High Throughput Traction Force Microscopy Using PDMS Reveals Dose-Dependent Effects of Transforming Growth Factor-β on the Epithelial-to-Mesenchymal Transition

Summary

We present a high throughput traction force assay fabricated with silicone rubber (PDMS). This novel assay is suitable for studying physical changes in cell contractility during various biological and biomedical processes and diseases. We demonstrate this method's utility by measuring a TGF-β dependent increase in contractility during the epithelial-to-mesenchymal transition.

Abstract

Cellular contractility is essential in diverse aspects of biology, driving processes that range from motility and division, to tissue contraction and mechanical stability, and represents a core element of multi-cellular animal life. In adherent cells, acto-myosin contraction is seen in traction forces that cells exert on their substrate. Dysregulation of cellular contractility appears in a myriad of pathologies, making contractility a promising target in diverse diagnostic approaches using biophysics as a metric. Moreover, novel therapeutic strategies can be based on correcting the apparent malfunction of cell contractility. These applications, however, require direct quantification of these forces.

We have developed silicone elastomer-based traction force microscopy (TFM) in a parallelized multi-well format. Our use of a silicone rubber, specifically polydimethylsiloxane (PDMS), rather than the commonly employed hydrogel polyacrylamide (PAA) enables us to make robust and inert substrates with indefinite shelf-lives requiring no specialized storage conditions. Unlike pillar-PDMS based approaches that have a modulus in the GPa range, the PDMS used here is very compliant, ranging from approximately 0.4 kPa to 100 kPa. We create a high-throughput platform for TFM by partitioning these large monolithic substrates spatially into biochemically independent wells, creating a multi-well platform for traction force screening that is compatible with existing multi-well systems.

In this manuscript, we use this multi-well traction force system to examine the Epithelial to Mesenchymal Transition (EMT); we induce EMT in NMuMG cells by exposing them to TGF-β, and to quantify the biophysical changes during EMT. We measure the contractility as a function of concentration and duration of TGF-β exposure. Our findings here demonstrate the utility of parallelized TFM in the context of disease biophysics.

Introduction

Acto-myosin contractility is an essential element of active cell mechanics, impacting cell behaviors from motility and proliferation to stem cell differentiation. In tissues, contractility drives activity from polar separation in embryogenesis, to airway constriction and cardiac activity. Critically, to generate tension, cells must first adhere to their extracellular environment. In doing so, this contractility generates traction forces on their surroundings. Traction Force Microscopy (TFM) has emerged in a multitude of forms as a way to quantify these forces from diverse cells under different conditions.

The field of TFM has seen an exceptional breadth of innovation and application, and the results have paved the way for new perspectives in biology, which incorporate mechanics and physical forces. Starting with wrinkling silicone substrates¹, researchers have applied various techniques to measure cell traction forces. These approaches have been continuously improved and have now reached a level of resolution on the order of several microns². However, one principal problem has emerged, which is the difficulty in creating substrates of suitably low moduli using the available silicones. To circumvent this problem, polyacrylamide was adopted as a replacement due to the ease of creating substrates on the order of 1-20 kPa³. We recently implemented very compliant silicones in TFM⁴, allowing us to fabricate the same range of moduli as polyacrylamide, but with the advantages of inert and robust silicone.

TFM approaches have enabled valuable mechano-biological discoveries, however, a persistent shortcoming is their complexity, often restricting their use to researchers in the engineering or physical sciences disciplines. This is due in large part to the detailed calibrations and challenging calculations that are required to quantify contractility. Another significant challenge is that TFM methods are largely low-throughput and therefore ill-suited to study many different conditions or populations simultaneously⁵. This has presented a bottleneck, which has hampered transfer of TFM from a specialist biophysics setting into broader biological and pharmacological applications.

We have recently developed a multi-well format TFM plate, which allows researchers to parallelize their TFM measurements for faster quantification of contractility metrics, while exploring the impact of different compounds and also using less reagents⁴. This methodology has broad utility in diverse mechanobiology studies, from evaluating the effects of compounds on cellular activity, to quantifying the contractile changes in differentiation or disease.

One area of biomedical research that will benefit greatly from TFM is the study of how physical cues impact the malignant phenotypes of cancer cells. Metastasis, responsible for 90% of cancer-related deaths, is characterized by cancerous cells leaving their original tumor site and colonizing a secondary site. For cells to migrate through tissue and pass in and out of the vascular system, they must radically change their shapes to squeeze through these physical barriers while generating substantial forces to pull their way along extracellular matrix or move between other cells. These forces are transmitted to the substrate through focal adhesion interactions^2,3, and can be quantified using TFM. While cancers are biochemically exceptionally diverse, with an expanding repertoire of known mutations and protein changes, some common physical changes have been observed; in a variety of cancers, including breast, prostate, and lung cancers, metastatic cells have been shown to exert 2-3 times the traction forces of non-metastatic cells^6,7,8. These results suggest that there may be a strong correlation between metastatic progression and the traction forces exerted by cells; however, the detailed time-dependent changes in contractility are difficult to examine.

The epithelial-to-mesenchymal transition (EMT) is a process whereby cells reduce adherens- and tight-junction mediated cell-cell adhesion, becoming more migratory and invasive. In addition to physiological functions that include wound healing and developmental processes, EMT is also a process exploited during metastasis, making it a useful model system to study this process. Using TGF-β, we can induce the EMT in murine mammary epithelial cells (NMuMG)⁹ to directly quantify the physical changes during this transformation, and characterize the time and dose-dependent effects of TGF-β on EMT and cell contractility. In this article, we demonstrate the utility of this approach by measuring the changes in contractility during an induced EMT.

Protocol

NOTE: The following protocol will guide researchers in fabricating and using the multi-well TFM dish shown in Figure 1.

1. Preparation of PDMS silicone substrates

1. Preparation of PDMS silicone rubber mixture based on a composite mixture of two commercially available kits.
   1. Add Part A and Part B of PDMS kit (e.g., GEL-8100, see the Table of Materials) in a 1:1 weight ratio into the 50 mL tube.
      NOTE: The mixture is mixed on a rotator at a speed slow enough for the mixture to flow back and forth during revolution to ensure complete mixing.
   2. Add the required amount of curing agent for the desired modulus of the substrate.
      NOTE: The amount of curing agent to be added to the mixture depends on the desired modulus of the substrate and may typically range from 0.1% to 1.8%. Refer to Table 1 and Figure 3 for a guide to specific crosslinker concentrations and resulting moduli.
   3. Mix the formulation on the rotator for 30-45 min; ensure the rotation is slow enough for thorough mixing.
2. Bottom layer: coating PDMS substrates on the glass slide
   1. Place the custom-built chuck illustrated in Figure 2 on the spin-coater. Clean the glass with ethanol or isopropanol, and dry with a lint-free wipe. Place the glass slide in the chuck and turn on the vacuum to hold the slide in place.
   2. Apply uncured PDMS approximately 1 cm from the edges and work in towards the center. Apply enough (3-4 mL) PDMS to ensure the whole surface will be covered.
      NOTE: To ensure that the PDMS is evenly spread on the surface of the glass, a pipette tip may be helpful to spread the PDMS mixture from the center to the edges.
   3. Spin the glass with the PDMS mixture on a spin-coater with the following protocol.
      1. To spread the uncured PDMS on the slide, accelerate at 50 rpm/s from 0 to 200 rpm; hold at 200 rpm for 1 min.
      2. To achieve a 100 µm thick PDMS layer, accelerate at 50 rpm/s to 300 rpm and hold for 1 min at 300 rpm. Different desired thicknesses other than 100 µm will require specific speed values.
      3. To remove, decelerate at 50 rpm/s to 0 rpm. Disable vacuum and remove the coated slide, taking care not to touch the coated surface.
         NOTE: It is important to include the acceleration and deceleration steps to ensure a smooth continuous surface.
         CAUTION: To ensure that the sample does not fly off the chuck, the custom-made holder should be used to hold the slide, and not rely simply on vacuum and the existing flat chuck. The details and specifications of this holder are given in Figure 2.
   4. Place the spin-coated sample in the oven at the manufacturer recommended temperature (100 °C) for 2 h.
      NOTE: The surface of the oven where the sample is placed should be solid (i.e., not a wire rack) and level surface to ensure the uniform heating and thickness of the sample. A ceramic or steel plate makes an ideal surface.
3. Top bead layer
   1. Add bead solution in the appropriate ratio to the remaining PDMS mixture.
      NOTE: This ratio depends on the concentration of the stock bead solution and the desired bead density on the sample. Typical final values are 9.2 x 10¹¹ beads/mL and 0.05-0.2 beads/µm², and an excess of beads is generally preferable to an inadequate amount.
   2. Mix the bead solution with the uncured PDMS. This may be accomplished by placing the tube on a rotator for approximately 30 min, vortexing for 1-2 min, or sonication for 30 min. These methods may be combined.
      NOTE: In our application, we find that sonication is effective in breaking bead aggregates, and rotation is effective in mixing. Synthesized fiduciary beads may aggregate in storage. Prior to use, they may be resuspended in hexane and sonicated. If there are significant large aggregates, one can filter the bead suspension through a 5 µm syringe filter. This filtration step is optional; it helps to coat the sample with monodispersed beads, but a significant fraction of beads may be lost in the filter.
   3. Take out the slide from the oven, allow to cool to room temperature, and place it on the spin-coater.
   4. Add 3-4 mL of the bead and uncured PDMS mixture onto the surface of the coated sample.
      NOTE: The mixture with beads added is less viscous due to the hexane. Make sure not to touch the surface of the substrate as it may damage the already-coated PDMS substrate. Additionally, the bead mixture may initially not wet the surface; take care that the mixture does not immediately flow off the cured PDMS surface.
   5. Spin the sample with the following protocol.
      1. To spread the bead and uncured PDMS mixture, accelerate at 100 rpm/s from 0 to 500 rpm; hold at 500 rpm for 1 min.
      2. To achieve a thin layer of bead-embedded PDMS (~1 µm), accelerate at 200 rpm/s from 500 to 5000 rpm; hold at 5000 rpm for 10 s.
      3. To remove, decelerate at 100 rpm/s to 0 rpm. Disable vacuum and remove coated slide, taking care not to touch the coated surface.
   6. Place the spin-coated sample in the oven at 100 °C for 1 h.
      NOTE: Temperatures above 100 °C or durations longer than 1 h can reduce the bead fluorescence. Make sure that the oven temperature is set to 100 °C and not higher.
   7. The protocol can be paused here. To store the sample, cover the surface to avoid dust and light exposure. Make sure that nothing touches the surface. The sample is shelf-stable at room temperature indefinitely.
4. Assembling the plate
   1. Add Part A (base) and Part B (curing agent, e.g., Sylgard) of a PDMS elastomer kit in a 10:1 weight ratio into the 50 mL tube.
   2. Mix the mixture on the rotator for 30-45 min.
      NOTE: For one plate, mix 5 mL of base with 0.5 mL of curing agent. Up to 1 mL of hexane can be added to reduce the viscosity of the mixture.
   3. Apply the mixture to the bottom of the divider and spread the mixture.
      NOTE: The divider should be placed upside-down.
   4. Lay the substrate onto the divider upside-down.
   5. Place the sample upside-down in the oven at 65 °C for 2 h.
   6. Take out samples from the oven and clean the bottom of the glass with 70% ethanol or isopropanol to remove any PDMS residue.
      NOTE: The protocol can be paused here. To store the sample, place a lid on the substrate and wrap the device in aluminum foil to avoid exposure to the light. The sample is shelf-stable at room temperature indefinitely. The divider utilized in this method is in a 96-well format; however, researchers may employ other formats (384-well, 2-well, 4-well, 8-well, etc.) depending on desired experiment setups and availability of dividing structures. Some further optimization may be required.

2. Surface functionalization

1. Dissolve 80 µL of a Sulfo-SANPAH aliquot in 40 mL of 0.1 M HEPES buffer (pH 7-9).
   NOTE: Prepare Sulfo-SANPAH aliquots by dissolving 100 mg of Sulfo-SANPAH powder in 2 mL of sterile dimethyl sulfoxide (DMSO). Prepare 0.1 M HEPES buffer by diluting 50 mL of HEPES in 450 mL of sterile deionized water and filter through a 0.22 µm pore filter.
2. Add 200 µL of diluted Sulfo-SANPAH solution to each well of the 96-well plate.
3. Expose the plate to UV (300-460nm) light at appropriate distance and duration.
   NOTE: After UV exposure, the color of the solution should be darker. UV exposure distance and duration depend on the UV lamp power. In our application, we expose for 10-15 min at a distance of 5 cm.
4. Remove the Sulfo-SANPAH solution from the wells and add 200 µL of 5 µg/mL fibronectin solution to each well.
   NOTE: Researcher-specified protein can be used for surface coating. Some commonly used proteins are collagen, fibronectin, and laminin. We have found Sulfo-SANPAH to be the most effective method, and plasma cleaning while sometimes employing in PDMS is discouraged as it creates a silicon dioxide layer and visibly damages the surface.
5. Incubate the plate at 4 °C overnight.
   NOTE: Different incubation methods can be applied depending on the protein used for coating.
6. Remove the fibronectin solution and wash each well with phosphate-buffered saline (PBS) twice.
7. Place a lid on the sample.
8. Add 200 µL of PBS to each well.
   NOTE: The protocol can be paused here. The fibronectin-coated samples can be stored at 4 °C for up to 2 weeks.

3. UV sterilization

1. Sterilize the sample under UV light in a biological safety cabinet for 30 min.
   NOTE: Longer UV exposure times may negatively impact the bead fluorescence. All subsequent steps must be performed under sterile conditions.

4. Cell culture

1. Remove PBS from each well and add 200 µL of cells suspended in culture media to each well.
   1. Plate the cells at the desired cell density. Cell density depends on the desired experiment. For single cell studies, cells should be minimum of 50 µm apart and cells near the edges of the imaging window should be not be included in the TFM measurement. For monolayer cells, the imaging window should have the viewing field covered with a confluent layer of cells.
   2. Prepare the complete growth culture media for NMuMG cells by supplementing DMEM with 5% FBS, 10 mM HEPES, 10 µg/mL insulin, 1% penicillin-streptomycin, 1 mM L-glutamine, and 0.5 µg/mL amphotericin B.
   3. Prepare insulin stock by reconstituting in acidified water (2.5 mL of glacial acetic acid in 130 mL of deionized water) to a concentration of 10 mg/mL. Store the stock solution at 4 °C. Wait until the solution is clear, and then filter through a 0.22 µm pore filter.
2. TGF-β addition
   1. To prepare of TGF-β stock, dissolve 2 µg of TGF-β in 100 µL of 10 mM citric acid (pH 3.0) and filter sterilize with 0.22 µm pores. Vortex the tube and aliquot into the desired volumes. Store the aliquots at -80 °C.
   2. Add 1.5 µL of TGF-β stock solution to 10 mL of the complete cell culture media to constitute the cell culture media with the final TGF-β concentration of 3 ng/mL.
      NOTE: To make 10 mM pH 3.0 citric acid, dilute the acid in water and adjust pH to 3.0 by adding HCl.

5. Data acquisition

1. For each position, acquire at least one image of fiduciary particles and cells. Focus on the bead layer.
   NOTE: Pixel size should be optimized based on size of the fiduciary particles and image processing method being used. In this application, the authors use a 10x 0.4 NA objective, and images are acquired with 1024 x 1024 resolution, with 455 nm/pixel. In general, it is helpful to retain a resolution of at least approximately 1-5 pixels per bead; here, beads are polydisperse and have an individual size of 300-500 nm. It is critical that the fluorescent fiduciary beads be in focus for images to be used for TFM calculations. The focus and imaging quality of the beads should be prioritized over imaging the cells themselves. There should be no cross-talk between different channels, particularly any fluorescence not from the fiduciary beads which appears in the imaging spectra of the beads.
2. Once all the positions of interest have been recorded, add detachment solution to each well to acquire force-free reference images of the fiduciary particles.
   1. To prepare cell detachment solution, mix an aqueous solution containing 2% TritonX-100, 50 mM sodium azide, and 500 mM potassium hydroxide.
      NOTE: The above is provided as an example of an effective detachment solution. Different detachment solutions at researchers' discretion may be used to detach the cells off the substrate surface.

6. Image analysis

1. Perform appropriate image analysis as desired.
   NOTE: Analysis software was developed in-house. Image analysis may be done with custom-made software or software available online.

7. Bead synthesis

NOTE: The following protocol is based on the synthesis method described by Klein et al.¹⁰.

1. Under a fume hood, prepare the three-neck flask with a water-cooled reflux condenser.
   CAUTION: Set up a synthesis in a well-ventilated chemical fume hood.
2. Add 0.5 mL of PDMS stabilizer and fluorophore to the flask.
3. Equip one neck with a rubber septum with a nitrogen inlet needle and an outlet needle and equip the other neck with a rubber septum for adding reagent with a syringe.
4. Add 100 mL of anhydrous hexane in 250 mL to the flask and add a small magnetic stir bar.
5. Place the flask in the mineral oil bath at 75 °C and purge it with nitrogen gas for 1 h.
6. Add 6 mL of methylmethacrylate to a 25 mL round bottom flask.
7. Add 0.100 g of 2,2'-azobisisobutyronitrile (AIBN) to the round bottom flask and purge the mixture with nitrogen gas for 1 h.
   NOTE: Flush methylamethacrylate through prepacked column to remove inhibitors before use. Add methylamethacrylate and AIBN mixture to the three-neck flask.
8. The solution initially becomes cloudy and turns milky. Let the reaction run for 3 h after the solution becomes cloudy.
9. After 3 h, place the flask in an ice water bath.
10. Vacuum-filter the solution with coarse filter paper.
11. Centrifuge the filtrate and re-suspend the particles in hexane.
    NOTE: The volume of the hexane to be added depends on the desired concentration of the bead solution. Sonication facilitates re-dispersion of the bead particles in hexane solution. Beads produced by the authors have polydisperse diameters of approximately 300-500nm. Due to the use of cross-correlation pattern tracking to determine displacements, monodisperse beads are not required.

8. Rheology measurement protocol

NOTE: Rheology is not required for every researcher or experiment, but is necessary to quantify the moduli for new formulations of PDMS. In this protocol, we employ a shear rheometer to measure the effects of crosslinker, frequency, and strain on moduli of PDMS samples. Depending on the available tools and expertise, moduli may also be measured using many other mechanical analysis approaches. Additionally, researchers using this protocol may elect to use our published moduli presented in Table 1, Figure 3 and Figure 4.

1. Use a rheometer with a 25 mm diameter parallel plate geometry. Other geometries may be used.
2. Initialize the system and calibrate the device and measuring system (parallel plate, d = 25 mm). After measuring the zero gap, begin loading the PDMS sample.
3. As soon as the PDMS elastomer and crosslinking agent have been mixed, pipet the mixture onto the bottom plate of rheometer.
   1. Move the spindle down to completely contact the top of the PDMS sample.
   2. Carefully trim the loaded sample excess from the bottom plate.
4. In a strain sweep test, apply increasing strains for each composition with different crosslinking density to ensure the polymer structure remains in the linear viscoelastic regime during all shear measurements.
   NOTE: Strain values relevant for cell studies are typically in the range of 0.1-10%. We have found PDMS to be linear up to approximately 100% strain.
5. Measure the dynamic shear storage modulus (G'), and loss modulus (G′′) of the PDMS network in a time sweep test with frequency of 1 Hz and oscillatory shear strain of 0.5% at 100 °C.
6. To determine the viscoelasticity and time dependency of the final PDMS network, apply a frequency sweep test with frequency ranging 0.1-100 Hz and oscillatory shear strain of 0.5%.

Results

Before addition of TGF-β, a confluent monolayer of cells has a cobblestone like shape and is tightly packed. Upon TGF-β treatment, cells become more elongated in morphology, enlarging the cell area and acquiring a more mesenchymal phenotype. Utilizing the multi-well device fabricated with soft PDMS elastomers, the physical properties of cells in a total 17 different conditions were studied. The cells were treated with four different TGF-β concentrations (0.5, 1, 2, and 4 ng/mL) and four d...

Discussion

For the success of this method, it is critical to have a uniformly coated sample with a constant thickness of approximately 100 µm. The modulus should be carefully chosen to examine the physical significance of the biological system of interest. When fabricating a top layer, the concentration of the fiducial fluorescent particles should be optimized for accurate analysis of displacement and traction stress. Analyzing isolated single cells requires a denser fiduciary layer than measuring confluent monolayers. Additio...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Plate
GEL-8100	Nusil Technology	GEL-8100	High Purity Dielectric, Soft Silicone Gel kit
Dow Corning Sylgard 184 Silicone Encapsulant Clear 0.5 kg Kit	Ellsworth Adhesives	184 SIL ELAST KIT 0.5KG	curing agent
Custom Cut Glass	Hausser Scientific Company		109.6mm± x 72.8mm± x 1mm thickness
Target 2TM Nylon Syringe Filter	ThermoFisher Scientific	F2513-4
96-well Stripwell Egg Crate Strip Holder	Corning	2572
Polystyrene Universal Microplate Lid With Corner Notch	Corning	3099
Ethyl alcohol	Greenfield Global	P016EA95	0.95
2-Propanol	Sigma-Aldrich	190764	ACS reagent, ≥99.5%
Surface Coating
Sulfo-SANPAH Crosslinker	Proteochem	c1111-100mg
Fibronectin bovine plasma	Sigma-Aldrich	F1141-1MG	solution, sterile-filtered, BioReagent, suitable for cell culture
PBS, 1X	Wisent	319-005-CL	pH 7.4, without calcium and magnesium
DMSO	Sigma-Aldrich	472301
Cell Culture
DMEM, 1X	Wisent	319-005-CL	4.5g/L glucose, with L-glutamine, sodium pyruvate and phenol red
FBS (Fetal Bovine Serum)	Wisent	080-150	Premium Quality, Endotoxin <1, Hemoglobin <25
HEPES	Wisent	330-050-EL	1M, free acid
Human Insulin Recombinant	Wisent	511-016-CM	USP grade
Penicillin-Streptomycin Solution	Wisent	450-201-EL	100 X, sterile filtered for cell culture
L-Glutamine solution	Wisent	609-065-EL	200mM solution, sterile filtered for cell culture
Amphotericine B	Wisent	450-105-QL	250μg/ml, sterile filtered for cell culture
Recombinant Human TGF-β1	Peprotech	100-21	HEK293 Derived
Acetic acid	Sigma-Aldrich	537020	Glacial, ≥99.85%
Cictric acid	Sigma-Aldrich	251275	ACS reagent, ≥99.5%
NMuMG	ATCC	CRL-1636	Mouse Mammary Gland Cell Line
Sodium azide	Fisher Schientific	AC190385000	99%, extra pure, ACROS Organics
Potassium hydroxide	Sigma-Aldrich	221473	ACS reagent, ≥85%, pellets
TritonX-100	Sigma-Aldrich	X100	laboratory grade
Bead Synthesis
1,1′-Dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI)	Sigma-Aldrich	468495-100MG	97%
Methyl methacrylate	Sigma-Aldrich	M55909-500ML	contains ≤30 ppm MEHQ as inhibitor, 99%
Inhibitor Remover	Sigma-Aldrich	306312-1EA	Prepacked column for removing hydroquinone and monomethyl ether hydroquinone
Methacryloxylpropyl Terminated Polydimethylsiloxane	Gelest	DMS-R31 (25,000g/mol)	Polydimethylsiloxane stabilizer, 25,000g/mol, 1,000 cSt
2,2′-Azobis(2-methylpropionitrile) (AIBN)	Sigma-Aldrich	441090-25G	98%
Hexane	Sigma-Aldrich	296090-2L	anhydrous, 95%
Hexane, mixture of isomers	Sigma-Aldrich	227064-1L	anhydrous, ≥99%
Whatman qualitative filter paper, Grade 1	Sigma-Aldrich	WHA1001055	circles, diam. 55 mm,
Equipment
Laurell WS-650Mz-23NPPB	Laurell Technologies
UVP Handheld UV Lamp Model UVGL-58	VWR	21474-622
Rheometer	Anton Paar	MCR 302 WESP