All reagents were purchased from Sigma (Sigma, St. Louis, MO) unless indicated otherwise.
Preparation of collagen surfaces
The preparation of millimeter thick hydrogels of native fibrillar collagen and thin films of collagen on gold coated coverslips has been described previously [18, 27]. Briefly, purified type I collagen was purchased as a solution of acid-stabilized monomer (Vitrogen; Cohesion Technologies, Inc., Palo Alto, CA). The collagen solution (0.8 mL, ≈3 mg/mL, 4°C) was neutralized with 0.1 mL of 10× Dulbecco's phosphate buffered saline (DPBS) (4°C) and 0.1 mL of NaOH (0.1 M, 4°C), and was kept at 4°C to minimize polymerization. To generate the fibrillar gels gaskets (≈16 mm inside diameter, 22 mm outside diameter) were cut from paper labels (Fasson label material, Avery Dennison, Brea, CA) and adhered onto cleaned coverslips before sterilization with 70% ethanol. Neutralized native collagen solution (100 μL) was applied to the dried coverslip making sure that solution fully contacted the paper gaskets. The samples were placed at 37°C overnight to initiate fibrillogenesis. The native collagen hydrogels were carefully rinsed and stored in DPBS at 4°C until they were used.
To prepare thin films for collagen and fibronectin, coverslips were coated with a 5 nm layer of chromium and a 15 nm – 20 nm layer of gold by magnetron sputtering. The semi-transparent gold-coated coverslips were immersed in 0.5 mM 1-hexadecanethiol (Aldrich, Milwaukee, WI) in ethanol for at least 8 hours before being rinsed with ethanol and dried with filtered N2. Thin films of fibrillar collagen were generated by adding neutralized solutions of native collagen solution (4°C, ≈0.4 mg/mL) to the alkanethiol-treated gold-coated coverslips and incubating overnight at 37°C. After incubation, the samples were slowly lifted out of the gelled collagen solutions, and rinsed with a stream of DPBS and then water from sterile Teflon squirt bottles (Nalgene Nunc, Rochester, NY). Once all loosely adhered gel was removed, the samples were dried under a stream of filtered N2 and immediately placed back into a DPBS solution. Thin films were also produced from a lower solution concentration of collagen by immersing alkanethiol-treated gold-coated glass coverslips into a solution of ≈0.025 mg/mL collagen in DPBS. The samples were incubated at 37°C for at least 12 hours, rinsed in DPBS and water, dried under a stream of filtered N2 and stored in DPBS at 4°C before they were used with cells. Ellipsometry studies of the thin films of collagen show that the preparations generated from the 0.4 mg/mL native collagen solution have an average thickness of 40 nm while those prepared from the lower solution of collagen (0.025 mg/mL) are 6 nm on average with a standard deviation of film thickness of less than 10% for both preparations. The homogeneity of surface coverage, evaluated by comparing the thickness of several areas on each sample by ellipsometry indicated that the standard deviation across the samples was less than 4% . To produce thin films of fibronectin, alkanethiol-treated gold-coated glass coverslips were incubated in the presence of 25 μg/mL solution of bovine fibronectin for at least 5 hours at 4°C, rinsed in DPBS and water, dried under a stream of filtered N2 and stored in DPBS at 4°C. Ellipsometry indicates that the fibronectin layer has a thickness of 4.5 ± 0.5 nm. Prior to seeding cells on the ECM substrates, the thin films and thicker hydrogels were conditioned for a minimum of 30 minutes with cell culture medium containing 5% (v/v) fetal bovine serum (FBS; Gibco Invitrogen, Carlsbad, CA).
NIH3T3 fibroblasts (ATCC, Manassas, VA), were maintained in Dulbecco's Modified Eagles Medium (DMEM; Mediatech, Herndon, VA) supplemented with nonessential amino acids, glutamine, penicillin (100 units/mL), streptomycin (100 μg/mL), 10% by volume FBS and 25 mM HEPES, and maintained in a humidified 5% CO2 balanced-air atmosphere at 37°C. Sub-confluent cultures were switched to supplemented DMEM containing 5% (v/v) FBS 24 hours prior to an experiment. Cells were removed from tissue culture polystyrene flasks by trypsinization, washed with DMEM/5% FBS, centrifuged for 5 minutes at 200 g and plated in DMEM/5% FBS onto the ECM preparations at a density of 2000 cells/cm2. Care was taken to ensure the seeding density was homogeneous over the surface of the substrates. This level of serum of 5% was chosen due to the results of preliminary experiments to examine the relative strength of the serum response element of the TN promoter compared to the biomechanical force and matrix response elements. We chose 5% serum to assure that cell response to the matrix was not overshadowed by the response to serum.
GFP reporter construct
The TN-1/pEGFP construct containing a 4.1 kilobase fragment of the TN-C promoter was generously provided by Dr. Peter L. Jones (University of Pennsylvania) . The sequences encoding enhanced GFP were removed from this construct using restriction enzymes Sal I and Not I followed by gel purification. A fragment of DNA encoding a degenerate variant of enhanced GFP, with a half-life of 2 hours as reported by the manufacturer, was removed from the vector pd2EGFP-1 (Clontech Laboratories, Palo Alto, CA) using similar restriction enzymes, was gel purified and the resulting 860 base pair fragment was ligated into to the TN-C promoter generating TN-1/pd2EGFP-1. NIH3T3 cells were transfected with the degenerate enhanced GFP-TN-C promoter based reporter construct using Lipofectamine reagent (Invitrogen, Carlsbad, CA) as per manufacturer's recommendation. Briefly, 1.5 × 106 cells were plated into a 6 well plate 16 hours prior to the transfection in medium lacking antibiotics. A mixture containing 1.5 μg DNA and 8 μL Lipofectamine was added dropwise to cells. Following a 6 hour incubation, fresh medium containing 10% FBS was added to the cells. Cells were maintained in selection medium containing G418/Geneticin (Gibco Invitrogen, Carlsbad, CA) (400 ug/mL) for fourteen days. Two rounds of limited dilution cloning were used to assure that the resulting cell line was derived from a single cell. This clonal population was used for all reporter based experiments. Passage number was designated from the time of isolation of the final subclone. The clone used for this study has been cryopreserved at low passage numbers to allow us to use relatively low passage number cells for replicate experiments at different times. In the course of this study, we have also begun to examine how passage number and cryopreservation quantitatively influence cell response. For the data shown in Figures 2, 4, 5 and 6, cells were used at passage number 12. For the data shown in Additional File 1, the cells were used at passage number 29.
Cell fixation and staining for automated data analysis
Substrates were removed from the incubator after 24 hours, rinsed with Hanks Balanced Salt Solution (HBSS; ICN Biomedicals, Costa Mesa, CA) supplemented with 10 mM HEPES and fixed for 24 hours at room temperature in 100 mM PIPES, 1 mM EGTA, 4% PEG  containing 100 ug/mL 3-malemido-benzoic acid-NHS ester (MBS, Sigma) as the cross-linker . Cells were permeabilized in 0.05% (v/v) Triton X-100 in DPBS, rinsed in DPBS, and incubated with DPBS containing Texas Red-C2-Maleimide (1 ng/mL) as a general stain and 0.05% (v/v) 4',6-diamidino-2-phenylindole (DAPI, Sigma) as a nuclear counter stain. After 2 hours at room temperature, 1% Bovine Serum Albumin (BSA) was added to quench the conjugation reaction. The substrates were then rinsed with DPBS and were mounted onto thin glass slides with 9:1 glycerol:Tris, pH 8.0. In separate control experiments, we have determined that this fixation procedure results in the retention of >90% of the original GFP of the cells (Elliott et al., in preparation). The fixed and stained cells were examined by phase contrast and fluorescence microscopy using a 10× objective on an inverted microscope (Zeiss Axiovert S100TV, Thornwood, NJ) outfitted with a computer controlled stage (LEP, Hawthorne, NY), an excitation filter wheel (LEP, Hawthorne, NY), and a CCD camera (CoolSnap fx, Roper Scientific Photometrics, Tucson, AZ). Hardware operation, and image digitization and analysis were under software control (ISee Imaging, Cary, NC) [18, 27]. A modular software routine controlled automated movement of the stage, auto-focusing, and collection of data from 50 to100 independent fields (of 870 μm × 690 μm in size) of cells per sample. Auto-focusing was performed on each field while examining the Texas Red fluorescence. The computer-controlled routine involves sequential steps of moving the objective lens relative to the sample to find the objective position that results in maximum variance in intensities within the field. For each field, cellular fluorescence from Texas Red, FITC or GFP, and DAPI were collected by automated switching of the appropriate excitation and emission filters and passing the emitted light through a multipass beam splitter (set# 84000; Chroma Technology Inc., Brattleboro, VT). The area of each individual cell, and the numbers of cells in each field were determined with image analysis software (ISee Imaging, Cary, NC) as previously described [18, 27]. Briefly, appropriate thresholding criteria allowed cell areas, as determined by areas containing Texas Red fluorescence, to be accurately segmented from the non-fluorescent non-cell areas. Fluorescence background around each cell in the GFP and FITC phalloidin stained images was determined by measuring the average background intensity in a defined region just outside the cell area mask described above. The defined region around each cell was generated by systematically enlarging (dilating) the original cell mask The number of nuclei, and therefore the number of cells, was determined from the corresponding images collected with the DAPI filter. Total intensity for the fluorophore in the images was calculated as follows: mean intensity outside the cell was subtracted from the mean intensity within the cell, and the result was multiplied by the cell area. It is likely that each entire cell is sampled because, in addition to auto-focusing, the objective lens used (10× magnification and 2.5 numerical aperture) has a depth of field of 8.5μm, which is on the order of the full thickness of the fibroblasts. Images were collected by moving the stage automatically in 1 mm steps. Morphology and fluorescence intensity were determined for at least 150 individual cells on each substrate.
Immunofluorescence and phalloidin staining
For immunostaining of actin, paxillin, and phosphorylated FAK Y397, cells were fixed in 4% formaldehyde, permeabilized with 0.1% (v/v) Triton X-100 in DPBS (5 minutes), blocked with DPBS containing 30 mg/mL BSA (blocking solutions) and 50 mg/mL donkey or goat serum at room temperature (1 hour). Appropriate samples were stained overnight at 4°C with 5 μg/mL rabbit anti-phospho FAK Y397 (Biosource, Camarilo, CA) or 20 μg/mL anti-paxillin (Upstate Biotechnology, Lake Placid, NY), in blocking solution. The samples were rinsed multiple times in DPBS and incubated with 10 μg/mL rhodamine-labeled donkey anti-rabbit antibody (Chemicon, Temecula, CA) or Alexa 556 goat anti-mouse as appropriate, in blocking solution at room temperature (1 hour). For filamentous actin (F-actin) staining, cells fixed and permeabilized as above were blocked with DPBS containing 30 mg/mL BSA (30 minutes), stained with Alexa 488 phalloidin (Molecular Probes, Eugene OR) in blocking solution (250 nM, 1 hour), and rinsed with blocking solution. The immunostained and phalloidin-stained samples were rinsed extensively with DPBS and samples were mounted on slides in Tris buffered saline containing 90% glycerol, 2.5 mg/mL DABCO and 0.05 μg/mL DAPI. Cells were imaged with phase and fluorescence microscopy using the appropriate filter sets.
Differences between mean values of GFP intensities and cell area for cells on the various surfaces were tested for statistical significance using a one way analysis of variance test in Sigma Stat software (SPSS Inc. Chicago, IL) followed by a Student-Newman-Keuls post hoc test for pair-wise comparison of cellular behavior (area or GFP intensity) on each substrate. A calculated p value of < 0.05 was considered significant in these comparisons.
The correlation between two variables was tested for statistical significance using Spearman rank order correlation test in Sigma Stat software (SPSS Inc. Chicago, IL). The Spearman correlation measures the strength of association between pairs of variables which cannot be described as having a normal distribution with constant variance, which is the case for the cell area data, the phalloidin data and the GFP intensity data The correlation coefficient, r is a number that varies between -1 and +1 with a correlation of +1 indicating a perfect positive relationship between the two variables, with both variables always increasing together. A P value (the probability of being incorrect in concluding that there is a true association between the variables) was calculated and the lower the P value the greater the probability that there is a direct relationship between the two variables. P < 0.05 was considered significant in these comparisons.