to 16,000 Pa (f). Fibroblasts on soft
materials had no stress fibers compared with fibro-
blasts on stiffer materials that do have articulated
stress fibers. Scale bar ⫽ 10 ␮m. g,h: NIH 3T3
fibroblasts on soft gels (180 Pa) are fixed with 4%
paraformaldehyde and their F-actin stained with
rhodamine phalloidin. The isolated fibroblast (g)
appears to have no stress fibers as in a. When the
fibroblasts are able to make cell-cell contact (h),
stress fibers form. Scale bar ⫽ 15 ␮m.
GFP-actin, fibroblasts grown on gels softer than 1,600 Pa A similar dependence of cell shape on stiffness was had no detectable stress fibers or other actin bundles also seen with bovine aorta endothelial cells. These cells, visible by fluorescence microscopy. At stiffnesses above shown after 1 day in culture in Figure 3, also ranged from 3,600 Pa, such actin fibers were a nearly universal feature round to well spread as the stiffness was increased about of these cells.
a few thousand Pa. As with fibroblasts, the round cells on The absence of stress fibers on soft gels was not an soft surfaces were capable of division and were able to indication of toxicity since cell growth rates were only form confluent monolayers indistinguishable from those slightly slower on 180-Pa gels compared to the rates on formed by cells on stiffer surfaces. Figure 3d–f shows the stiffest gels (55,000 Pa) and rates on gels between monolayers developed after 3 days of culture on surfaces 1,600 and 3,600 Pa were equal to or greater than those on of stiffness varying from 180 to 29,000 Pa. Unlike the the stiffest gels. The absence of stress fibers was also obvious difference in shape of single cells, these endo- only evident on single cells on the soft gels. As the cell thelial cells lose their morphologic difference once they density increased over time, cells that made cell-cell make cell-cell contact. The density of cells in the mono- contact frequently became elongated and developed layers also was independent of substrate stiffness, with 927, stress fibers. Figure 2g,h shows two examples of fibro- 898, and 895 adherent cells counted on equal-size areas blasts grown 48 h on a 180-Pa gel and stained with (600 ⫻ 490 ␮m) of gels with elastic moduli of 180, 2,900, rhodamine phalloidin. The single cell (Fig. 2g) shows and 28,600 Pa, respectively.
phalloidin staining consistent with the images of GFP- The dependence of cell shape on stiffness is shown actin under these conditions showing amorphous F-actin quantitatively in Figure 4. Figure 4a shows the average distribution and no evidence of stress fibers. Comparable circumference of 3T3 fibroblasts grown for 1 day on gels of cells (Fig. 2h) on the same gel stiffness that had made varying stiffness that were laminated with either fibronectin intercellular contact show a more flattened morphology (circles) or type 1 collagen (triangles). The circumference of and abundant stress fibers.
these cells changes by more than a factor of 4 as the

Cell Morphology and Substrate Stiffness
Effect of substrate mechanical properties on endothelial cell morphology. Bovine aortic endo- thelial cells (BAECs) were plated on polyacrylamide gels. Projected cell area increases with substrate
stiffness (a– c). As cells reach confluence (d–f) a monolayer forms in all groups and the morphologies
become indistinguishable.
substrate stiffness increases and reached a value almost In contrast to both fibroblasts and endothelial identical to that on glass for FN-laminated gels with stiff- cells, human neutrophils exhibit no dependence of cell ness above 10,000 Pa. On soft gels, 3T3 fibroblasts have shape on substrate stiffness over the stiffness range approximately the same size when bound either to FN or accessible by this system. Figure 4c shows that the collagen, but on gels with 10,000 Pa, they show more mod- circumference of neutrophils is constant on gels with est spreading on collagen than when bound to FN. Similar stiffness from 2 to 1,000 Pa and equal to that seen on results were obtained when plotting adherent area instead of glass. Moreover, the fMLP-stimulated increase in cell circumference (data not shown). Controls on glass could not spreading is as large on the softest gel that could be be done for collagen since in this case FN from the serum- made (2 Pa) as it is on glass.
containing medium would also bind the glass surface and The difference in morphology observed after provide these cells with an additional adhesion protein. The prolonged incubation on gels with different stiffness non-monotonic dependence of circumference (or area) with arises initially from large differences in the rates of gel stiffness is unexplained but apparently not simply an spreading after initial adhesion of cells to the surface.
artifact of measuring mean areas of populations with a large Figure 5 shows time courses of the spreading of two variance, since in repeated studies using different fibroblast representative fibroblasts after initial binding to gels samples and gel preparations, maxima in cell spreading with stiffness of 180 (Fig. 5a) or 55,000 Pa (Fig. 5b).
were seen at intermediate stiffnesses around 3,000 Pa.
The change in adherent area with time for 5 such cells Endothelial cells also show a stiffness-dependent under each condition are shown in Figure 5C and D.
spreading, but the change is not as large as for fibroblasts There is a rather broad range of spreading rates under as quantified either by cell circumference (Fig. 4b) or both conditions, but, on average, cells on soft gels do area (data not shown). As with fibroblasts, spreading on not spread fast or very far compared to cells on stiffer collagen is somewhat less than on FN-coated gels.
gels. On the stiff gels, a few cells fail to spread after Yeung et al.
Measurement of cell cir- cumference for different cell types
on substrates of varying rigidities.
NIH 3T3 fibroblasts (a) and BAECs
(b) were plated on flexible substrates
coated with fibronectin (circles) or
type I collagen (triangles) and im-
aged after 24 h in culture. Both fi-
broblasts and BAECs increased in
circumference as substrate stiffness
increased. Cells plated on fibronec-
tin reached a larger apparent circum-
ference than those plated on colla-
gen. A histogram of circumference
measurements (inset in a) shows that
despite a large variation in cell cir-
cumference within a given cell pop-
ulation, the distributions measured at
different stiffnesses overlap very lit-
tle. Human blood neutrophils (c)
were also plated on varying rigidities
and left untreated (open circles) or
stimulated
fMLP. Neutrophil circumference re-mains essentially stable despite gelstiffness. Error bars denote S.E.M.
from measurements of 10 to 200cells for each data point.
initial attachment but most spread at a rate never (Fig. 6a). The increase in ␣5 integrin expression on observed on the soft gel. The average rate of spreading stiff gels may indicate an increase in adhesivity on as quantified by the initial rate of area versus time stiffer materials. To determine if ␣5 integrin expres- (Fig. 5c, inset) is more than 5 times larger on the sion was sufficient to cause cell spreading, even on stiffer gel.
soft surfaces, the ␣5 subunit was expressed as an Coincident with the rather abrupt spreading of EGFP fusion in NIH 3T3 fibroblasts. The exogenous fibroblasts on FN-coated surfaces stiffer than 2,000 Pa expression of EGFP-␣5 had no obvious effect on cell was a large increase in expression of ␣5 integrin spreading, as cells on soft and stiff gels followed the

Cell Morphology and Substrate Stiffness
Cell spreading dependence on substrate stiffness. The initial cell spreading as measured by cell area of NIH3T3 fibroblasts on soft (a,c) and stiff gels (b,d) shortly after plating. a,b: Representative phase
images of fibroblasts spreading on both gels show that cells on soft gels (a) spread very little compared
with the extent of spreading on stiff gels (b). Quantification of adherent cell area shows that cells on soft
gels are smaller and exhibit a slower spreading rate (c) compared to the average of cells on stiff gels (d).
Inset in c: Rate of spreading as quantified by initial rate of area versus time on soft and stiff gels.
same morphological trend as untransfected cells (Fig.
6b,c). On soft gels, ␣5 was evident along the plasmamembrane, but the cell did not spread as it did on stiff Most cell types in multicellular organisms are at- gels where the adherent area was larger and the ␣5 had tached to soft materials, either other cells or extracellular a punctate distribution.
matrices, but most of what is known about cell structure

Yeung et al.
Role of ␣5-integrin in fibro- blasts on flexible substrates. a: Fibro-
blasts plated on varying rigidities were
lysed after 24 h in culture, and ␣5 ex-
pression was determined by Western
blot analysis as shown in the inset. Fi-
broblasts on stiff materials show a 5-fold
increase in protein expression compared
with cells on softer gels. b,c: Exogenous
expression of ␣5 by transfection with
GFP-␣5 had no effect on cell spreading.
Cells on soft gels (b) showed that ␣5
was present and localized to the cell
membrane, but cells remained small and
rounded. In contrast, cells on stiff gels
expressing exogenous ␣5 (c) were well
spread.
and function in vitro derives from studies of cells plated in different cell types and depends on the nature of the on rigid substrates such as plastic or glass, laminated adhesion receptor by which the cell binds its substrate.
with a thin film of protein. As a result, some prominent There are also important differences between cells grown aspects of cell structure, such as the fan-shaped morphol- on two- and three-dimensional adhesive materials, but ogy of cultured fibroblasts or the prevalence of large even when confined to adhesion on flat surfaces, the actin-containing stress fibers that are commonly studied elastic constant of the surface can determine cell mor- in vitro, are rarely if ever seen in vivo. One well-recog- phology and protein expression over a very wide range.
nized and studied reason to account for such differences Recent studies with aortic smooth muscle cells have is that many cell types in vivo function within a three- shown that substrate stiffness is a more important deter- dimensional matrix, in contrast to the growth of cultured minant for cell shape than is the density of adhesive cells at a solid/liquid interface [Bard and Hay, 1975; ligand to which the cell binds [Engler et al., 2004].
Greenburg and Hay, 1982], but an additional indepen- In addition to relatively well-recognized mechani- dent difference relates to differences in elasticity of bio- cal signals produced at cell-matrix adhesion sites, cells in logical surfaces [Lo et al., 2000; Pelham and Wang 1997; mechanically active environments also form extensive, Wang et al., 2000a]. The studies reported here show that cadherin-mediated intercellular junctions that are impor- cellular response to matrix stiffness may be very different tant in tissue remodeling and differentiation. A relevant Cell Morphology and Substrate Stiffness
finding in the work reported here is that when endothelial able to generate large traction forces that it can generate cells or fibroblasts make cell-cell junctions on soft sub- during phagocytosis [Evans et al., 1993] and that appear strates, they convert to the morphology seen on stiffer to be essential for extension of fibroblast cytoskeletons.
substrates. Specifically, fibroblasts in contact with other Another example that runs counter to the expectation that cells spread and develop stress fibers even on surfaces of cell protrusion requires traction forces exerted on a stiff 100-Pa stiffness, when neighboring single cells maintain material is observed with neurons derived from embry- a radically distinct rounded morphology (Fig. 2h). Like- onic murine spinal cord. These cells extend processes wise, endothelial cells when confluent have indistin- that branch most effectively on soft surfaces with maxi- guishable morphologies on soft and hard substrates while mal branching on gels with stiffness of 50 Pa [Flanagan their structures are easily distinguishable when they are et al., 2002]. The detailed differences in how neutrophils sparse enough to lack cell-cell junctions.
and other cell types employ actin polymerization and The simplest explanation for this change is that cells acto-myosin contractility to move are not known, but have a binary sensor at their membrane junction sites that they must be variable enough to allow robust spreading signals for a relaxed rounded morphology when the surface and motility of neutrophils and neurons on surfaces that is softer than the cell's intrinsic elastic modulus, and signals are far too soft to allow fibroblasts or endothelial cells to for another phenotype with increased contractility and stress spread. There are clearly defined differences in the num- fiber formation when the external material, either another ber, size, and stability of integrin-based adhesions in cell or an extracellular matrix, is stiffer than the cell or as different cell types [Entschladen and Zanker, 2000], and stiff as the cell itself. An alternate but not exclusive expla- it seems probable that large focal adhesions characteristic nation is that the signals for mechanically induced morpho- of fibroblasts on stiff surfaces do not form nor could they logic differences come only from extracellular ligand-acti- function on soft gels [Balaban et al., 2001]. Different cell vated integrins, and that activated cadherins override these types may have adapted responses to different ranges of signals to establish a distinct program. Adherens junctions stiffness to match their function in vivo.
in connective tissue fibroblasts or other cell types may The magnitude of stiffness over which fibroblasts transmit mechanical signals and coordinate multicellular switch from being round and free of stress fibers to the adaptations to physical forces [Ko et al., 2001; Nelson and fan-shaped form with abundant stress fibers seen when Chen, 2003; Philippova et al., 1998]. The involvement of grown on rigid surfaces may help differentiate among dif- cadherin-based adhesions sites in sensing and responding to ferent possible structures responsible for stiffness sensing.
the mechanical properties of the tissue is strongly suggested For fibroblasts on fibronectin-coated surfaces, the transition by the finding that in wound healing, intercellular contacts occurs around a value for the shear modulus of 1,000 to involving interactions between cadherins, actin, and myosin 3,000 Pa. This stiffness is similar to the elastic modulus of are implicated in generating the forces required for wound the fibroblast itself. Measurements by atomic force micros- closure [Adams and Nelson, 1998]. Notably, maintenance copy of Young's modulus (equal to 3 times the shear mod- of the structure of intercellular contacts critically depends ulus for simple incompressible solids) of an NIH 3T3 fi- upon contractile forces generated by the actin cytoskeleton broblasts on a rigid surface are between 3 and 12 kPa [Danjo and Gipson 1998] that act on intercellular contacts in [Rotsch et al., 1999]. The similarity of these elastic con- adjacent cells [Gloushankova et al., 1998], and the magni- stants suggests that fibroblasts may undergo spreading only tude and spatial distribution of these forces depend on the when internally generated forces, which are reported to be viscoelastic properties of the cells as well as the underlying independent of collagen matrix stiffness [Freyman et al., 2002], are exerted on a surface that is stiff enough so that One clear outcome of these studies is that not all the resulting deformation is partly within the cytoskeleton cells appear to use stiffness as a cue for morphology or and not only within the external matrix.
motility, and the well-documented increase in spread The measured shear moduli of the polyacrylamide- area in fibroblasts [Lo et al., 2000] as matrix stiffness or based gels used in these studies is somewhat smaller than density of adhesions sites increases is not universally the elastic moduli reported from measurements of uniaxial observed. Figure 4, for example, shows that, unlike fi- extension or indentation reported in other studies. The broblasts or endothelial cells, neutrophils appear to be forces that cells apply to a flat surface are shear forces, and insensitive to matrix stiffness. In the inactive state, they therefore the shear modulus measured here is the relevant are relatively round on all surfaces tested and can spread parameter quantifying resistance of the material to this and protrude after stimulation by a chemotactic peptide geometry of force application. The Young's modulus, as equally well on a gel with elastic modulus 2 Pa as they do measured by extensional or indentation methods, is related on glass. The ability of neutrophils to spread on such a to the shear modulus by a function that requires knowledge soft matrix is remarkable because it means that the cell of the Poisson's ratio of the material. The somewhat smaller extends its cortical actin-rich periphery without being values of shear modulus measured directly may help resolve Yeung et al.
a question related to the mechanism by which cells produce Engler A, Bacakova L, Newman C, Hategan A, Griffin M, Discher D.
the internal stress that allows them to probe matrix mechan- 2004. Substrate compliance versus ligand density in cell on gel ical properties. Magnitudes of this stress, reported to be as responses. Biophys J 86:617– 628.
Entschladen F, Zanker KS. 2000. Locomotion of tumor cells: a mo- large as 30 kPa in some studies, are difficult to account for lecular comparison to migrating pre- and postmitotic leuko- by a purely acto-myosin dependent process because the cytes. J Cancer Res Clin Oncol 126:671– 681.
concentration of cellular myosin and the known range of Evans E, Leung A, Zhelev D. 1993. Synchrony of cell spreading and forces each motor can generate are not sufficient to produce contraction force as phagocytes engulf large pathogens. J Cell this level of force per surface area, suggesting that other Flanagan LA, Ju YE, Marg B, Osterfield M, Janmey PA. 2002. Neurite force-generating mechanisms such as osmotic stress may be branching on deformable substrates. Neuroreport 13:2411–2415.
involved. However, the relevant elastic constants may be Flory P. 1953. Principles of polymer chemistry. Ithaca: Cornell Uni- smaller than those used to calculate internal stress, and the versity Press. 672 p.
transition from round to spread morphology of fibroblasts Freyman TM, Yannas IV, Yokoo R, Gibson LJ. 2002. Fibroblast shown in Figure 4 suggests that internal stresses of 3,000 Pa contractile force is independent of the stiffness which resists thecontraction. Exp Cell Res 272:153–162.
may be sufficient for mechanosensing.
Gloushankova NA, Krendel MF, Alieva NO, Bonder EM, Feder HH, Vasiliev JM, Gelfand IM. 1998. Dynamics of contacts betweenlamellae of fibroblasts: essential role of the actin cytoskeleton.
Proc Natl Acad Sci USA 95:4362– 4367.
The stiffness of the surface to which cells adhere Greenburg G, Hay ED. 1982. Epithelia suspended in collagen gels can lose polarity and express characteristics of migrating mesen- can have a profound effect on cell structure and protein chymal cells. J Cell Biol 95:333–339.
expression, but these mechanical effects vary with dif- Ko KS, Arora PD, McCulloch CA. 2001. Cadherins mediate intercel- ferent cell types, and depend on the nature of the adhe- lular mechanical signaling in fibroblasts by activation of sion receptors by which the cells bind their substrate.
stretch-sensitive calcium-permeable channels. J Biol Chem Fibroblasts and endothelial cells develop a spread mor- phology and actin stress fibers only when grown on Laukaitis CM, Webb DJ, Donais K, Horwitz AF. 2001. Differential dynamics of alpha 5 integrin, paxillin, and alpha-actinin during surfaces with an elastic modulus greater than 2,000 Pa, formation and disassembly of adhesions in migrating cells.
with a greater effect seen when bound to fibronectin J Cell Biol 153:1427–1440.
compared to collagen. In contrast, neutrophils appear to Lo CM, Wang HB, Dembo M, Wang YL. 2000. Cell movement is be insensitive to stiffness changes over a very wide guided by the rigidity of the substrate. Biophys J 79:144 –152.
range. The stiffness-dependence of fibroblasts and endo- Nelson CM, Chen CS. 2003. VE-cadherin simultaneously stimulates and inhibits cell proliferation by altering cytoskeletal structure thelial cells is no longer evident when cells become and tension. J Cell Sci 116:3571–3581.
confluent, or in the case of fibroblasts even when two Ory DS, Neugeboren BA, Mulligan RC. 1996. A stable human-derived cells make contact suggesting either that mechanosens- packaging cell line for production of high titer retrovirus/ ing uses the cells' internal stiffness as a criterion or else vesicular stomatitis virus G pseudotypes. Proc Natl Acad Sci that signaling from cadherins in cell-cell contacts over- USA 93:11400 –11406.
Pelham RJ, Jr., Wang Y. 1997. Cell locomotion and focal adhesions rides signals from the cell-matrix adhesion complexes.
are regulated by substrate flexibility. Proc Natl Acad Sci USA94:13661–13665.
Philippova MP, Bochkov VN, Stambolsky DV, Tkachuk VA, Resink TJ. 1998. T-cadherin and signal-transducing molecules co- Adams CL, Nelson WJ. 1998. Cytomechanics of cadherin-mediated localize in caveolin-rich membrane domains of vascular cell-cell adhesion. Curr Opin Cell Biol 10:572–577.
smooth muscle cells. FEBS Lett 429:207–210.
Balaban NQ, Schwarz US, Riveline D, Goichberg P, Tzur G, Sabanay I, Rossi FM, Guicherit OM, Spicher A, Kringstein AM, Fatyol K, Mahalu D, Safran S, Bershadsky A, Addadi L, Geiger B. 2001.
Blakely BT, Blau HM. 1998. Tetracycline-regulatable factors Force and focal adhesion assembly: a close relationship studied with distinct dimerization domains allow reversible growth using elastic micropatterned substrates. Nat Cell Biol 3:466 – 472.
inhibition by p16. Nat Genet 20:389 –393.
Bao G, Suresh S. 2003. Cell and molecular mechanics of biological Rotsch C, Jacobson K, Radmacher M. 1999. Dimensional and me- materials. Nat Mater 2:715–725.
chanical dynamics of active and stable edges in motile fibro- Bard JB, Hay ED. 1975. The behavior of fibroblasts from the devel- blasts investigated by using atomic force microscopy. Proc Natl oping avian cornea. Morphology and movement in situ and in Acad Sci USA 96:921–926.
vitro. J Cell Biol 67:400 – 418.
Wakatsuki T, Kolodney MS, Zahalak GI, Elson EL. 2000. Cell mechanics Cunningham CC, Vegners R, Bucki R, Funaki M, Korde N, Hartwig studied by a reconstituted model tissue. Biophys J 79:2353–2368.
JH, Stossel TP, Janmey PA. 2001. Cell permeant polyphosphoi- Wang HB, Dembo M, Wang YL. 2000a. Substrate flexibility regulates nositide-binding peptides that block cell motility and actin growth and apoptosis of normal but not transformed cells. Am J assembly. J Biol Chem 276:43390 – 43399.
Physiol Cell Physiol 279:C1345–C1350.
Danjo Y, Gipson IK. 1998. Actin "purse string" filaments are anchored Wang LZ, Gorlin J, Michaud SE, Janmey PA, Goddeau RP, Kuuse R, by E-cadherin-mediated adherens junctions at the leading edge Uibo R, Adams D, Sawyer ES. 2000b. Purification of salmon of the epithelial wound, providing coordinated cell movement.
clotting factors and their use as tissue sealants. Thromb Res J Cell Sci 111:3323–3332.

Source: http://cytothesis.us/3.0/Oil_Cell-Morphology.pdf