Cell culture

Madin-Darby Canine Kidney (MDCK) type II cells (purchased from Millipore Sigma) were maintained in low-glucose Dulbecco’s modified Eagle’s Medium (DMEM, 10–014, Corning) supplemented with 10 fetal bovine serum (FBS, Corning) and 1% Penicillin-Streptomycin (Corning). Human keratinocytes (HaCaTs, provided by Professor Kristyn Masters’ Lab) were maintained in high-glucose Dulbecco’s modified Eagle’s Medium (DMEM, 10–013, Corning) supplemented with 10 fetal bovine serum (FBS, Corning) and 1% Penicillin-Streptomycin (Corning). Cells were maintained at 37^°C and 5% CO₂. Experiments involving human induced pluripotent (hiPS) stem cells reported in this paper were approved by the Human Stem Cell Research Oversight Committee (hSCRO) at the University of California Irvine, study no. 2018–1072. We used early passages (50) of hiPS cell line CTR217 (which were a gift from Kenneth Kosik’s lab and first used for this study on 04/15/2025). The cell line was previously karyotyped as normal. The hiPS cells were cultured with mTESR+ medium (Stemcell Technologies 100–0276) on Matrigel (Corning 354277)-coated 6-well plates. Cell culture medium was exchanged daily, and cells were maintained at 37^°C and 5% CO₂.

Method 1 Sample Preparation and optical microscopy

Polyacrylamide (PA) gels of 6 kPa Young’s modulus were prepared using previously described methods [12,28,32]. To begin, the surface of a glass bottom dish with a 1.5 cover glass was covered with a solution containing 0.2 (v/v) acetic acid and 0.3 (v/v) 3-(Trimethoxysilyl)propyl methacrylate 98 in ultrapure water and allowed to sit for 30 min. The dishes were then washed in ultrapure water and dried. A solution of polyacrylamide and bisacrylamide (Bio-Rad) (see [33] or similar literature for recipes with different Young’s modulus) containing fluorescent particles (Life Technologies) was prepared. Once the ammonium persulfate is added, it is important to work quickly, and polymerization times might need to be adjusted depending on the user. This can be done by changing concentrations of N,N,N´,N´-tetramethylethylenediamine (TEMED, Bio-Rad) and ammonium persulfate (Bio-Rad). 20 μL of the polyacrylamide solution was pipetted onto the center of the dish, and a 18 mm coverslip was gently placed over the top of the gel solution. It is important that the coverslip be in the middle of the dish and not touching the plastic on the sides to minimize leakage. Next, the dishes were covered with a lid, stacked, taped together, and placed upside down in a centrifuge. It is important to prevent any rattling of the dishes as it can contribute to leakage. To this end, dishes were stacked and wrapped with paper towels before placing them into the centrifuge buckets. Dishes were centrifuged upside down for 14 min at 60–80 relative centrifugal force. The gel typically polymerized in about 5–7 min, and the longer centrifugation time was used as a precaution. Ultrapure water was added to each dish to cover the gels, and the gels were allowed to swell overnight. Next, we followed our stencil-based micropatterning protocol to micropattern 0.1 mg/mL type I rat tail collagen (BD Biosciences) and seed the cells. Samples were then imaged using an Eclipse Ti-E microscope (Nikon) running on Elements AR software with a 10 × numerical aperture 0.5 objective (Nikon) and an Orca Flash 4.0 digital camera (Hamamatsu).

Some experiments imaged only the collagen I bound to the substrate. For these experiments, samples were prepared according to the protocol, but no cells were seeded on the surface. Collagen I was fluorescently labeled using a polyclonal collagen I antibody 1:300 (ThermoFisher, cat. PA 1–36145), and dishes were incubated overnight at 4^°C in the antibody solution. A secondary antibody, Alexa Fluor 488 goat anti-rabbit antibody (Invitrogen, cat. R37116), was added according to manufacturer instructions, and incubated at 4^°C for 4 hr. Samples were imaged on a Nikon A1R confocal microscope with a numerical aperture 0.13 objective (Nikon) using NIS-Elements Ar software (Nikon) and an Orca Flash 4.0 digital camera (Hamamatsu).

Traction force microscopy and monolayer stress microscopy

Cell velocities were computed using Fast Iterative Digital Image Correlation (FIDIC) [34] on successive phase-contrast images of the cell monolayers using 64 × 64 pixel subsets with spacing of 16 pixels and dividing by the time between images (10 or 15 min).

For traction experiments, the particles in the substrate were imaged via fluorescence and the cells were imaged via phase contrast every 10 or 15 min. Then the cells were removed from the substrate with trypsin, allowing the substrate to recover to an undeformed reference state, and a reference image of the fluorescent particles was collected. Cell-induced substrate displacements were computed by correlating the cell-deformed image of fluorescent particles to the reference one with FIDIC [34] using the same subset size and spacing as described above. Cell-substrate tractions were computed via unconstrained Fourier Transform Traction Cytometry [35] with a correction for finite substrate thickness [36,37]. The in-plane components of the stress tensor were computed using Monolayer Stress Microscopy [32,38,39]. The maximal and minimal eigenvalues of the stress tensor are the first and second principal stresses, and , respectively. The maximal shear stress was computed by taking half the difference in principal stresses ().

Method 2 sample preparation and optical microscopy

Soft elastomeric silicone substrates were prepared using established protocols. In brief, a silicone elastomer was synthesized by mixing a 6:5 weight ratio of CY52-276A and CY52-276B polydimethylsiloxane (Dow Corning Toray). After degassing for 5 min, the elastomer solution was spin-coated onto a 22 mm round coverslip (10200−036, VWR) for 60 s at 500 rpm followed by a spin-off step of 1000 rpm for 3 s. A 35 mm plastic bottomed petri-dish (Corning 351008) with a custom pre-drilled 20 mm hole was then placed onto the PDMS coated coverslip to create a sealed PDMS-coated glass-bottomed dish. These dishes were then cured at 65^°C overnight. The substrates were kept in a clean dust-free and dry environment and always used within 4 weeks of fabrication.

After curing, the soft PDMS substrates were treated with (3-aminopropyl)triethoxysilane (APTES, Sigma-Aldrich, cat. no. A3648) diluted at 10 in absolute ethanol (Sigma, 459844) for 4.5 min, rinsed 3 times with ethanol, followed by a rinse with deionized water and carefully aspirated without touching the substrate surfaces until thoroughly dry. To attach fluorescent beads as fiducial markers, PDMS substrates were incubated for 17 hr with a solution containing 400-nm-diameter sky-blue fluorescent carboxylate-modified beads (1:50 v/v dilution, Spherotech, CFP-0570–2), MES buffer (0.5M pH 5.0, ThermoFisher J62081.AE), and 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC, 0.4 mg/ml, Sigma-Aldrich). Next, substrates were rinsed 3 times with deionized water and immediately used in photopatterning steps.

Images were recorded on a Leica SP8 confocal microscope with a 633 nm laser at 0.75 laser power utilizing a HyD detector with a 643–743 nm bandpass detector window. Images were recorded with 20 × magnification using a resolution of 2048 × 2048 pixels at 8-bit depth.

Method 1

Both complete protocols can be found in Supporting Information Files 1 and 2. In Method 1, a photomask with the desired geometries is designed and printed. Then, SU-8 is heated, and a ∼400 μm thick layer is spread onto a silicon wafer with a razor blade. This 400 μm thickness is larger than the thickness commonly used for many microfluidics assays, and it is chosen to facilitate handling of the PDMS masks in later steps. The wafer undergoes a soft bake. Once the wafer has returned to room temperature, the photomask is secured over the wafer, and the wafer is exposed to UV light, which crosslinks the exposed SU-8. The SU-8 that went unexposed is developed away, leaving the desired features behind. The wafer is then silanized to prevent PDMS from sticking to the surface. A mask is created by spin coating a thin layer of PDMS onto the chip and curing overnight in the oven. The mask is then removed, with the result being a pliable PDMS sheet with through-holes in the shape of the features. Next, the masks were treated with oxygen plasma and attached to the surface of a polyacrylamide hydrogel. Sulfo-SANPAH is added over the mask, and exposed to UV light. Collagen I is then added and allowed to covalently bind to the substrate overnight. At this point, the mask is removed and the gel is treated with bovine serum albumin to prevent non-specific adhesion. Finally, cells are seeded onto the gel, and are confined to the desired geometries. A schematic of this procedure is illustrated in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Schematic of Method 1.

https://doi.org/10.1371/journal.pone.0344657.g001

To assess the fidelity of Method 1, we compared the intended geometries with both the deposited extracellular matrix and the resulting confined cell monolayers. For this, we designed three patterns, a circle, a square, and a five-pointed star, all having widths of approximately 1 mm (Fig 2 a,b,c). The three patterns are useful for testing, as they provide a range of complexity, with the circle being the simplest and the star being the most complex. We began by patterning the polyacrylamide substrate with collagen I and labeling it fluorescently to visualize how well the deposited collagen matched the initial design. Across all geometries, the collagen deposition generally matched the design (Fig 2 d,e,f). The collagen transfer was not as strong at sharp corners, as seen in the corners of the square and the tips of the star (Fig 2 e,f). On separate micropatterned substrates, we seeded MDCK cells on the patterns, as described in the protocol. The resulting monolayers adhered well to the patterned regions (Fig 2 g,h,i). Cell confinement was robust in all cases, although some bridging was observed in regions of acute angles in the star pattern (Fig 2i). It is important to note that the passivation of non-patterned regions with bovine serum albumin (BSA) prevents nonspecific cell adhesion and limits spreading outside the collagen-treated areas. To ensure this method is compatible with other cell types, we confined human keratinocytes to 1 mm circular islands (S3 Fig S1 in S3 File).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Pattern fidelity of Method 1.

(a) Schematic of the 1 mm circular pattern. (d) Patterned collagen I. (g) Epithelial cells confined by the pattern. (j) Binarized cell island. Panels (b,e,h,k) and (c,f,i,l) repeat the patterning for square and star geometries, respectively. (m) Dice coefficient for circle, square, and star geometry. (n) Area difference between initial design and confined cell monolayers for square, star, and circle geometry. Gray dots represent individual cell monolayers, and black bars indicate means. At least 7 monolayers were used per shape.

https://doi.org/10.1371/journal.pone.0344657.g002

To quantify pattern fidelity, we calculated the Dice coefficient between the designed patterns and the confined cell monolayer. The Dice coefficient was calculated according to , where A is the designed confinement mask (Fig 2 a,b,c), B is the binarized mask of the confined cell monolayers (Fig 2 j,k,l), and is the intersection of A and B. Each point in A and B is a pixel, and each pixel has a width of 0.65 μm. The Dice coefficient ranges from 0 to 1, with 1 indicating perfect overlap between A and B, and 0 indicating no overlap. The mean Dice coefficient was greater than 0.9 for both the circle and square patterns, indicating high pattern fidelity (Fig 2m). For the star pattern, the mean Dice coefficient was ≈ 0.8, due to reduced fidelity in sharp and narrow regions. All patterns had Dice coefficients above 0.8, demonstrating that the confined cell monolayer matched the designed confinement well. We next quantified the area difference between the designed pattern and the confined cell monolayer. The area difference is calculated as , where C is the area of the designed confinement mask in pixels, and D is the area of the binarized mask of the cell monolayer. The circular pattern showed the smallest difference in area of ∼5 (Fig 2n). The square and star pattern had larger area differences of ∼10 and ∼30 , respectively. As expected, increasing pattern complexity led to decreased pattern fidelity, as more intricate designs were more susceptible to incomplete pattern transfer or cell bridging. However, the Dice coefficient and area difference metrics show that our protocol achieves high pattern fidelity, particularly for geometries with smooth, continuous boundaries.

Method 2

Method 2 (Supporting Information File 2), which is used to pattern PDMS substrates, has been previously established in similar systems [40–44]. Briefly, a PDMS substrate is fabricated on a glass sheet. A photopatterning solution composed of PLL-PEG is placed between the substrate and a photomask. Next, the sample is exposed to UV, which degrades the PEG in the illuminated regions, allowing for protein conjugation. Finally, the sample is washed and an ECM is added, which binds to the UV-exposed areas, as illustrated in Fig 3. Cells are then seeded onto the substrate and allowed to adhere. A description of the method, and a custom UV setup is described in the Supporting Information.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Schematic of Method 2.

https://doi.org/10.1371/journal.pone.0344657.g003

We also quantified the pattern fidelity for Method 2. Fluorescently labeled bovine serum albumin (BSA) of protein deposition and confined monolayers showed strong alignment with the designed pattern (Fig 4a-c). The Dice coefficient was calculated as described above, and was greater than 0.9, indicating high pattern fidelity (Fig 4e). Area difference was quantified as described above, and was less than 5, demonstrating that the confined cell monolayer matched the designed confinement well (Fig 4f). Together, these results show that both Method 1 and Method 2 achieve high pattern fidelity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Pattern fidelity of Method 2.

(a) Schematic of the 300 μm circular pattern. (b) Patterned BSA-488. (c) Epithelial cells confined by the pattern. (d) Binarized cell island. (e) Dice coefficient for circle geometry. (f) Area difference between initial design and confined cell monolayers for the circle geometry. Gray dots represent individual cell monolayers, and black bars indicate means. Results are from 6 cell monolayers.

https://doi.org/10.1371/journal.pone.0344657.g004

To illustrate how this protocol can be used to study tissue mechanics and collective cell behavior, we applied Method 1 to confine MDCK monolayers to circular islands of three different diameters, 300 μm, 400 μm, and 1 mm (Fig 5a-c), and imaged over several hours. The 300 μm and 400 μm islands were generated using lithography-based micropatterning, while the 1 mm island was created using a biopsy punch, which is a convenient alternative for producing patterns with features mm. Both approaches are detailed in Method 1 (Supporting Information File 1). For each island size, cells filled the island. Occasionally, some cells appeared outside of the micropatterned area. These cells had rounded morphology, suggesting that they were not well adhered. This system allows for controlled study of how the size of an epithelial monolayer influences the magnitude and spatial organization of cellular velocity. For each cell island, we quantified the cell velocity using digital image correlation. The resulting heat maps of velocity magnitude overlaid with velocity vectors are shown in Fig 5d-f. Smaller cell islands exhibited smoother and more coherent patterns of motion compared to the 1 mm island, which had visible vortices in the velocity field. Consistent with prior observations [9,10] the 300 μm island showed rotational motion about the center of the island (Fig 5d). Interestingly, the average velocity magnitude in the 300 μm and 400 μm island was approximately half that of the 1 mm islands, suggesting that the confinement caused by the smaller island size reduced the average cellular motion. Additionally, further analysis can be done to quantify the motion of the cell islands. For example, the distribution of velocity magnitude and direction can be seen in Fig 5g,h. These methods have been previously used to quantitatively characterize cell motion and its relationship to force generation in epithelial monolayers, including analyses of stress-strain rate relationships and stress-cell shape relationships [45–47]. Together, these results demonstrate that our protocol enables controlled investigation of confinement size-dependent collective motion in epithelial monolayers.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Representative experiment showing effects of island size on cell velocity.

(a-c) Phase contrast image of 300 μm, 400 μm, and 1 mm diameter cell islands. (d-f) Heat map of velocity magnitude overlaid with arrows indicating velocity direction and magnitude for 300 μm, 400 μm, and 1 mm cell islands. (g) Histogram of the velocity magnitude over 1 cell island over 15 hr. (h) Histogram of the direction of velocity over one cell island over 15 hr.

https://doi.org/10.1371/journal.pone.0344657.g005

To demonstrate that this protocol can be used in conjunction with traction force microscopy, we measured forces exerted by cell monolayers prepared using Method 1. First, we patterned a 1 mm diameter circular island (Fig 6a) and quantified the traction magnitude over space (Fig 6b). Tractions were spatially heterogeneous, consistent with previously reported patterns of force generation in epithelial monolayers [37,47,48]. Also consistent with prior observations [5,12,32,37,49], tractions applied by cells to the substrate pointed radially inward at the edge of the island, indicating that epithelial cells have a tendency to pull themselves toward free space (Fig 6c). We also used monolayer stress microscopy to compute the in-plane components of the stress tensor. Here, we plot the maximum shear stress by where and are the first and second principal stresses, respectively. Similar to traction magnitude, the maximum shear stress was spatially heterogeneous (Fig 6d) [45]. We next applied the same approach to an elongated cell island, illustrating how this protocol can be used to examine force generation in more complex geometries (Fig 6e). The traction magnitude was heterogeneous in space. Because circular symmetry does not apply to the elongated island, we decomposed the tractions into components parallel and perpendicular to the long axis of the island to assess whether traction organization was aligned with geometric confinement. Similar to the circular pattern, cells at the edges of the elongated islands pulled inward on the substrate, meaning they pulled themselves towards the free space (Fig 6g). Finally, we computed the maximum shear stress of the elongated island, which was visually smoother over space than that of the circular island (Fig 6h). Additionally, these methods have been used previously to relate cell motion, force generation, and alignment in epithelial monolayers [45,47,50] These data demonstrate use of our protocol for experiments measuring cell-substrate tractions and monolayer stresses and suggest that cellular confinement geometry can influence the spatial organization of force production and transmission in epithelial monolayers.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Representative experiment showing effect of island shape on force production and transmission within the cell monolayer.

(a) Phase contrast image of 1 mm cell island. (b) Heat map of traction magnitude. (c) Heat map of the radial component of traction applied by the cells to the substrate, with outward being positive. (d) Heat map of maximum shear stress. (e) Phase contrast image of an elongated cell island. (f) Heat map of traction magnitude. (g) Heat map of traction components parallel and perpendicular to the long axis of the elongated island. Arrows indicate direction of positive traction. (h) Heat map of maximum shear stress.

https://doi.org/10.1371/journal.pone.0344657.g006