Strain Stiffening of Fibrillar Collagen during ... - ACS Publications

Apr 29, 2016 - ABSTRACT: The multistep process of cell migration requires cells to dynamically couple to extracellular interfaces and generate tractio...
0 downloads 0 Views 5MB Size
Forum Article www.acsami.org

Strain Stiffening of Fibrillar Collagen during Individual and Collective Cell Migration Identified by AFM Nanoindentation Sjoerd van Helvert† and Peter Friedl*,†,‡,§ †

Radboud University Medical Centre, Radboud Institute for Molecular Life Sciences, Department of Cell Biology, Nijmegen, The Netherlands ‡ David H. Koch Center for Applied Research of Genitourinary Cancers, The University of Texas MD Anderson Cancer Center, Houston, Texas 77030, United States § Cancer Genomics Center (CGC.nl), 3584 CG Utrecht, The Netherlands S Supporting Information *

ABSTRACT: The multistep process of cell migration requires cells to dynamically couple to extracellular interfaces and generate traction force or friction for displacement of the cell body. When deformed, biopolymer networks, including fibrillar collagen and fibrin, undergo a nonlinear elasticity change that is termed strain stiffening and is commonly measured by bulk rheology. It remains poorly characterized, however, whether forces generated by moving cells suffice to induce strain stiffening. To detect strain stiffening at the leading edge of normal and tumor cells moving across fibrillar type I collagen, we combined AFM nanoindentation and differential field probing with confocal reflection microscopy. In different cell models, gradient-like fiber realignment, densification, and elevation of Young’s modulus ahead of the leading edge were observed, with peak increases of up to 1.15 kPa near the leading edge. Moving fibroblasts generated a larger anterograde strain field with a higher amplitude and up to 6-fold increased cumulative strain stiffening (52 kPa) compared with mesenchymal HT1080 fibrosarcoma cells (8.8 kPa) and epithelial SCC38 cancer cells (9.8 kPa). Collectively moving SCC38 cells produced 4-fold increased cumulative strain stiffening (38 kPa) compared with individually moving SCC38 cells in a β1 integrin- and actomyosin-dependent manner. This indicates that the extent of strain stiffening by the leading edge of moving cells scales with cell type, multicellular cooperativity, integrin availability, and contractility. By straining, migrating cells realign and densify fibrillar extracellular matrix and thus adopt an autonomous strategy to move on a “traveling wave” of stiffened substrate, which reaches levels sufficient for mechanosensory activation and self-steering of migration. KEYWORDS: strain stiffening, collagen, cell migration, AFM nanoindentation, Young’s modulus



INTRODUCTION Individual and collective cell migration are fundamental physicochemical processes that are critically involved in morphogenesis, wound healing, and immune function as well as cancer metastasis. As the first step of migration, cells develop a leading edge that mechanically couples to extracellular tissue interfaces. Initial adhesion is followed by cell contraction, gradual sliding of the cell rear, and translocation of the cell body.1 The mechanical traction force required for cell displacement is generated by actomyosin-mediated contraction and transmitted through integrin-based adhesion toward the extracellular matrix (ECM) substrate.1,2 Besides providing force © XXXX American Chemical Society

for movement, adhesive mechanocoupling induces intracellular signaling in response to local ECM mechanics.3,4 Mechanosensing influences cell function, including cell morphology and cytoskeletal structure,5 differentiation,6,7 and migration.8,9 Consequently, defects in mechanotransduction are implicated Special Issue: Interfaces for Mechanobiology and Mechanochemistry: From 2-D to 3-D Platforms Received: February 10, 2016 Accepted: April 18, 2016

A

DOI: 10.1021/acsami.6b01755 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

ACS Applied Materials & Interfaces



in the development of various diseases10 and contribute to malignant cell transformation and cancer progression.11 The material stiffness is defined by the resistance to deformation, which for elastic materials is commonly measured as the Young’s modulus (E, in Pa).12 The elastic moduli of tissues exhibit great variability, ranging from 100 Pa in brain to over 1 MPa in bone,13 and largely depend upon the quantity and organization of fibrillar collagens and integrated mineral.14 In addition to baseline stiffness, most biopolymer networks possess strain stiffening properties, defined by an increased elastic modulus at small to moderate strains.15 Strain stiffening is reversible, confers compliance to small forces and resistance to large and potentially harmful deformations, and thereby contributes to shape dynamics and functioning of tissues.16 Strain stiffening and fiber alignment enable cells to sense mechanical signals over a distance of several cell lengths and could permit long-range cell−cell communication,17−19 whereas these effects are lacking in linear synthetic materials.20 Macro-scale rheology measurements reveal actomyosin-mediated contraction of fibroblasts as sufficient to drive fibrin gels into a strain-stiffened state, coinciding with the onset of cell elongation.21 Likewise, fibroblasts and mesenchymal stem cells can locally introduce stiffness gradients in fibrin gels by actomyosin contractility,18 indicating global and local strain stiffening as a consequence of cell contraction in fibrin gels. Whereas high forces generated by cell spreading tune the stiffness of nonlinear ECM networks, it is unclear whether forces generated by moving cells equally result in physiologically relevant strain stiffening. The stress−strain behavior of biopolymer networks is routinely measured by shear rheology.21,22 These measurements, however, are performed on a macroscopic scale and thus lack spatial sensitivity toward local heterogeneity of cellgenerated strain stiffening. Traction force microscopy (TFM) and micropillar devices provide a solution for local strain and force mapping (as reviewed in refs 23 and 24). These techniques have demonstrated that anisotropy in strain energy directs cancer cell invasion25 and that each individual cell generates force toward the substrate in collective epithelial migration,26 as indicated by balanced waves of mechanical stress between cadherin-based intercellular junctions.27,28 The active tensile modulus of an epithelial sheet scales linearly with the number of its constituent cells,29 which cooperatively strain their local environment, impacting tissue tension and movement. Because of the complex stress−strain behavior of biopolymer networks, these approaches often rely upon synthetic materials, which, however, lack the complex mechanical properties of physiological biopolymers networks, or use simplifying assumptions by recording deformation as an indirect readout of local forces.23 Thus, a direct readout of the magnitude of strain stiffening on a cellular level and in the context of cell migration remains poorly defined. To directly address local elasticity changes generated by moving cells on fibrillar type I collagen as an ECM environment, we here apply spatially resolved atomic force microscopy (AFM) nanoindentation in live-cell culture. We provide quantitative heat maps and reveal strain stiffening at the leading edge of normal and tumor cells in single and collective cell migration. The data show that strain stiffening is linked to actomyosin-coupled adhesion, depends upon the cell type and migration mode, and strongly impacts the topography of the force field encountered by moving cells.

Forum Article

EXPERIMENTAL PROCEDURES

Cell Lines and Matrix Culture. The following cell lines were used: human mammary fibroblasts (HMF) (ScienCell Research Laboratories), lentivirally transduced with LifeAct-GFP; human HT1080 wild-type fibrosarcoma (ACC315, DSMZ Braunschweig); and human SCC38 head and neck squamous cell carcinoma (SCC) derived from glottic larynx (UT-SCC38, kindly provided by R. Grénman, University of Turku, Finland). Cells were cultured at 37 °C with 10% CO2 humidified atmosphere in Dulbecco’s modified Eagle’s medium (DMEM) (Invitrogen) supplemented with 10% fetal calf serum (FCS) (Sigma-Aldrich), penicillin (100 units/mL) and streptomycin (100 μg/mL), and L-glutamine (2 mM) and sodium pyruvate (1 mM) (both Invitrogen). Multicellular spheroids (500 cells) were generated from subconfluent culture flasks after detachment with EDTA (2 mM) and trypsin (0.075%) (both Invitrogen) by overnight hanging-drop incubation in 10% methylcellulose (Sigma).30 Fibrillar collagen gels were prepared from acid-solubilized rat-tail collagen I (Corning; 5 mg/mL final concentration), as described.31 For AFM and confocal microscopy, gels of 0.1−0.3 mm thickness were polymerized onto glass-bottom dishes (Willco Wells) and in silicon spacers (Electron Microscopy Sciences) in 24-well plates, respectively. Low-density single-cell suspensions or multicellular spheroids were plated on top of the collagen lattice, allowed to attach (30−60 min), overlaid with medium, and incubated overnight before AFM probing or fixation. Samples prepared for AFM measurements were maintained in bicarbonate-free DMEM (Sigma) supplemented with HEPES (20 mM, Invitrogen) becaue of the absence of CO2 pressure on the AFM system. As a consequence, these samples were also incubated in the absence of CO2 while the other culture conditions were retained. Cell probing in the presence of anti-β1 integrin monoclonal antibody (mAb) 4B4 (Beckman Coulter) using a dose that inhibited collagen contraction by 50% (final concentration 1 μg/mL)32 or blebbistatin (Sigma; final concentration 20 μM) was performed after a preincubation period of 2 or 3 h, respectively. AFM Nanoindentation at Physiologic State. AFM field probing was performed with a Catalyst BioScope (Bruker) coupled to a confocal microscope (TCS SP5II, Leica), using the “point and shoot” procedure (Nanoscope software, Bruker). A fluorescent 10 μm polystyrene bead (Invitrogen) was glued to silicon nitride cantilevers with nominal spring constants of 0.06 N/m (NP-S type D, Bruker).33 The system was calibrated in cell-free medium at 37 °C prior to each experiment by measuring the deflection sensitivity on a glass surface, which allowed the cantilever spring constant to be determined using the thermal noise method.34 Before placement of the sample, the xy movement of the sample stage was calibrated using the NanoScope software. After the bead position was registered, three or four force− distance curves were sequentially acquired from each point to form a two-dimensional (2D) array. The approach and retraction velocities were kept constant at 10 μm/s, ramping the cantilever by 4 μm with a 2 nN threshold in a closed z loop to achieve a contact area of approximately 50 μm2. To retrospectively assess cell displacement in order to confirm probing of the leading edge, bright-field images were taken before and after data acquisition (Figure S1). The maximum duration of live-cell probing was restricted to 2 h to minimize any negative impact of medium evaporation. Automated Fitting of Force−Indentation Curves and Data Processing. AFM force−distance curves were both transformed to force−indentation curves and fitted using the PUNIAS software (http://punias.free.fr). Because contact point determination is inherently uncertain when probing soft substrates,35 curves were fitted from the end downward (Figure S2) to eliminate errors imposed by illdefined contact points using the contact-point-independent linear Hertz model:36

(Fbead)2/3 =

⎛4 E ⎞2/3 ⎜ ⎟ R δ ⎝ 3 1 − ν2 ⎠

where Fbead is the force, R is the bead radius, E is the Young’s modulus, δ is the indentation depth, and ν is the Poisson ratio, which was set to 0.5. The Hertz model assumes the following: (1) infinite sample B

DOI: 10.1021/acsami.6b01755 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Forum Article

ACS Applied Materials & Interfaces

Figure 1. AFM−confocal setup for spatial mapping of Young’s moduli during live-cell 3D matrix culture. (A) Experimental setup for AFM spectroscopy of the field (red squares) directly adjacent to the leading edge of moving cells seeded on a thick fibrillar collagen lattice. Custom functionalization of the cantilever with 10 μm diameter polystyrene beads, resulting in an array of probed collagen surface points with an area of up to 22.5 × 103 μm2. The bead position (green sphere) was coregistered using its fluorescent emission. Force−distance curves were obtained, and the Young’s modulus was calculated per 10 μm × 10 μm probing field. (B) Bright-field image of polarized HMF cell expressing LifeAct-GFP. Red squares show the locations of data acquisition. (C) Frequency distribution of background Young’s moduli in cell-free regions lacking strain stiffening. Data sets were normally distributed (D’Agostino and Pearson omnibus normality test) and are represented as stacked bars after binning from three independent experiments. (D) Heat map representing the Young’s modulus of each probing position, not corrected for background, obtained by fitting force−indentation curves (for details, see Experimental Procedures). The LifeAct-GFP signal (maximum intensity projection) was obtained after AFM scanning. (E) Differential distribution of Young’s moduli in the anterograde direction ahead of the leading edge. Values from three independent cells were grouped for 20 μm steps relative to the distance from the leading edge (0 μm). Data represent the medians and 25/75 (box) and 5/95 (whiskers) percentiles. Scale bars: 10 μm. modulus was calculated as the average of triplicate or quadruplicate measurements per point. Values exceeding 2 kPa were assumed to unintentionally include the edge of the cell and were omitted from the data set. To correct for the sample-to-sample variation of background elasticity of fibrillar collagen, the mean background Young’s modulus was established from cell-free areas without signs of cell-induced stiffening. All of the quantifications except for heat map images represent background-corrected data calculated by subtracting the mean background from all of the Young’s moduli for that sample. To determine the differential distribution of Young’s moduli extending in the anterior direction from the leading edge, data points representing the direction of force generation were included, as determined from the cell orientation and anterograde force field above background. The cumulative cell-induced strain stiffening was defined as the sum of all background-corrected Young’s moduli per heat map multiplied by a factor of 1.91−2.54 in a sample-specific manner to extrapolate the

thickness, which was approximated by the use of small indentations of 1.67 ± 0.25 μm on 100−300 μm thick gels, guaranteeing that the underlying glass dish did not interfere; (2) disregard of the fact that the Poisson ratio, which reflects the behavior of the material under compression and is generally set to 0.5 for soft biological samples,37 may be asymmetric and nonlinear for collagen;38 (3) linear elasticity, which underrepresents the complex stress−strain relations of biological materials. To prevent these assumptions from influencing the outcome, fitted curves that did not follow Hertzian behavior were excluded by applying a stringent threshold for data inclusion. Automated curve fitting was applied using a fitting range defined by manual analysis of a batch of force−distance curves. A maximum fit range was selected from the end of the curve downward until R2 dropped below 0.995, showing that 816 ± 269 nm was the maximum range to maintain R2 ≥ 0.995. For outliers, the linear fit range was manually adjusted until a coefficient of 0.995 was reached. Curves with R2 < 0.995 despite manual adjustment were excluded. The Young’s C

DOI: 10.1021/acsami.6b01755 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Forum Article

ACS Applied Materials & Interfaces

Figure 2. Cell-line-specific anisotropy of collagen fibril morphology and stiffness at the leading edges. (A) Collagen fiber alignment and enrichment at the leading edge of fixed single cells, as detected by confocal reflection microscopy: left panels, overview; right panels, close-up in false color (after masking and cropping of the reflection signal from regions with cell overlap). (B) Differential Young’s moduli of the collagen surface adjacent to the leading edge of moving cells. Heat maps were superimposed on the corresponding positions of bright-field images. (C) Distribution of Young’s moduli in the anterograde direction ahead of the leading edge for three cell lines. Values from 4 to 13 independent cells per cell line were grouped as in Figure 1E. Data represent the medians and 25/75 (box) and 10/90 (whiskers) percentiles. (D) Area of strain-stiffened collagen at the leading edge of single cells, calculated by multiplying the pixel area by the number of pixels with amplitudes 2 standard deviations above the background. (E) Highest-amplitude Young’s modulus per individual heat map after background correction. (F) Cumulative strain stiffening generated by the leading edge for three cell lines. The area and amplitude of strain stiffening are expressed as a single data point per heat map (for details, see Experimental Procedures). Red lines in D−F indicate medians. **, P < 0.01; *, P < 0.05; n.s., nonsignificant (Mann−Whitney test). Scale bars: 10 μm. reflection analysis of fibrillar collagen, except for densitometry of cellfree collagen samples, where a wavelength of 488 nm was used. Intensities were measured by average-intensity line plots (100 pixels width). Images were processed from 3D stacks using FIJI.39 Curve fitting was performed using the lowess function (GraphPad Prism 5.01), unless stated otherwise.

bead−substrate contact area to the pixel area, thus correcting for slight variations in probing density. For comparison of two groups without Gaussian distributions, the nonparametric Mann−Whitney test was used. All of the analyses were performed in GraphPad Prism 5.01. Confocal Fluorescence and Reflection Microscopy. Samples for confocal microscopy were fixed (4% paraformaldehyde in phosphate buffer) for 15 min at room temperature and subsequently washed in phosphate-buffered saline. Samples were then stained with 4′,6-diamidino-2-phenylindole (DAPI) (Roche) and Alexa Fluor 488conjugated phalloidin (Invitrogen). High-magnification confocal imaging was performed in an Olympus FV1000 scanner (60×/1.35 NA oil objective for cell-containing gels and 40×/0.80 NA water objective for cell-free gels). Reflection signal intensities were quantified from maximum projections of confocal stacks for an average scan depth of 5 μm from the collagen surface for cell-containing gels and exact 3 μm for cell-free gels. A wavelength of 559 nm was used for



RESULTS Spatial Mapping of Young’s Moduli of Fibrillar Collagen Surfaces. A combined AFM−confocal setup was used to collect spatial arrays of force−distance curves by probing the surface of thick fibrillar collagen I lattices under conditions compatible with live-cell culture (Figure 1A). Cantilevers with nominal spring constants of 0.06 N/m were functionalized with a 10 μm fluorescent polystyrene bead33 to D

DOI: 10.1021/acsami.6b01755 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Forum Article

ACS Applied Materials & Interfaces

Figure 3. Strain stiffening during collective migration of SCC38 cells with dependence on β1 integrins and myosin II. (A) Collagen fiber alignment and enrichment at the leading edge of a fixed sample, as detected by confocal reflection microscopy. Image processing and data analysis were performed as described in Figure 2. Asterisks indicate nuclei. (B) Increased Young’s moduli at the leading edge of collectively moving SCC38 cells: (left) bright-field overview with probing region (white frame) and (right) heat map of the Young’s modulus superimposed on a bright-field image of the corresponding region. (C) Distribution of Young’s moduli in the anterograde direction ahead of the leading edge of collectively moving SCC38 cells before and after treatment with mAb 4B4 (1 μg/mL). Values from four to seven independent spheroids were grouped as in Figure 1E. Data represent the medians and 25/75 (box) and 10/90 (whiskers) percentiles. (D−F) Quantifications of the (D) extent, (E) amplitude, and (F) cumulative strain stiffening of collectively moving SCC38 cells before and after 4B4 treatment. (G) Cumulative strain stiffening after myosin II inhibition by blebbistatin (20 μM) or DMSO control. Black dotted lines indicate reference values obtained from individually moving SCC38 cells (medians from Figure 2C). Quantifications were performed as described in Figure 2. Scale bars: (A) 10 μm; (B) 50 μm.

force is generated at the leading edge of moving cells40,41 (Figure 1D,E). Next, we compared whether and to what extent migrating mesenchymal and epithelial single cells induce strain stiffening. Fibroblasts generated larger fields of fiber alignment compared with mesenchymal HT1080 sarcoma or epithelial SCC38 cancer cells, as detected by confocal reflection microscopy in fixed samples (Figure 2A). Systematic spatial mapping of Young’s moduli of the same region in live samples revealed that fiber alignment is accompanied by stiffening of the substrate (Figure 2B). The cell-specific maximum Young’s modulus was adjacent to the leading edge and gradually decreased in the anterograde direction as a consistent feature across cell types (Figure 2C). This stiffness distribution is consistent with the force density of individual cancer cells moving in 3D fibrillar collagen, which generate the peak force within 20 μm of the cell edge.38 In direct comparison, moving fibroblasts generated significantly larger areas of increased Young’s modulus than HT1080 and SCC38 (Figure 2D). Likewise, the peak Young’s moduli differed among cell types, with fibroblasts inducing 2-fold higher values compared with HT1080 and 1.3-fold compared with SCC38 cells (Figure 2E). To integrate both the field size and the amplitude of the Young’s modulus in a single parameter, the cumulative cellinduced strain stiffening was calculated as the sum of background-corrected Young’s moduli per individual heat map (Figure 2F) as a measure of the total change in Young’s modulus and a global parameter for comparing stiffness modulation among cell types and experimental conditions.

probe the Young’s modulus of the collagen surface on a cellular level. The location and dimensions of the probing array anterior of the leading edge of polarized cells were defined by brightfield microscopy. The array consisted of up to 256 measurement points, each representing an area of 100 μm2 (Figure 1B). Young’s moduli from regions lacking cell-induced stiffening, which were routinely probed from each sample, followed normal distributions with slight sample-to-sample variability (Figure 1C). To account for such sample-specific variability, background corrections were performed on all of the data in a sample-specific manner. Human mammary fibroblasts (HMF) stably expressing LifeAct-GFP were used as a reference model for high contractility and long-range collagen alignment, as detected by confocal reflection microscopy (compare with Figure 2A). Spatial mapping of Young’s moduli revealed that the region of fiber alignment ahead of the cell overlapped with the local anisotropy of elasticity, visible as a gradient-like field of increased Young’s modulus reaching peak values of 1.2 kPa near the cell border (Figure 1D). Increased Young’s moduli consistently extended in a radial manner and gradually decreased in the forward direction of the length axis of the cell (Figure 1E). Thus, the 2D anisotropy maps of collagen elasticity revealed a cell-generated stiffness gradient in the anterograde direction. Differential Force Generation by Individually Moving Cells. Measurements of cellular forces on synthetic materials as well as on 3D fibrillar collagen cultures suggest that the highest E

DOI: 10.1021/acsami.6b01755 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Forum Article

ACS Applied Materials & Interfaces Moving fibroblasts generated strongly increased cumulative strain stiffening (52 kPa) compared with HT1080 (8.8 kPa) and SCC38 (9.8 kPa) (Figure 2F). Comparison of fibroblasts with HT1080 and SCC38 cells showed that changes in field size and cumulative strain stiffening were not proportional to the change in peak Young’s modulus, suggesting that strain stiffening reaches a maximum near the cell attachment site with variable propagation in the forward direction. Thus, cell types differ in their ability to stiffen collagen during migration, with up to 6-fold higher cumulative strain stiffening induced by fibroblasts compared with moving tumor cells. The Extent of Strain Stiffening Scales with Multicellular Cooperativity. We and others have demonstrated extensive collagen alignment at the front of collectively migrating cell groups,42,43 indicative of the critical involvement of high physical forces. When plated as multicellular spheroids onto a fibrillar 3D collagen interface, SCC38 cells often evaded first as finger-like collective strands guided by one or few leader cells and tightly connected follower cells with a typical strand width of 2−4 cells (Figure 3A) followed by occasional singlecell detachment (movie S1 and Figure S3). Thus, SCC38 cells were used as a model for direct comparison of the magnitudes of strain stiffening for individual and collective cell migration. Collagen fiber alignment and enrichment were prominent ahead of and lateral to the leader cell(s) and more pronounced in collectively moving compared with individually moving SCC38 cells (Figure 3A; compare with Figure 2A). The corresponding AFM force curves again revealed a region of elevated Young’s modulus extending from the leading edge (Figure 3B,C; the single-cell reference is denoted as a dotted line) with up to 4-fold increased cumulative stiffness, peak values, and force field for collective compared with single-cell migration (Figure 3D−F; compare with Figure 2D−F). This indicates that multicellular groups mechanically cooperate and generate higher strain stiffening than individually moving cells. Cell-Induced Strain Stiffening of Fibrillar Collagen Networks Is Dependent on Traction through β1 Integrins and Myosin II. Integrins represent the dominant receptor system physically linking extracellular interfaces to the cytoskeleton and have key functions in supporting multicellular migration.42,44 Adhesion to collagen fibrils depends upon β1 integrin function, and therefore, we added adhesion-perturbing anti-β1 integrin mAb 4B4 in moderate concentration (1 μg/ mL) and tested whether the strain stiffening was diminished. In single-cell culture, this dose of mAb 4B4 resulted in complete loss of cell spreading and the force field, thus rendering meaningful probing obsolete (data not shown). In cell groups, mAb 4B4 rapidly affected cell migration, including diminished migration speed, retraction of multicellular strands, and rounding of cells (movie S1 and Figure S3), as described.42 Cell clusters treated with mAb 4B4 revealed diminished collagen stiffening when probed by AFM nanoindentation (Figure 3C), with reductions in the stiffness field (2.3-fold), amplitude (2.3-fold), and cumulative strain stiffening (2.6-fold) compared with untreated cell groups (Figure 3D−F). To transmit force, integrins, via focal adhesions, couple to actomyosin motors, which mediate cell contraction to mechanically pull on adhesion sites and propagate cells and cell groups forward.45 Migrating cell clusters treated with blebbistatin, an inhibitor of myosin II-mediated contractility but not adhesion to the collagen substrate, developed significantly reduced cumulative strain stiffening (Figure 3G) similar in magnitude to the diminished stiffness after mAb 4B4 treatment.

Thus, local force generation and strain stiffening by moving tumor cells depend upon the β1 integrin−collagen interaction and myosin II-mediated contractility.2,42,44 Multiple Parameters Contribute to Cell-Induced Collagen Stiffening. As a standardized reference for the scaling behavior of both Young’s modulus and reflection intensity in response to varying collagen density, we probed the surfaces of cell-free collagen lattices at different concentrations by AFM nanoindentation and confocal reflection microscopy, respectively (Figure 4A). Whereas the intensity of the reflection signal (I) increased linearly with collagen concentration (I ∼ C0.92; Figure 4A), as reported previously,46 the Young’s modulus (E) depended on the collagen concentration by a power law function (E ∼ C2.62; Figure 4A), consistent with theoretical predictions47 and the stiffness behavior of fibrillar collagen measured by shear rheology.48,49 As a quantitative estimate of altered local collagen density near the leading edge of moving fibroblasts, the reflection signal as a function of the distance from the leading edge revealed a gradient-like decrease in the anterograde direction (Figure 4B,C), indicating peak density of collagen fibrils directly adjacent to the leading edge. We finally tested whether the local change in Young’s modulus ahead of the leading edge is sufficiently explained by this local increase in collagen fiber density. Using the dependence of the Young’s modulus and reflection intensity on the collagen concentration found in strain-free gels (Figure 4A), we calculated the predicted reflection intensity corresponding to the local Young’s modulus measured at fibroblast leading edges (Figure 4C, dashed line). The predicted values follow a gradient-like decrease away from the leading edge, reaching the baseline value with increasing distance, but in magnitude range below the slope of the measured reflection intensities (Figure 4C, shaded area). This discrepancy between the measured and predicted reflection intensities indicates that besides the local collagen density, additional parameters contribute to the locally increased reflection signal, such as strain-induced fibril alignment perpendicular to the light path,50 consistent with the radial alignment of collagen fibrils toward the leading edge (Figures 2A and 4B).



CONCLUSION AND DISCUSSION Quantitative stiffness maps of thick fibrillar collagen networks obtained by AFM nanoindentation reveal the amplitude and field of strain stiffening generated by the leading edge of individually or collectively moving cells. The extent of strain stiffening depends upon the cell type, multicellular cooperativity, actomyosin contractility, and integrin availability and thus represents a sensitive determinant of the mode and physics of cell migration. Differential AFM nanoindentation adjacent to moving cells was sufficiently fast and sensitive to reach elasticity probing of coherent fields with high spatial resolution (50 μm2 for 10 μm bead diameter) and to quantify the cell-induced local stiffness change of fibrillar collagen as a physiologically relevant nonlinear substrate. Taking into account the spatial variations of background stiffness of reconstituted lattices, internal background corrections performed for each individual scan resulted in a normalized readout of cell-generated strain stiffening. To further improve the spatial resolution and identify subregions of peak strain, a smaller bead can be used, whereas a larger probe diameter supports faster acquisition (with decreased resolution) for scanning of larger fields. AFM nanoindentation complements bulk rheology measurements by F

DOI: 10.1021/acsami.6b01755 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Forum Article

ACS Applied Materials & Interfaces

measurements conserving the gel volume and hence the fiber density during probing,48,51−53 this cell-based AFM probing defines local gradients of substrate tension and further responds to locally varying fiber density imposed by cell-induced matrix realignment. Networks composed of semiflexible filamentous proteins, including fibrillar collagen and fibrin, undergo stiffening when pulled, and collagen networks stiffen even at low to intermediate strain.15,51 When pulled by cells, collagen fibrils align and straighten in the direction of force.38,41 Consequently, cell-induced strain stiffening occurs within a local strain field extending up to 100 μm ahead of the leading edge, as determined from elevated Young’s moduli and increased fiber density, with peak activity near the leading edge.38 The locally increased fiber density and tension thus cross-talk in the same strain field as interdependent parameters, yet they likely represent distinct consequences of cell-induced strain. Fibrillar collagen networks undergo densification and alignment in response to external deformation,48 and cell-scale simulations predict local fiber compaction and alignment along the direction of tension.19,54 Near the leading edge, the measured local reflection signal intensity ranged clearly above the values predicted from the Young’s moduli using standard curves from cell-free gels. This difference is best explained by collagen fiber deflection and realignment in the horizontal orientation, which maximizes the reflection signal.50 This argues in favor of tension across the fibrillar network as a consequence of cellinduced strain, particularly in “high-force” regions where the local Young’s modulus and reflection intensity steeply diverge (