Controlling cell shape on hydrogels using lift-off protein patterning
Controlling cell shape on hydrogels using lift- off protein patterning
Jens Moeller 2 3 4
Aleksandra K. Denisin 1 2 3 4
Joo Yong Sim 2 3 4
Robin E. Wilson 2 3 4
Alexandre J. S. Ribeiro 2 3 4
Beth L. Pruitt 0 1 2 3 4 5
☯ These authors contributed equally to this work. 2 4
0 Stanford Cardiovascular Institute, Stanford University , Stanford, California , United States of America
1 Department of Bioengineering, Stanford University , Stanford, California , United States of America
2 Funding: This project was supported in part by the National Science Foundation (EFRI-MIKS 1136790, CMMI 1662431, ECCS-1542152), National Institutes of Health (R01EB006745, graduate and post-doctoral research fellowships from the National Science Foundation, ILJU Foundation, Stanford BioX, Stanford Office of the Vice Provost for Graduate Education , American
3 Department of Mechanical Engineering, Stanford University , Stanford, California , United States of America
4 Editor: Nic D. Leipzig, The University of Akron , UNITED STATES
5 Department of Molecular and Cellular Physiology, Stanford University School of Medicine , Stanford, California , United States of America
Polyacrylamide gels functionalized with extracellular matrix proteins are commonly used as cell culture platforms to evaluate the combined effects of extracellular matrix composition, cell geometry and substrate rigidity on cell physiology. For this purpose, protein transfer onto the surface of polyacrylamide hydrogels must result in geometrically well-resolved micropatterns with homogeneous protein distribution. Yet the outcomes of micropatterning methods have not been pairwise evaluated against these criteria. We report a high-fidelity photoresist lift-off patterning method to pattern ECM proteins on polyacrylamide hydrogels with elastic moduli ranging from 5 to 25 kPa. We directly compare the protein transfer efficiency and pattern geometrical accuracy of this protocol to the widely used microcontact printing method. Lift-off patterning achieves higher protein transfer efficiency, increases pattern accuracy, increases pattern yield, and reduces variability of these factors within arrays of patterns as it bypasses the drying and transfer steps of microcontact printing. We demonstrate that lift-off patterned hydrogels successfully control cell size and shape and enable long-term imaging of actin intracellular structure and lamellipodia dynamics when we culture epithelial cells on these substrates.
Data Availability Statement: All relevant data are
within the paper and its Supporting Information
Cell culture substrates patterned with extracellular matrix (ECM) are widely used to mimic the
spatial organization and rigidity of the in vivo cell microenvironment in vitro. These cell
culture platforms enable reductionist studies of the mechanobiology of healthy and diseased
tissues under physiological stiffness conditions [
]. Specifically, polyacrylamide (PAAm)
hydrogels are commonly used because these substrates can be functionalized with ECM
proteins and tuned in their mechanical properties to replicate different tissue stiffness ranging
from ~0.1 kPa to ~40 kPa [
]. Yet techniques for patterning proteins on PAAm have lacked
quantitative assessment, which is critical for developing and comparing protocols to reliably
Heart Association, and Stanford Graduate
Fellowship. The funders had no role in study
design, data collection and analysis, decision to
publish, or preparation of the manuscript.
restrict cells to user-defined shapes. The spatial resolution and accuracy of the protein patterns
will directly impact the cellular response which is of particular importance for
mechanobiological studies on the organization and force transduction within the actin cytoskeleton [
cellular adhesions [
Broadly, two main strategies exist to pattern ECM on PAAm gels (reviewed in [
selective activation of the gels for covalent attachment of proteins to activated regions (e.g. direct
surface functionalization using UV-reactive sulfo-SANPAH crosslinkers [
] or polymerizing
N-hydroxyacrylamide into the hydrogel surface [
]) and ii) co-polymerization of ECM
proteins into the gels during gelation through direct contact of the acrylamide precursor mix with
a protein-patterned coverslip. The first method, direct surface functionalization, uses
expensive functionalization reagents and also depends on reagent quality and reaction time as the
chemicals are unstable in aqueous media and in the presence of oxygen [
]. The method of
copolymerizing ECM proteins relies on patterning glass coverslips with protein and placing
them in direct contact with the hydrogel during polymerization. Although the detailed
molecular mechanism of protein incorporation into the polymerizing gel is unknown, this method
has successfully been applied to functionalize hydrogels with a variety of ECM proteins [7±9].
Protein patterns on glass coverslips are often created by microcontact printing (μCP) using
elastomeric `stamps' [
]. μCP involves casting polydimethylsiloxane (PDMS) on
microfabricated master structures created by photolithography to create stamps by replica molding [
Most groups use μCP since PDMS casting and contact printing protocols are straightforward
once the master structures on silicon wafers are made [
7, 12, 13
]. However, μCP relies on the
transfer of dried proteins from a deformable PDMS stamp and thus the accuracy, resolution,
and pattern design are limited and critically depend on the PDMS stamp preparation and
handling [14±16]. PDMS can also be micromachined to create a stencil which can be used to
selectively adsorb proteins to specific regions of the glass coverslip and keep proteins hydrated
throughout the process [
], but the fabrication of a high-resolution stencil requires
reactive ion etching which is not available to all laboratories  and stencil alignment and
conformal contact with the substrate is required for successful protein patterning. A technique
similar to μCP called ªstamp-offº allows for proteins to first be adsorbed on a surface and then
selectively being removed by a PDMS stamp placed in contact with the substrate [
stamp-off can enable patterning multiple proteins, it still suffers from the same lateral size
limitations as μCP. To improve the accuracy of μCP, alternative nanopatterning methods have
been developed for patterning proteins on substrates (reviewed in [
]). For example, dip-pen
nanolithography, AFM-based patterning, and nanografting enable direct writing of proteins
on flat, solid substrates with nanometer precision [
]. However, these nanopatterning
methods are serial and have not been used to directly functionalize hydrogels.
The challenge of serial patterning can be overcome using approaches to pattern protein
features over large areas in parallel using a molecular adlayer to define biopassive regions and
backfilling exposed regions with protein. Biopassive, `non-fouling' Poly(l-lysine)-graft-poly
(ethylene glycol) (PLL-g-PEG) copolymers can be used to control protein adsorption to
engineered substrates [
]. To pattern proteins on glass, PLL-g-PEG can then be selectively
oxidized by deep UV irradiation through a photomask or via projection lithography [
Such patterns can subsequently be transferred to a hydrogel  thereby decoupling pattern
generation from hydrogel functionalization. Those methods however require either access to a
collimated deep UV light source or a UV projection system, which are not readily available in
most laboratories. Further, the PLL-g-PEG layer must either be dried prior to the UV
irradiation, which requires a rehydration step prior to protein incubation [
], or a photoinitiator
must be added during the UV exposure step that has to be removed completely to re-establish
the biopassive properties of the adlayer [
]. Another method using PLL-g-PEG involves
2 / 17
creating a sacrificial mask on glass and uncovering the protected regions for later incubation
steps. In combination with reactive ion etching, Falconnet and colleagues nanoimprinted a
poly(methyl methacrylate) (PMMA) film on a glass coverslip, incubated the substrate with a
biotinylated PLL-g-PEG copolymer to facilitate the selective binding of avidin-functionalized
proteins, and lifted off the nanoimprinted film prior to backfill with biopassive PLL-g-PEG to
achieve 100 nm features [
]. Nanoimprinting and reactive ion etching can thus be used in
tandem with blocking of nonspecific protein adsorption using biopassive polymers, e.g.
PLLg-PEG, to produce nanometer-scale features on large areas but the process is complex and
requires microfabrication equipment not readily available in many laboratories.
In this work, we present a photoresist lift-off patterning (LOP) method to control the shape
of cells on PAAm hydrogels with high fidelity. Our method integrates advances in: i) contact
photolithography and photoresist lift-off widely used in the semiconductor and
microfabrication industry [
], ii) the molecular assembly and patterning of biopassive PLL-g-PEG
coatings on glass[29±31], and iii) the protein transfer from glass to PAAm hydrogels [
We create protein-patterned glass coverslips by photoresist lift-off-assisted patterning of
PLL-g-PEG and transfer the protein pattern to PAAm gel surfaces by co-polymerization. To
demonstrate the utility of this approach, we successfully controlled the shape of MDCK cells
cultured on patterned hydrogels and followed the cells' cytoskeletal and membrane dynamics.
We benchmark the LOP technique to the widely used μCP across a range of physiologically
relevant hydrogel stiffness (5 kPa, 10 kPa and 25 kPa) and analyze the pattern accuracy and
transfer efficiency from the glass to the PAAm gel. We find that the LOP protocol improves
both the pattern transfer efficiency and the pattern accuracy, thereby reducing the pattern
variability and increasing the predictability of the engineered in vitro cell culture models.
Materials and methods
Photoresist lift-off assisted patterning of ECM proteins (LOP)
ECM patterned glass coverslips were fabricated by photoresist lift-off (see process flow in Fig
1, full protocol in S1 Text). We cleaned coverslips with acetone, isopropanol, and water,
followed by thoroughly drying them on a hot plate. We then spin-coated S1818 photoresist
(Microchem) on coverslips using standard contact photolithography and photopatterned the
2μm thick resist layer (40±50 mJ/cm2 at 365 nm, OAI Instruments) (Fig 1A and 1B). Following
plasma activation, we incubated the S1818 patterned glass coverslips with 0.1 mg/ml
(poly(llysine)-graft-poly(ethylene glycol) (PLL(20)-g[3.5]-PEG(2), SuSoS AG) in PBS (pH 7.4) for
one hour to allow for self-assembly of the densely packed, biopassive PLL-g-PEG polymeric
brush adlayer on the exposed surface areas of the glass substrate. The PLL(20)-g[3.5]-PEG(2)
copolymer comprises of linear PEG chains (Mw = 2 kDa) grafted to a PLL backbone (Mw = 20
kDa) at a grafting ratio g = 3.5 . Non-adsorbed PLL-g-PEG was removed by washing in
PBS (pH 7.4). After photoresist lift-off in 1-methyl-2-pyrrolidone (NMP, Sigma 328634), we
backfilled the PLL-g-PEG patterns with 100 μg/ml of Oregon Green-488 or Alexa Fluor
568-labeled gelatin solution in PBS pH 7.4 for 1 hour in the dark (Thermo Scientific, G13186,
A10238) (Fig 1C±1E). The slides were washed thoroughly with DI water and excess liquid was
removed by blotting on filter paper immediately prior to gel transfer. We chose gelatin,
hydrolyzed collagen I, as a model ECM protein to mimic the epithelial basement membrane because
the Arg-Gly-Asp (RGD) sequence that is critical for cell adhesion, migration and proliferation
is preserved [
]. Gelatin, in contrast to collagen I, is available commercially with fluorescent
labels or can be functionalized with standard protein labeling kits.
3 / 17
Fig 1. LOP fabrication of protein patterns on polyacrylamide gels. (A,B) Photoresist patterns are
fabricated by standard contact photolithography on glass coverslips. Inset at right shows array of S1818
photoresist features after development. (C) Unspecific protein adhesion to the resist-patterned coverslip is
blocked by incubating with biopassive PLL(20)-g[3.5]-PEG(2) copolymer. (D,E) Following photoresist lift-off,
the resulting PLL-g-PEG pattern is backfilled with the ECM protein of interest. Inset at right shows a
fluorescence micrograph of labeled gelatin on glass after backfill. (F) To transfer the protein pattern to the
PAAm gel, the gel is polymerized between the protein patterned glass coverslip and a silanized coverslip. (G)
After gel polymerization, the top coverslip is removed from the PAAm gel. Inset at right shows a fluorescence
micrograph of a labeled protein transferred to a PAAm gel. (H) Inset at right shows pairs of epithelial cells on
the patterned PAAm gel restricting the geometry of the protein functionalized regions.
Microcontact printing of ECM proteins (μCP)
We prepared PDMS stamps by casting Sylgard 184 PDMS (10:1 base to curing agent, Dow
Corning) in a 9 μm deep mold microfabricated by standard photolithography using SU-8
4 / 17
negative resist [
]. We incubated the PDMS stamps (1 cm2 squared stamps with 45 μm2
patterns with 80 μm spacing) with 100 μg/ml fluorescently labeled gelatin solution for one hour in
the dark. Following protein incubation, we aspirated excess protein solution and dried the
stamps gently using low nitrogen gas flow. Prior to μCP, we cleaned glass coverslips with 2%
v/v Hellmanex solution (Hellma Analytics) in DI water for at least 30 minutes. We then rinsed
the coverslips 5 times with DI water and dried them with compressed air prior to stamping.
We put the PDMS stamps in contact with the cleaned coverslips for 5 minutes and removed
the stamps by carefully forcing a tweezer between the coverslip and the edge of the stamp.
Preparation of ECM patterned polyacrylamide gels
We transferred the protein patterns from the glass coverslip to the surface of PAAm gel for
both the LOP and μCP protocols by co-polymerization (Fig 1F±1H). Polyacrylamide gels of
varying stiffness were polymerized between the protein patterned glass coverslip and a
silanized bottom coverslip. The bottom coverslip was silanized to ensure covalent bonding of gels
to this bottom glass layer, following a method by Guo and colleagues [
]. Briefly, 30 μL of
working solution (3 μl bind-silane (3-methacryloxypropyl-trimethoxysilane, Sigma-Aldrich,
M6514), 950 μL 95% ethanol, and 50 μL of glacial acetic acid) were applied to the coverslip,
allowed to incubate for 5 min, and then rinsed with ethanol and dried in a desiccator.
Polyacrylamide gels of three different stiffness were used for experiments: 5 kPa, 10 kPa,
and 25 kPa as determined by Tse and Engler [
]. MilliQ water, acrylamide (0.5 g/mL stock,
Sigma-Aldrich, 01696 FLUKA, 71.08g/mol), and bis-acrylamide (0.025 g/mL stock,
SigmaAldrich, 146072, 154.17 g/mol) were combined to yield 5% w/v acrylamide and 0.15% w/v
bisacrylamide for 5 kPa gels, 10% w/v acrylamide and 0.1% w/v bis-acrylamide for 10 kPa gels,
and 10% w/v acrylamide and 0.25% w/v bis-acrylamide for 25 kPa gels. The precursor solution
was degassed in a vacuum desiccator for 1 hr. To initiate gelation, 5 μL of 10% w/v ammonium
persulfate (APS, Sigma-Aldrich, A9164) was added to ~995 μL of gel precursor solution
followed by 0.5 μL of N,N,N0,N0-Tetramethylethylenediamine accelerator (TEMED,
SigmaAldrich, 411019). We mixed the solutions by gentle pipetting, dispersed 50 μL of the solution
on the activated coverslip, and then placed the protein-functionalized coverslip on top,
creating a sandwich (Fig 1F). Gels were left undisturbed at room temperature for 30 minutes to
polymerize. After polymerization, the gels were immersed in PBS for at least 1 hour and the
glass coverslip was removed from the top of the gels (Fig 1G).
Analysis of pattern transfer efficiency
To assess the protein transfer efficiency from the patterned glass coverslip onto the PAAm gel,
we imaged the coverslips before transfer and compared it to the resulting patterns on the
PAAm gel surface after gelation and coverslip removal using the same microscope image
acquisition parameters (1 second exposure, images of 1 series acquired at the same day to
avoid variability in lamp power). Prior to imaging, we avoided photobleaching by keeping
samples in the dark. Our patterns were arranged in labelled arrays so we decreased
photobleaching by exposing each area to light only when focusing the image and during image
capture. All acquired images were processed by ImageJ software (http://rsb.info.nih.gov/ij/). We
analyzed 150 individual patterned features by measuring the difference between the same
feature on the coverslip before and after transfer, using the cvMatch_Template ImageJ plugin
]. The average background signal was determined outside the protein pattern and
subtracted for each image. We measured the average pixel intensity within a region of interest
defined as our theoretical patterning shape and calculated the transfer efficiency as the average
5 / 17
intensity of the protein pattern on the gel image divided by the average intensity of the pattern
on the coverslip.
Analysis of the geometric accuracy of protein patterning
To compare the accuracy of patterns generated by LOP and μCP, we calculated the
cross-correlation coefficient between the theoretical pattern shape and the binarized patterned features
using the corr2 function in Matlab (R2014b, Mathworks). The binarized stacks (n = 150
patterns) were created with ImageJ by de-noising the images using the built-in despeckle function
followed by automated binarization of each pattern using Otsu thresholding. Profile column
average plots were analyzed from the binarized pattern stacks using ImageJ. To perform yield
analysis, we selected around 389±416 features for each gel stiffness type, created a montage,
and then used cross correlation with a threshold of 0.84 to find acceptable features. We divided
the number of acceptable features by the total number of features analyzed for each gel stiffness
to calculate the yield.
Analysis of surface energy using water contact angle
We used a contact angle goniometer (Rame-Hart 290) to measure the hydrophilicity of
substrates used in LOP and μCP. We dispensed 4 μl of deionized water on the surface, equilibrated
for 1 minute at room temperature before taking a photograph of the water contact angle to
standardize between measurements and ensure equal evaporation of the liquid. We evaluated
the water contact angle of PLL-g-PEG coated glass coverslips before and after polymerization
of 25 kPa PAAm gels. Polymerized gels were incubated with PBS overnight at 4ÊC before
dissociating the top coverslip and performing the water contact angle measurements. We
calculated the contact angle using the DropShape ImageJ plugin [
]. Water contact angle
measurements are a direct read-out of the surface energy and thus can provide insights of the
conformation states of the proteins during both patterning processes.
Madin-Darby Canine Kidney (MDCK) type II G cells were transfected with LifeAct-GFP
(ibidi, 60101) using the Amaxa Biosystem Nucleofector II system and transfection kit (Lonza,
VCA-1005). The LifeAct-GFP MDCK cells were maintained in low glucose DMEM
(Invitrogen, 11885) containing 1 g/l sodium bicarbonate, 1% Penicillin-Streptomycin (PenStrep,
ThermoFisher, 15140122), 0.5 mg/ml G418 selection reagent (Sigma-Aldrich, G418-RO Roche),
and supplemented with 10% (vol/vol) fetal bovine serum (FBS). 25 kPa PAAm gels patterned
with 100 μg/ml collagen I (Gibco, A1048301) mixed with 20 μg/ml Alexa Fluor 568 labeled
gelatin were cast into Mattek dishes (14 mm glass, Mattek P35G-0.170-14-C). MDCK cells were
trypsinized and seeded on the PAAm gels for 16 hours before imaging experiments. Prior to
imaging, the media was replaced to low glucose DMEM with no phenol red (ThermoFisher,
11054001) and supplemented with 1% PenStrep, 10% FBS, and 25 mM HEPES buffer. Cells
were imaged on a Leica DMI6000B microscope with heated incubation unit at 5 minute
intervals using a 40x air objective, NA = 0.6.
We compare the LOP and μCP methods by analyzing the efficiency of protein transfer from
the surface of coverslips onto the surface of PAAm gels (Fig 2) and the geometrical accuracy of
the created patterns (Fig 3). We use a square `frame' pattern shape to compare how both
protocols resolve corners and edges of a complex shape. We show pattern arrays of glass and PAAm
6 / 17
Fig 2. Quantification of protein transfer efficiency to PAAm gels of varying stiffness. (A,B) Arrays of 45 μm2 square protein
patterns on 25 kPa PAAm gels created by LOP and μCP before and after transfer to gel surface. (C) Quantification of protein
transfer efficiency from glass coverslips to PAAm gel of varying stiffness. Differences between LOP and μCP for each stiffness are
statistically significant (p-value < 2.2E-16, Mann-Whitney-Wilcoxon test). Substantially more protein is transferred from patterns
created by photoresist lift-off. Data are represented as box plots. The median, 1st and 3rd quartile, and minimum and maximum
values are shown, n = 150 for each method and stiffness shown. (D) Overview of μCP method to pattern proteins on PAAm gels.
samples normalized for contrast to aid visual comparison of the transfer efficiency for LOP
and μCP techniques in Fig 2A and 2B.
Protein patterns created by the LOP method are transferred more efficiently from the
coverslips to gels for all gel stiffness we tested (Fig 2C). We find significant differences in
transfer efficiency between LOP and μCP when comparing both protocols at each stiffness
(p-value < 2.2E-16 using the Mann-Whitney-Wilcoxon to compare the 5 kPa, 10 kPa, and
25 kPa PAAm gel samples). However, the protein transfer efficiency in both methods is
considerably lower for 5 kPa when compared to 10 kPa and 25 kPa PAAm gels. To explain this
observation, we analyzed and compared the size properties of gelatin and polyacrylamide gel
formulations used in this study to those commonly used in mechanobiology and
electrophoresis (see S1 Text). The 10 kPa and 25 kPa gel formulations we use contain 10% total polymer
which is twice that of the 5 kPa gel formulation (5%). Due to the lower total polymer content,
we hypothesize that fewer sites are available for protein integration during polymerization in
the 5 kPa gels. This effect is independent of the patterning technique and thus can be a limiting
factor for the protein functionalization of soft polyacrylamide gels.
To evaluate the accuracy of the features transferred to PAAm gels, we compare the corners
of the square frame patterns for both protocols (Fig 3A) and show the difference between the
actual and theoretical shape (Fig 3B). Protein patterns created by LOP exhibit greater
definition in the pattern edges and corners than protein patterns created by μCP. Cross-correlation
analysis of the patterns compared to the theoretical pattern shape on the photomask shows
7 / 17
Fig 3. Comparison of pattern accuracy between LOP and μCP methods. (A) Average images of 150
binarized protein patterns created by LOP and μCP on 25 kPa gels. (B) Difference images calculated by
comparing the average images and the theoretical pattern mask. Edges and corners are resolved
substantially better in patterns created by LOP. (C) Theoretical pattern shape with a region highlighted
corresponding to where profile column average scans were taken. (D) Profile column average scans across
150 binarized patterns show that the variation in protein signal at the pattern edges is strongly reduced in LOP
patterns. Plotted are the median (line), 1st / 3rd quartile (box) and 5±95% (whisker) of the probability of protein
present across the pattern width.
that LOP results in patterns that more accurately recapitulate the theoretical shape. Correlation
coefficients are as follows (n = 150 patterns; mean +/- standard deviation): μCP (5, 10, 25 kPa):
0.84±0.05; 0.87±0.02; 0.89±0.02; LOP (5, 10, 25 kPa): 0.91±0.04; 0.94±0.02; 0.93±0.01). The
higher fidelity of the pattern edges becomes evident when we compare profile scans across the
average of 150 patterns for both methods to the theoretical pattern shape (Fig 3C and 3D;
similar to methods by Vignaud and colleagues [
]). The variation in the protein signal at the
pattern edges is strongly reduced in the LOP patterns. These results are also supported by a
crosscorrelation analysis where we tested the variability and yield of acceptable features across
about 400 total feature samples for each gel formulation. We applied a correlation coefficient
threshold for acceptable features of 0.84 to match the lowest correlation coefficient in our
analysis above. LOP resulted in a greater number of acceptable features than μCP and acceptable
feature yield varied from 59% to 98% for LOP and from 4% to 72% for μCP for different gel
formulations (see S2 Fig and S2 Table for summary of data).
We demonstrate that single cells as well as pairs of cells attach exclusively to the ECM
patterned areas of PAAm gels patterned using the LOP method (Fig 4A and 4B; S1 and S2
Movies). Areas between patterns exhibit anti-adhesive properties and prevent cells from binding
outside the protein features. To test if the cytoskeletal architecture and remodeling are
8 / 17
Fig 4. LOP yields sharper cell edges with localized actin bundles compared to μCP patterned gels. Time-lapse acquisitions of
MDCK cells transfected with Lifeact-GFP (actin label) grown on 25 kPa PAAm gels showed similar intracellular actin structures on LOP (A,
B) and μCP (C,D) protein patterns. Cell doublets rotated around each other on the patterns for both techniques (B,D).
different for cells attached to patterns created by LOP or μCP, we followed the actin dynamics
of LifeAct-GFP transfected MDCK cells using live cell fluorescence microscopy on 25 kPa
substrates. We chose to conduct our analysis on gels with a 25 kPa elastic modulus because this
value is close to the measured stiffness of a MDCK monolayer by micro-indentation: 33 ± 3
]. Consistent with previous literature reports [
], we found that cell doublets on the
frame patterns rotated around each other (Fig 4B and 4D; S2±S4 Movies). While this was
observed independent of the patterning method used, the cell edges were more clearly defined
for cells adhering to patterns created by LOP than gels patterned by μCP (Fig 5), which was
consistent with the higher pattern accuracy (Fig 3).
In this work, we introduce a photoresist-based LOP technique to pattern ECM proteins on
polyacrylamide hydrogels to control the shape of cells with high-fidelity and compare it with
the widely used μCP protocol. We found the LOP method to be more efficient and accurate in
reproducing complex micrometer-sized patterns (Figs 2 and 3). To illustrate that the improved
fidelity of LOP patterns translates to greater control over cell shape, we cultured MDCK
epithelial cells on patterned 25 kPa gels for up to 16 hours. The shape of single cells and cell pairs
on LOP pattern reflected the theoretical shape with greater accuracy as compared to cells
on μCP patterns (Figs 4 and 5; S1±S4 Movies).
The difference between μCP and LOP protein transfer efficiency is most likely arising from
how the two methods change substrate surface energy to facilitate protein adsorption. Protein
transfer in μCP is based on surface energy differences between the PDMS stamp and the glass
]. PDMS is hydrophobic with low free energy (water contact angle 91±111Ê) while
plasma-treated or Hellmanex-cleaned glass is a high-energy hydrophilic surface (contact angle
9 / 17
Fig 5. Lamellipodia are more exploratory for cells on LOP than on μCP patterned substrates. Phase contrast time lapse imaging of 3
representative MDCK cells on 25 kPa PAAm gels patterned by LOP (A) and μCP (B). Cells on substrates produced by LOP follow the protein
pattern border more accurately (dotted line panels A, B) and reveal more pronounced lamellipodia. (C, D) Kymograph analysis of lamellipodia
kinetics along pattern edge. Single cell on LOP pattern shows increased lamellipodia protrusions and retractions within a 10 μm wide region of
interest outside the protein pattern edge compared to a cell on a μCP pattern (S5 and S6 Movies). Kymographs show cells depicted in bottom
row of panel A and B. Regions of interest are straightened and distorted regions at the pattern corners are cleared.
of 0Ê) (S3 Fig). The hydrophobicity of PDMS causes proteins to denature and lose their
conformation thereby decreasing the μCP protein transfer to the glass and to the polyacrylamide gel.
In contrast, LOP does not depend on a surface energy gradient and none of the substrates
involved in LOP are hydrophobic. We found that the glass had hydrophilic properties at all
stages of the LOP protocol with water contact angle of 29±36Ê (see S3 Fig).
Another difference between the protocols which can lead to differences in protein transfer
to the PAAm gels is the drying of the PDMS stamp after protein incubation in the μCP
protocol. Drying the protein-inked stamp is essential for μCP to be successful to maintain the
gradient in surface energy to increase affinity of the adsorbed protein for the hydrophilic glass and
to ensure accurate patterning without blurring by diffusion [
]. Yet, drying the protein leads
to changes in protein conformation, causing some protein to irreversibly adsorb to the
]. Experiments which increased the hydrophilicity of PDMS stamps found that the
quality of μCP protein transfer decreased due to the presence of polar functional groups
attracting the protein to the PDMS rather than the desired glass substrate [
The differences in pattern fidelity on the hydrogels stem from methodological differences
in the patterning of the glass coverslips. LOP relies on the direct molecular assembly of the
biopassive PLL-g-PEG copolymer on the S1818 photoresist-patterned glass substrates while μCP
involves PDMS replica molding from SU-8 master structures and stamping of the protein to
the coverslip. The ideal spatial resolution that is achievable by contact photolithography to
prepare both the S1818 patterns for LOP and the SU-8 master for μCP can be estimated using the
following relation between the ideal spatial resolution (R), exposure wavelength (λ) and the
10 / 17
photoresist thickness (z) [
Thus, differences in resist thickness substantially contributes to a decrease in pattern
accuracy. We use a 2 μm positive S1818 resist layer for LOP and a 9 μm negative SU-8 resist layer
for the fabrication of the PDMS μCP master structures because the resists serve different
purposes for each method. The thin positive resist in LOP can be removed by NMP lift-off
while keeping the adsorbed PLL-g-PEG as a patterned, biopassive adlayer on the glass
coverslip. 9 μm thick negative resist was chosen to yield PDMS feature heights that comply
with μCP design rules by Qin and colleagues [
] who suggest that the aspect ratio of a 10:1
Sylgard 184 PDMS stamp must be 0.5 < H/L< 5 and H/D > 0.05, where H is the height of
PDMS features, L is the critical feature dimension (line width L = 9 μm for 45μm2 frame
pattern used in this study), and D is the longest distance between features (D = 120 μm, diagonal
distance between frame pattern corners). Those design rules avoid lateral pairing and buckling
of stamp features, as well as stamp roof collapse to yield successful protein transfer. In contrast
to μCP, LOP enables the design and fabrication of arbitrary pattern geometry and spatial
organization (e.g. large pattern-to-pattern distance) as it circumvents the PDMS stamping.
Additional sources for error in μCP can arise from non-uniform contact of the stamp with the glass
In addition to the resist layer thickness, the type of resist used in both protocols further
contributes to pattern fidelity. Positive resist has a higher contrast (γ = 2.2) as compared to
negative resist (γ = 1.5) [
]. This difference contributes to positive resist usually yielding higher
resolution features with less distortion than negative resist. The negative SU8 resist used
in μCP to create the PDMS mold is also typically under-developed and thus resolving the
edges of features is a challenge (see S1 Fig) which explains why μCP results in features which
are smaller than the theoretical specifications (see Fig 3A). LOP uses the positive S1818 resist
which tends to be over-developed and thus features fabricated by this method tend to be larger
than theoretical specifications but with better-resolved corners (see Fig 3A).
We found the lamellipodia of MDCK cells to be more exploratory and dynamic on LOP
than μCP substrates, extending up to 5 μm outside of the protein pattern (Fig 5). Epithelial
cells have been shown to extend lamellipodia for several micrometers past areas with ECM
during wound healing and cell migration [
]. Our observations of lamellipodia extending
up to 5 μm beyond adhesive regions and the pronounced actin bundles at the pattern edges
match well with the spatio-temporal organization of the cytoskeleton and focal adhesions at
the leading edge of migrating cells . Recently, it was shown that epithelial cells migrating
from clusters respond to geometrical constraints by altering their speed and that the
acto-myosin contractile ring structures of leader cells differ at near sharp corners to help polarize and
guide the direction of migration [
]. We were interested to note the dynamic way cells
explored LOP patterns due to the increased pattern fidelity on the pattern edges and sharp
corners. We thus expect the LOP method to enable future studies on the role of ECM organization
on cell migration and lamellipodia extension during embryonic development and cancer
An open question for patterns created by LOP is whether the PLL-g-PEG blocking agent on
the glass coverslips is transferred to the PAAm gel. In control experiments, we used
TRITClabeled PLL-g-PEG and we were not able to trace fluorescently labeled PLL-g-PEG transferring
to the gels nor did we record any loss of TRITC-labeled PLL-g-PEG on the glass coverslip after
using it for gel polymerization (see S4 Fig). Additionally, the water contact angle for
PLL-g11 / 17
PEG coated glass before and after gel polymerization remained constant. This data strongly
suggests that the PLL-g-PEG does not transfer from the glass to the gel during gel
polymerization. Yet, it remains open if any interactions of PLL-g-PEG copolymer and polyacrylamide
occur during gel polymerization. Further molecular level studies are needed to directly test
this hypothesis and are outside the scope of this work. Regardless of PLL-g-PEG transfer to the
PAAm surface, LOP results in functionalized PAAm gels with non-adhesive regions between
protein patterns. We noted that removing the glass coverslips from polymerized gels was easier
for samples created by LOP than for μCP (Fig 1G) and we hypothesize that this effect is due to
the high water content of the PLL-g-PEG adlayer on the coverslips [
] (see also contact angle
data in S3 and S4 Figs).
In summary, our LOP method facilitates advanced cell culture techniques that require precise
patterning of single or multiple cells into shapes of arbitrary geometry on PAAm hydrogel
substrates of varied stiffness. High pattern accuracy and defined ECM density within the protein
patterns are essential to compare cell phenotypes on different patterns and reduce the
systematic error of pooled measurements. This is of particular importance for studies focusing on
complex, multivariate cell-ECM signaling pathways and the cytoskeletal response to different
cell geometries and substrate stiffness [
]. Overall, local ECM density, cell shape, and
substrate stiffness have been shown to regulate the structural organization of focal adhesion
], the force balance between cell-cell and cell-ECM adhesions , the nuclear
], mesenchymal stem cell stiffness [
], stem cell fate [
], leader cells during
collective migration , and the contractile properties of cardiomyocytes [
S1 Table. Polyacrylamide gel formulations used in this study.
S2 Table. Acceptable feature yield results.
S1 Text. File with detailed protocols for the microcontact printing and lift-off methods.
We also include comparisons between polyacrylamide pore size and gelatin protein molecular
size to aid discussion of protein transfer efficiency.
S1 Fig. Overview of the SU8 master and PDMS stamps used for microcontact printing. The
photolithography mold (A) and PDMS stamp cast from this mold (B) show rounded corners
where the edges of the pattern meet, both in the inner and outer regions of the pattern. The
height of the SU8 mold (~9 μm) may be limiting the pattern accuracy achievable with
S2 Fig. LOP results in higher yield of acceptable features than μCP. By setting a threshold of
a 0.84 correlation coefficient, the LOP protocol resulted in more acceptable features than μCP
(highlighted in green). We selected 389±416 features for each gel sample and then performed
cross correlation analysis on the collected feature montage. Acceptable feature yield varied
from 59% to 98% for LOP and from 4% to 72% for μCP for different gel formulations. See
S2 Table for summary of data.
12 / 17
S3 Fig. μCP depends on surface energy differences while substrates used for LOP have
similar surface energies. The water contact angle of substrates used in μCP differs substantially
from average of 111Ê for PDMS (n = 12 measurements) to approximately 0Ê for
Hellmanexcleaned glass (the substrate used for μCP). The Hellmanex treated glass sample was super
hydrophilic making an exact measurement of the low water contact angle difficult. Untreated
glass is shown as comparison with an average water contact angle 75Ê (n = 8 measurements).
The substrates used for LOP varied little in water contact angle. The ªUV-exposedº sample
corresponds to glass cleaned with acetone-isopropanol-water, coated with S1818 resist,
floodexposed to UV, developed, and processed with NMP for lift-off. In the LOP protocol, areas
that adsorb the PLL-g-PEG adlayer have been treated with the same procedure. The ªmaskedº
sample corresponds to glass cleaned with acetone-isopropanol-water, coated with S1818 resist,
no UV exposure, developed, and processed with NMP for lift-off. This substrate thus replicates
the surface areas that adsorb protein in the LOP protocol. See insets from our LOP protocol
and mask design for clarification. We recorded average water contact angles of 36Ê for glass
cleaned in a series of acetone-isopropanol-water (n = 48 measurements), 34Ê for ªUV exposedº
samples (n = 38 measurements), and 29Ê for ªmaskedº samples (n = 12 measurements).
For μCP, protein must be transferred from the hydrophobic PDMS to the hydrophilic
Hellmanex-cleaned glass. For LOP, protein would be adsorbed to the areas masked by S1818 after
those areas are exposed by lift-off and we found these areas to be hydrophilic. Insets show
examples of water droplets on the corresponding substrates.
S4 Fig. PLL-g-PEG remains on the glass slide after gel polymerization due to similar water
contact angle before and after gel polymerization and using a fluorescent PLL-g-PEG. A.)
We measured the contact angle of PLL-g-PEG coated glass before and after polymerizing a
polyacrylamide gel. The average water contact angle is similar with 27Ê for PLL-g-PEG glass
(n = 46 measurements) and 23Ê for PLL-g-PEG glass after gel polymerization (n = 42
measurements). B.) We also used TRITC-labeled PLL-g-PEG on the LOP patterned glass and measured
the intensity of the fluorescent signal before and after gel polymerization on the same
coverslip. We show a representative image showing the PLL-g-PEG-TRITC signal outside of the
protein features (dark frames in image). We subtracted the signal within the protein pattern
areas and divided the average PLL-g-PEG-TRITC signal `after' gel polymerization by the
`before' signal. Within the limits of the measurement, no loss in PLL-g-PEG-TRITC intensity
on the glass coverslip was observed (average 98% ± 2.6% of the initial signal remains on the
glass after gel polymerization, n = 80 regions analyzed). We were also unable to detect
PLL-gPEG on the surface of the resulting polyacrylamide gels. Together, our water contact angle and
fluorescence imaging data strongly suggest that PLL-g-PEG is not transferred to the PAAm gel
S1 Movie. Single MDCK on LOP gel. Three separate time-lapse acquisitions (5 minute
increments, time shown at upper left) of single MDCK cells on LOP-functionalized 25 kPa PAAm
gels. Three channels are shown (gelatin for protein patterning, phase for cell outline, and
LifeAct-GFP for actin structures). Scale bar is 45 μm wide.
S2 Movie. Doublet MDCK cell pairs on LOP gel. Three separate time-lapse acquisitions (5
minute increments, time shown at upper left) doublet MDCK cell pairs on LOP-functionalized
25 kPa PAAm gels. Three channels are shown (gelatin for protein patterning, phase for cell
13 / 17
outline, and LifeAct-GFP for actin structures). Scale bar is 45 μm wide.
S3 Movie. Single MDCK on μCP gel. Three separate time-lapse acquisitions (5 minute
increments, time shown at upper left) of single MDCK cells on μCP-functionalized 25 kPa PAAm
gels. Three channels are shown (gelatin for protein patterning, phase for cell outline, and
LifeAct-GFP for actin structures). Scale bar is 45 μm wide.
S4 Movie. Doublet MDCK cell pairs on μCP gel. Three separate time-lapse acquisitions (5
minute increments, time shown at upper left) doublet MDCK cell pairs on μCP-functionalized
25 kPa PAAm gels. Three channels are shown (gelatin for protein patterning, phase for cell
outline, and LifeAct-GFP for actin structures). Scale bar is 45 μm wide.
S5 Movie. Lamellipodia beyond pattern edge for a cell on LOP gel. A time-lapse acquisition
(5 minute increments, time shown at upper left) of an MDCK cell lamellipodia extending past
the protein pattern region created by LOP. The image is a `straightened' path along the cell
edge (on bottom) extending 10 μm past the protein pattern (away from the cell). Scale bar is
10 μm wide.
S6 Movie. Lamellipodia beyond pattern edge for a cell on μCP gel. A time-lapse acquisition
(5 minute increments, time shown at upper left) of an MDCK cell lamellipodia extending past
the protein pattern region created by μCP. The image is a `straightened' path along the cell
edge (on bottom) extending 10 μm past the protein pattern (away from the cell). Scale bar is
10 μm wide.
The authors thank Pruitt lab members for helpful discussion of results and Dr. Jeffrey Tok
from the Stanford Soft and Hybrid Materials Facility for his advice on characterizing substrates
using the contact angle goniometer.
Conceptualization: Jens Moeller, Aleksandra K. Denisin, Alexandre J. S. Ribeiro, Beth L.
Data curation: Jens Moeller, Aleksandra K. Denisin.
Formal analysis: Jens Moeller, Aleksandra K. Denisin.
Funding acquisition: Beth L. Pruitt.
Investigation: Jens Moeller, Aleksandra K. Denisin, Joo Yong Sim, Robin E. Wilson,
Alexandre J. S. Ribeiro.
Methodology: Jens Moeller, Aleksandra K. Denisin, Joo Yong Sim, Robin E. Wilson,
Alexandre J. S. Ribeiro.
Project administration: Beth L. Pruitt.
14 / 17
Resources: Beth L. Pruitt.
Supervision: Beth L. Pruitt.
Visualization: Jens Moeller, Aleksandra K. Denisin.
Writing ± original draft: Jens Moeller, Aleksandra K. Denisin, Beth L. Pruitt.
Writing ± review & editing: Jens Moeller, Aleksandra K. Denisin, Joo Yong Sim, Robin E.
Wilson, Alexandre J. S. Ribeiro, Beth L. Pruitt.
15 / 17
16 / 17
1. TheÂry M. Micropatterning as a tool to decipher cell morphogenesis and functions . J Cell Sci . 2010 ; 123 (Pt 24): 4201 ± 13 . https://doi.org/10.1242/jcs.075150 PMID: 21123618 .
2. Tse JR , Engler AJ . Preparation of hydrogel substrates with tunable mechanical properties . Curr Protoc Cell Biol . 2010 ; 47 ( 10 ): Unit 10.6 . https://doi.org/10.1002/0471143030.cb1016s47 PMID: 20521229 .
3. Kassianidou E , Kumar S. A biomechanical perspective on stress fiber structure and function . Biochim Biophys Acta . 2015 ; 1853 (11 Pt B): 3065 ± 74 . https://doi.org/10.1016/j.bbamcr. 2015 . 04 .006 PMID: 25896524 .
4. Sim JY , Moeller J , Hart KC , Ramallo D , Vogel V , Dunn AR , et al. Spatial Distribution of Cell-Cell and Cell-ECM Adhesions Regulates Force Balance while Maintaining E-cadherin Molecular Tension in Cell Pairs . Mol Biol Cell . 2015 ; 26 ( 13 ): 2456 ± 65 . https://doi.org/10.1091/mbc.E14-12-1618 PMID: 25971797 .
5. Ribeiro AJS , Denisin AK , Wilson RE , Pruitt BL . For whom the cells pull: Hydrogel and micropost devices for measuring traction forces . Methods . 2016 ; 94 : 51 ± 64 . https://doi.org/10.1016/j.ymeth. 2015 . 08 .005 PMID: 26265073 .
6. Grevesse T , Versaevel M , Circelli G , Desprez S , Gabriele S. A simple route to functionalize polyacrylamide hydrogels for the independent tuning of mechanotransduction cues . Lab Chip . 2013 ; 13 ( 5 ): 777 ± 80 . https://doi.org/10.1039/c2lc41168g PMID: 23334710 .
7. Rape AD , Guo W-H , Wang Y -l. The regulation of traction force in relation to cell shape and focal adhesions . Biomaterials . 2011 ; 32 ( 8 ): 2043 ±51. https://doi.org/10.1016/j.biomaterials. 2010 . 11 .044 PMID: 21163521 .
8. Polio SR , Parameswaran H , Canović EP , Gaut CM , Aksyonova D , Stamenović D , et al. Topographical control of multiple cell adhesion molecules for traction force microscopy . Integrative Biology . 2014 ; 6 ( 3 ): 357 ± 65 . https://doi.org/10.1039/c3ib40127h PMID: 24441735 .
9. Vignaud T , Ennomani H , TheÂry M. Polyacrylamide Hydrogel Micropatterning . Methods Cell Biol . 120 : Elsevier; 2014 . p. 93 ± 116 . https://doi.org/10.1016/B978-0 -12-417136-7 . 00006 -9 PMID: 24484659
10. Alom Ruiz S , Chen CS . Microcontact printing: A tool to pattern . Soft Matter . 2007 ; 3 ( 2 ): 168 ± 77 . https:// doi.org/10.1039/B613349E
11. Qin D , Xia Y , Whitesides GM . Soft lithography for micro- and nanoscale patterning . Nat Protoc . 2010 ; 5 ( 3 ): 491 ± 502 . https://doi.org/10.1038/nprot. 2009 .234 PMID: 20203666 .
12. Lee J , Abdeen AA , Zhang D , Kilian KA . Directing stem cell fate on hydrogel substrates by controlling cell geometry, matrix mechanics and adhesion ligand composition . Biomaterials . 2013 ; 34 ( 33 ): 8140 ±8. https://doi.org/10.1016/j.biomaterials. 2013 . 07 .074 PMID: 23932245 .
13. Tang X , Ali MY , Saif MTA . A Novel Technique for Micro-patterning Proteins and Cells on Polyacrylamide Gels . Soft Matter . 2012 ; 8 ( 27 ): 7197 ± 206 . https://doi.org/10.1039/C2SM25533B PMID: 23002394 .
14. Hui CY , Jagota A , Lin YY , Kramer EJ . Constraints on microcontact printing imposed by stamp deformation . Langmuir . 2002 ; 18 : 1394 ± 407 . https://doi.org/10.1021/la0113567
15. Huang YY , Zhou W , Hsia KJ , Menard E , Park J-U , Rogers JA , et al. Stamp Collapse in Soft Lithography. Langmuir . 2005 ; 21 ( 17 ): 8058 ± 68 . https://doi.org/10.1021/la0502185 PMID: 16089420
16. Delamarche E , Schmid H , Michel B , Biebuyck HA . Stability of molded polydimethylsiloxane microstructures . Advanced Materials . 1997 ; 9 : 741 ± 6 .
17. Park TH , Shuler ML . Integration of cell culture and microfabrication technology . Biotechnol Prog . 2003 ; 19 ( 2 ): 243 ± 53 . https://doi.org/10.1021/bp020143k PMID: 12675556 .
18. Masters T , Engl W , Weng ZL , Arasi B , Gauthier N , Viasnoff V . Easy Fabrication of Thin Membranes with Through Holes . Application to Protein Patterning. PLoS One . 2012 ; 7(8):e44261±7 . https://doi.org/ 10.1371/journal.pone. 0044261 PMID: 22952944
19. Chen W , Lam RHW , Fu J . Photolithographic surface micromachining of polydimethylsiloxane (PDMS) . Lab Chip . 2012 ; 12 ( 2 ): 391 ±5. https://doi.org/10.1039/c1lc20721k PMID: 22089984 .
20. Desai RA , Rodriguez NM , Chen CS . "Stamp-off" to micropattern sparse, multicomponent features . Methods Cell Biol . 2014 ; 119 :3± 16 . https://doi.org/10.1016/B978-0 -12-416742-1 . 00001 -9 PMID: 24439276 .
21. Christman KL , Enriquez-Rios VD , Maynard HD . Nanopatterning proteins and peptides . Soft Matter . 2006 ; 2 ( 11 ): 928 ± 12 . https://doi.org/10.1039/b611000b
22. Narui Y , Salaita KS . Dip-pen nanolithography of optically transparent cationic polymers to manipulate spatial organization of proteolipid membranes . Chem Sci . 2012 ; 3 ( 3 ): 794 ± 9 .
Wagner MS , Pasche S , Castner DG , Textor M. Characterization of poly(L-lysine)-graft-poly(ethylene glycol) assembled monolayers on niobium pentoxide substrates using time-of-flight secondary ion mass spectrometry and multivariate analysis . Anal Chem . 2004 ; 76 ( 5 ): 1483 ± 92 . https://doi.org/10.
1021 /ac034873y PMID: 14987107 .
24. Pasche S , De Paul SM, VoÈros J , Spencer ND , Textor M. Poly(L-lysine ) -graft-poly(ethylene glycol) assembled monolayers on niobium oxide surfaces: A quantitative study of the influence of polymer interfacial architecture on resistance to protein adsorption by ToF-SIMS and in situ OWLS . Langmuir . 2003 ; 19 : 9216 ± 25 . https://doi.org/10.1021/la034111y
25. Azioune A , Carpi N , Tseng Q , TheÂry M , Piel M. Protein Micropatterns . A Direct Printing Protocol Using Deep UVs . Methods Cell Biol . 2010 ; 97(C): 133 ± 46 . https://doi.org/10.1016/ S0091 -679X( 10 ) 97008 - 8 PMID: 20719269 .
26. Strale P-O , A zioune A , Bugnicourt G , Lecomte Y , Chahid M , Studer V . Multiprotein Printing by LightInduced Molecular Adsorption . Adv Mater Weinheim . 2016 ; 28 ( 10 ): 2024 ± 9 . https://doi.org/10.1002/ adma.201504154 PMID: 26689426 .
27. Falconnet D , Pasqui D , Park S , Eckert R , Schift H , Gobrecht J , et al. A Novel Approach to Produce Protein Nanopatterns by Combining Nanoimprint Lithography and Molecular Self-Assembly . Nano Lett . 2004 ; 4 ( 10 ): 1909 ±14. https://doi.org/10.1021/nl0489438
28. Fredericks EC , Kotecha HN , inventors; United States Patent, assignee. Photoresist lift-off process for fabricating semiconductor devices . United States patent US 06/567 , 034 . 1986 Feb 14 .
29. Falconnet D , Koenig A , Assi F , Textor M. A Combined Photolithographic and Molecular-Assembly Approach to Produce Functional Micropatterns for Applications in the Biosciences . Adv Funct Mater . 2004 ; 14 ( 8 ): 749 ± 56 . https://doi.org/10.1002/adfm.200305182
30. Anderegg F , Geblinger D , Horvath P , Charnley M , Textor M , Addadi L , et al. Substrate adhesion regulates sealing zone architecture and dynamics in cultured osteoclasts . PLoS One . 2011 ; 6 ( 12 ):e28583. https://doi.org/10.1371/journal.pone.0028583 PMID: 22162778 .
31. Moeller J , Emge P , Avalos Vizcarra I , Kollmannsberger P , Vogel V . Bacterial filamentation accelerates colonization of adhesive spots embedded in biopassive surfaces . New J Phys . 2013 ; 15 ( 12 ): 125016 . https://doi.org/10.1088/ 1367 -2630/15/12/125016 related:A02FwzaWv6kJ.
32. Zhu J , Marchant RE . Design properties of hydrogel tissue-engineering scaffolds . Expert Rev Med Devices . 2011 ; 8 ( 5 ): 607 ± 26 . https://doi.org/10.1586/erd.11.27 PMID: 22026626 .
33. Guo W-H , Wang Y -l. Micropatterning cell-substrate adhesions using linear polyacrylamide as the blocking agent . Cold Spring Harb Protoc . 2011 ; 2011 ( 3):prot5582 . https://doi.org/10.1101/pdb.prot5582 PMID: 21363946 .
34. Tseng Q , Wang I , Duchemin-Pelletier E , Azioune A , Carpi N , Gao J , et al. A new micropatterning method of soft substrates reveals that different tumorigenic signals can promote or reduce cell contraction levels . Lab Chip . 2011 ; 11 ( 13 ): 2231 . https://doi.org/10.1039/c0lc00641f PMID: 21523273 .
35. Stalder AF , Kulik G , Sage D , Barbieri L , Hoffmann P. A snake-based approach to accurate determination of both contact points and contact angles. Colloids and surfaces A, Physicochemical and engineering aspects. 2006 ; 286 ( 1 ±3): 92 ± 103 . https://doi.org/10.1016/j.colsurfa. 2006 . 03 .008
36. Schulze KD , Zehnder SM , Urueña JM , Bhattacharjee T , Sawyer WG , Angelini TE . Elastic modulus and hydraulic permeability of MDCK monolayers . J Biomech . 2017 . https://doi.org/10.1016/j.jbiomech. 2017 . 01 .016 PMID: 28173919 .
37. Tseng Q , Duchemin-Pelletier E , Deshiere A , Balland M , Guillou H , Filhol O , et al. Spatial organization of the extracellular matrix regulates cell±cell junction positioning . Proc Natl Acad Sci U S A . 2012 ; 109 ( 5 ): 1506 ± 11 . https://doi.org/10.1073/pnas.1106377109 PMID: 22307605
38. Ricoult SG , Sanati Nezhad A , Knapp-Mohammady M , Kennedy TE , Juncker D . Humidified Microcontact Printing of Proteins: Universal Patterning of Proteins on Both Low and High Energy Surfaces . Langmuir. 2014 ; 30 ( 40 ): 12002 ± 10 . https://doi.org/10.1021/la502742r PMID: 25222734
39. Kaufmann T , Ravoo BJ . Stamps, inks and substrates: polymers in microcontact printing . Polymer Chemistry . 2010 ; 1 ( 4 ): 371 ± 17 . https://doi.org/10.1039/b9py00281b
40. Tan JL , Tien J , Chen CS . Microcontact printing of proteins on mixed self-assembled monolayers . Langmuir . 2002 ; 18 ( 2 ): 519 ± 23 . https://doi.org/10.1021/la011351+
41. Madou MJ . Manufacturing Techniques for Microfabrication and Nanotechnology . 3 ed: CRC Press; 2011 .
42. Qin D , Xia Y , Whitesides GM . Soft lithography for micro- and nanoscale patterning . Nat Protoc . 2010 ; 5 ( 3 ): 491 ± 502 . https://doi.org/10.1038/nprot. 2009 .234 PMID: 20203666 .
43. Madou PM , Wang PC . Photolithography. Dordrecht: Springer Netherlands; 2012 . p. 2051 ± 60 .
44. Das T , Safferling K , Rausch S , Grabe N , Boehm H , Spatz JP . A molecular mechanotransduction pathway regulates collective migration of epithelial cells . Nat Cell Biol . 2015 ; 17 ( 3 ): 276 ± 87 . https://doi.org/ 10.1038/ncb3115 PMID: 25706233 .
45. Ravasio A , Cheddadi I , Chen T , Pereira T , Ong HT , Bertocchi C , et al. Gap geometry dictates epithelial closure efficiency . Nat Commun . 2015 ; 6 : 7683 . https://doi.org/10.1038/ncomms8683 PMID: 26158873 .
46. Burnette DT , Manley S , Sengupta P , Sougrat R , Davidson MW , Kachar B , et al. A role for actin arcs in the leading-edge advance of migrating cells . Nat Cell Biol . 2011 ; 13 ( 4 ): 371 ± 81 . https://doi.org/10.1038/ ncb2205 PMID: 21423177 .
47. Kollimada SA , Kulkarni AH , Ravan A , Gundiah N. Advancing Edge Speeds of Epithelial Monolayers Depend on Their Initial Confining Geometry . PLoS One . 2016 ; 11 ( 4 ):e0153471. https://doi.org/10.1371/ journal.pone.0153471 PMID: 27078632 .
48. Pasche S , VoÈros J , Griesser HJ , Spencer ND , Textor M. Effects of ionic strength and surface charge on protein adsorption at PEGylated surfaces . J Phys Chem B . 2005 ; 109 ( 37 ): 17545 ± 52 . https://doi.org/10. 1021/jp050431+ PMID: 16853244 .
49. Tseng P , Di Carlo D. Substrates with Patterned Extracellular Matrix and Subcellular Stiffness Gradients Reveal Local Biomechanical Responses . Adv Mater Weinheim . 2014 ; 26 ( 8 ): 1242 ±7. https://doi.org/10. 1002/adma.201304607 PMID: 24323894
50. Balaban NQ , Schwarz US , Riveline D , Goichberg P , Tzur G , Sabanay I , et al. Force and focal adhesion assembly: a close relationship studied using elastic micropatterned substrates . Nat Cell Biol . 2001 ; 3 ( 5 ): 466 ± 72 . https://doi.org/10.1038/35074532 PMID: 11331874 .
51. Cavalcanti-Adam EA , Volberg T , Micoulet A , Kessler H , Geiger B , Spatz JP . Cell Spreading and Focal Adhesion Dynamics Are Regulated by Spacing of Integrin Ligands . Biophys J . 2007 ; 92 ( 8 ): 2964 ± 74 . https://doi.org/10.1529/biophysj.106.089730 PMID: 17277192
52. Ihalainen TO , Aires L , Herzog FA , Schwartlander R , Moeller J , Vogel V . Differential basal-to-apical accessibility of lamin A/C epitopes in the nuclear lamina regulated by changes in cytoskeletal tension . Nat Mater . 2015 ; 14 ( 12 ): 1252 ± 61 . https://doi.org/10.1038/nmat4389 PMID: 26301768 .
53. Tee S-Y , Fu J , Chen CS , Janmey PA . Cell Shape and Substrate Rigidity Both Regulate Cell Stiffness. Biophys J . 2011 ; 100 ( 5):L25±L7 . https://doi.org/10.1016/j.bpj. 2010 . 12 .3744 PMID: 21354386
54. Trappmann B , Gautrot JE , Connelly JT , Strange DGT , Li Y , Oyen ML , et al. Extracellular-matrix tethering regulates stem-cell fate . Nat Mater . 2012 ; 11 ( 7 ): 642 ±9. https://doi.org/10.1038/nmat3339 PMID: 22635042 .
55. Ribeiro AJ , Ang YS , Fu JD , Rivas RN , Mohamed TM , Higgs GC , et al. Contractility of single cardiomyocytes differentiated from pluripotent stem cells depends on physiological shape and substrate stiffness . Proc Natl Acad Sci U S A . 2015 ; 112 ( 41 ): 12705 ± 10 . https://doi.org/10.1073/pnas.1508073112 PMID: 26417073 .