Extracellular matrix ligands can be patterned onto polyacrylamide hydrogels to enable the culture of human embryonic stem cells in confined colonies on compliant substrates. This method can be combined with traction force microscopy and biochemical assays to examine the interplay between tissue geometry, cell-generated forces, and fate specification.
Human embryonic stem cells demonstrate a unique ability to respond to morphogens in vitro by self-organizing patterns of cell fate specification that correspond to primary germ layer formation during embryogenesis. Thus, these cells represent a powerful tool with which to examine the mechanisms that drive early human development. We have developed a method to culture human embryonic stem cells in confined colonies on compliant substrates that provides control over both the geometry of the colonies and their mechanical environment in order to recapitulate the physical parameters that underlie embryogenesis. The key feature of this method is the ability to generate polyacrylamide hydrogels with defined patterns of extracellular matrix ligand at the surface to promote cell attachment. This is achieved by fabricating stencils with the desired geometric patterns, using these stencils to create patterns of extracellular matrix ligand on glass coverslips, and transferring these patterns to polyacrylamide hydrogels during polymerization. This method is also compatible with traction force microscopy, allowing the user to measure and map the distribution of cell-generated forces within the confined colonies. In combination with standard biochemical assays, these measurements can be used to examine the role mechanical cues play in fate specification and morphogenesis during early human development.
Human embryonic stem cells (hESCs) hold great promise for use in regenerative medicine and tissue engineering applications. The pluripotent nature of these cells gives them the ability to differentiate into any adult cell type. While great strides have been made in directing the fate of hESCs to particular cell types, it has remained very difficult to generate whole tissues or organs de novo1,2,3,4,5. This is due, in large part, to a limited understanding of the mechanisms that drive the formation of these tissues during human development. In order to fill this gap in knowledge, a number of methods have emerged in recent years to model the early embryo and subsequent stages of development with embryonic stem cells6,7,8,9,10,11,12,13.
Shortly after the derivation of the first hESC lines14, it was demonstrated that embryoid bodies formed from hESCs were capable of spontaneously producing cells of the three primary germ layers6. However, due to the inherent lack of control over the size and morphology of embryoid bodies, the organization of germ layers varied significantly and failed to match the organization of the early embryo. More recently, Warmflash et al. developed a method to confine colonies of hESCs on glass substrates via micropatterning, providing control and consistency over the size and geometry of the colonies8. In the presence of BMP4, an important morphogen in early development, these confined colonies were capable of self-organizing reproducible patterns of specification to fates representing the primary germ layers. Although this provided a useful model for studying the mechanisms by which primary germ layers are established, the patterns of fate specification did not precisely match the organization and morphogenesis observed during embryogenesis15. A more faithful recapitulation of early embryonic development was achieved by embedding hESCs in a three-dimensional extracellular matrix (ECM) of matrigel11, providing the strongest evidence to date for the ability of hESCs to self-organize and model the early stages of embryogenesis ex vivo. However, this method yields inconsistent results and is thus incompatible with a number of assays that could be used to reveal the underlying mechanisms of self-organization and fate specification.
Given these existing methods and their respective limitations, we sought to develop a method for reproducibly culturing hESC colonies of defined geometries in conditions that model the extracellular environment of the early embryo. To achieve this, we used polyacrylamide hydrogels of tunable elasticity to control the mechanical properties of the substrate. Using atomic force microscopy on gastrulation-stage chicken embryos, we found that the elasticity of the epiblast ranged from hundreds of pascals to a few kilopascals. Thus, we focused on generating polyacrylamide hydrogels with elasticity in this range to serve as the substrate for hESC colonies. We modified our previous methods for culturing hESCs on polyacrylamide hydrogels7,9 to provide robust control over the geometry of the colonies. We achieved this by first patterning ECM ligands, namely matrigel, onto glass coverslips through microfabricated stencils, as previously reported16. We then designed a novel technique to transfer the patterned ligand to the surface of polyacrylamide hydrogels during polymerization. The method we describe here involves using photolithography to fabricate a silicon wafer with the desired geometric patterns, creating stamps of theses geometric features with polydimethylsiloxane (PDMS), and using these stamps to generate the stencils that ultimately allow patterning of ligand onto the surface of glass coverslips and transfer to polyacrylamide.
In addition to recapitulating the mechanical environment of the early embryo, confining hESC colonies on polyacrylamide enables the measurement of cell-generated forces with traction force microscopy (TFM), as reported in our previous method9. In brief, fluorescent beads can be embedded in the polyacrylamide and used as fiducial markers. Cell-generated forces are calculated by imaging the displacement of these beads after seeding hESCs onto the patterned substrate. Furthermore, the resulting traction force maps can be combined with traditional assays, such as immunostaining, to examine how the distribution of cell-generated forces in confined hESC colonies may regulate or modulate downstream signaling. We expect these methods will reveal that mechanical forces play a critical role in the patterning of cell fate specification during early embryonic development that is currently overlooked.
All methods described here pertaining to the use of hESCs have been approved by the Human Gamete, Embryo and Stem Cell Research (GESCR) Committee at the University of California San Francisco.
1. Preparation of silicon wafer with geometric features
2. Preparation of coverslips
3. Generation of stencils for patterning ECM ligand
4. Patterning ECM ligand on coverslips
5. Transfer of ligand to polyacrylamide gel
6. Culturing hESCs on patterned gels
NOTE: If planning to fix samples for immunostaining after TFM, take images of unstressed microsphere positions prior to seeding cells. In this case, fluorescently tagged ligand should be used in step 4.3.1 so that the eventual locations of cells will be known before seeding and microspheres can be imaged in those regions.
7. Performing TFM
The main challenge to overcome in attempting to culture hESCs in colonies of controlled geometry on compliant substrates is to generate a homogenous pattern of ECM-ligand on the surface of the substrate. The strategy presented in this method involves first generating the desired pattern on the surface of a glass coverslip and then subsequently transferring that pattern to the surface of a polyacrylamide hydrogel during polymerization of the gel (Figure 1A). Thus, it is important to ensure the desired pattern is created successfully on the surface of the glass coverslip prior to proceeding to polymerization of the hydrogel and transfer of the pattern (Figure 1B). Based on imaging of fluorescent ligand patterns transferred to polyacrylamide, the ligand appears to be present only in a single plane at the surface of the polyacrylamide, though we did not precisely characterize the thickness of this layer. There are two common types of defects observed following pattern generation on the glass coverslip, each with its own source of error. The first is the appearance of fluorescent ligand extending beyond the margins of the desired pattern (Figure 1C, top), which results from leaking of the fluorescent ligand solution due to a small tear in the stencil or insufficient sealing of the stencil to the glass coverslip. The second is the appearance of an incomplete pattern (Figure 1C, bottom), which is typically due to an air bubble trapped at the coverslip interface that prevents adsorption of the fluorescent ligand.
The ultimate measure of success for this method is the ability to culture hESCs in desired geometries on the patterned hydrogels (Figure 2). In order to achieve this, hESCs are seeded at a relatively high density (300,000 cells/mL) in the presence of a Rho kinase inhibitor (Y27632) and incubated for 3 h to facilitate adhesion to the patterns of ligand. Media is then replaced to remove non-adhered cells. Over the course of 72 h, the Y27632 is gradually diluted out of the media by a series of media exchanges at 24 and 48 h post-seeding. Typically, the hESCs proliferate to complete the patterned geometries by 48-72 h, such that experiments can begin at 72 h post-seeding, once the Y27632 is completely removed from the media.
Culturing hESC colonies in confined geometries on polyacrylamide hydrogels permits the measurement of cell-generated traction forces using TFM. These measurements are made by embedding fluorescent microspheres in the hydrogel and imaging the positions of these beads before and after seeding hESCs (Figure 3A). The displacement of the beads following cell seeding is a function of cell-generated forces and the elasticity of the hydrogel, thus the images of the bead positions can be used to generate maps of bead displacements and subsequently used to calculate the underlying traction stresses. In circular colonies of hESCs, the largest traction stresses are found near the peripheral edge of the colonies, while the center of the colonies display uniformly low traction stresses (Figure 3B). Interestingly, the highest traction stresses are found in clusters near the edge of the colonies, rather than forming a continuous ring of maximal stress. This implies that although colony geometry plays a key role in determining the distribution of traction stresses, more localized regulation and feedback determines the precise locations of maximal stress. Additionally, so long as the image of microsphere positions without adhered cells is taken prior to cell seeding, hESC colonies can be fixed for immunostaining of proteins of interest following traction force measurements. Despite the observed non-uniform distributions of traction stresses, hESCs cultured as patterned circles in maintenance conditions display uniform expression of the pluripotency marker Oct3/4 and cell adhesion molecule E-cadherin (Figure 3C).
The presented method for using stencils to generate patterns of ligand is demonstrably superior to the common technique of microcontact printing for the relatively large geometries used in this method (i.e., for circular colonies 1 mm in diameter as well as triangles and squares of equivalent area). Microcontact printing patterns of ligand at this length scale results in heterogeneous transfer at the edges of patterns, with very little ligand deposited in the central regions of the patterns (Figure 4A). This is clearly insufficient for consistently producing hESC colonies of specific geometries. Reducing the cell seeding density also leads to inconsistency in achieving completed colonies. Cells seeded at 200,000 cells/mL, rather than 300,000 cells/mL, are not able to generate sufficient cell-cell contacts to survive the reduced concentration of Y27632 at 24 h post-seeding (Figure 4B). It may be possible to generate complete patterns with lower cell densities by extending the period of Y27632 dilution; however, overall it is more efficient to seed at the higher 300,000 cells/mL density. Occasionally, errors in pattern generation that are not detected earlier in the protocol become apparent upon seeding of hESCs. One such error is the leaking of ligand solution underneath the stencil due to poor contact with the coverslip. This ultimately results in ligand being transferred to regions of the polyacrylamide outside the desired geometries, and unconfined growth of hESCs upon seeding (Figure 4C).
Figure 1: Patterning of ECM ligand onto acid-washed coverslips and transfer to polyacrylamide hydrogels. (A) Schematic representation of the protocol for patterning ECM ligand on acid-washed coverslips and transferring the patterned ligand to polyacrylamide hydrogels. (B) Immunofluorescent images of patterned ECM ligand on acid-washed coverslips (left) and following transfer to polyacrylamide hydrogel (right). Biotin-tagged matrigel was patterned onto the coverslip and labeled with Alexa Fluor 555 streptavidin prior to transfer to polyacrylamide. Insets show zoomed-out view of the full patterns generated. (C) Representative fluorescent images demonstrating patterning defects of ligand on the glass coverslip. These result from common errors in the protocol, such as leaking of the ligand solution outside the patterned geometry (top, arrow indicates site of leak), and trapping of air bubbles inside the patterned geometries of the stencil upon adding the ligand solution (bottom, arrow indicates site of air bubble). All scale bars = 500 µm. Please click here to view a larger version of this figure.
Figure 2: Seeding of hESCs onto patterned polyacrylamide hydrogels. Representative brightfield images demonstrating successful seeding of hESCs onto polyacrylamide hydrogels with patterned ligand. The timeline at the top indicates the series of media changes used to remove unattached cells and dilute out the Y27632. Note that after initial seeding, cells adhere stochastically to various regions within the patterned ligand and then proliferate to fill the patterned regions over the course of 72 h. Scale bar = 500 µm. Please click here to view a larger version of this figure.
Figure 3: Regional localization of traction stresses and immunostaining of confined hESC colonies. (A) Fluorescent images of microspheres within the polyacrylamide hydrogel before and after seeding hESCs. (B) Representative particle image velocimetry (PIV) plot depicting the displacement of microspheres due to traction stresses (left) and corresponding reconstructed traction stresses (right). (C) Immunostaining of confined hESC colonies on patterned polyacrylamide hydrogels, demonstrating the ability to compare localization of proteins of interest to traction stresses. All scale bars = 200 µm. Please click here to view a larger version of this figure.
Figure 4: Non-ideal results yielded by alternative methods and errors. (A) Representative fluorescent images of ECM ligand patterned onto acid-washed coverslips via microcontact printing. (B) Brightfield images of hESCs that fail to form completed colonies due to insufficient adherence of cells. Large gaps remaining in the colonies at 24 h post-seeding (arrows) are an early indicator of this issue. (C) Brightfield images of hESCs that fail to form confined colonies due to errors in patterning that led to presence of ligand outside the desired geometries. All scale bars = 500 µm. Please click here to view a larger version of this figure.
Example SU8 Wafer Fabrication | |
1. Spin coat wafer with SU8-3050 | Spin at 1,000 RPM for 30 s |
2. Soft bake wafer on hot plate | i. 3 min at 65 °C |
ii. 45 min at 95 °C | |
iii. 3 min at 65 °C | |
iv. Cool to room temperature | |
3. Expose in mask aligner | i.Align wafer and photomask in mask aligner |
ii.Expose with energy of 250 mJ/cm2 | |
• E.g. for lamp with intensity of 11 mW/cm2, expose for 23 s | |
4. Post exposure bake on hot plate | i. 1 min at 65 °C |
ii.15 min at 95 °C | |
iii. 1 min at 65 °C | |
iv.Cool to room temperature | |
5. Develop | i. Agitate in SU8 Developer, approx. 5-10 min |
ii.Check development with isopropyl alcohol (IPA) | |
• If under-developed, IPA rinse will produce a white residue | |
6. Hard bake on hot plate (optional) | 1-2 h at 150 °C |
Table 1: Example SU8 wafer fabrication protocol. Silicon wafers used to generate PDMS stamps and subsequent stencils were fabricated using the steps outlined in this table. This protocol was generated using the SU8 3000 data sheet with the aim of creating a film thickness of approximately 100-250 µm.
Polyacrylamide Gel Formulations | |||||||||
Volumes for 1 mL complete solution | |||||||||
Elastic Modulus (Pa) | Final % acrylamide | Final % Bis-acrylamide | ddH2O (μL) | 40% acrylamide (μL) | 2% Bis-acrylamide (μL) | 10x PBS (μL) | 1% TEMED in ddH2O (μL) | Microspheres 0.5% solids in ddH2O (μL) | 1% PPS in ddH2O (μL) |
1050 | 3 | 0.1 | 565 | 75 | 50 | 100 | 75 | 60 | 75 |
2700 | 7.5 | 0.035 | 485 | 187.5 | 17.5 | 100 | 75 | 60 | 75 |
4000 | 7.5 | 0.05 | 477.5 | 187.5 | 25 | 100 | 75 | 60 | 75 |
6000 | 7.5 | 0.07 | 467.5 | 187.5 | 35 | 100 | 75 | 60 | 75 |
Table 2: Polyacrylamide gel formulations. Volumes of components for generating polyacrylamide hydrogels of various elastic moduli with fluorescent microspheres for TFM. Volumes can be scaled up or down based on amount of polyacrylamide solution needed. All components except the fluorescent microspheres and PPS are mixed together prior to degassing. PBS = phosphate-buffered saline, TEMED = tetramethylethylenediamine, PPS = potassium persulfate.
To simplify a long and detailed protocol, this method consists of three critical stages: 1) generating patterns of ECM ligand on glass coverslips, 2) transferring the patterns to polyacrylamide hydrogels during polymerization of the gel, and 3) seeding hESCs on the patterned hydrogel. There are critical steps that must be considered at each of these three stages. In order to generate high-fidelity patterns on the glass coverslips, the stencil must be firmly pressed onto the coverslip to prevent leaking of the ligand solution and all air bubbles must be removed after adding ligand solution to the surface of the stencil (Figure 1C). If ligand solution does leak through the stencil due to poor contact with the coverslip, ligand will be transferred to the entire surface of the polyacrylamide hydrogel and hESCs will not be confined to the desired colony geometry (Figure 4C). The most important step in transferring the patterned ligand to polyacrylamide is gently separating the top coverslip from the hydrogel while all components remain submerged in PBS. If the hydrogel does not remain submerged during separation, the patterns may be completely destroyed. Furthermore, if the separation occurs too rapidly, the surface of the hydrogel may tear or be otherwise damaged. The softer the hydrogel, the more it is at risk for damage during separation. Finally, the user must pipette very carefully when exchanging media during the seeding and culture of hESCs on the patterned hydrogels. The hESCs remain loosely adhered throughout the protocol and the patterned colonies can be easily disrupted by careless or rushed pipetting.
In addition to the critical steps discussed above, there are a number of other steps that may require modification and troubleshooting when adapting this protocol for different applications. While the patterns demonstrated here are on the length scale of hundreds of microns to a millimeter, generating patterned features on silicon wafers with transparency photomasks and negative photoresist allows for feature sizes all the way down to 7-10 µm. Thus, this protocol could be adapted for confining the geometry of single cells or smaller colonies of a few cells on compliant substrates.
Two parameters that will likely require optimization when adapting this protocol for different cell types are the type of ECM ligand used and the concentration of the ligand in solution during adsorption to the glass coverslip through the stencils. A total ligand concentration of 250 µg/mL was sufficient for producing homogenous patterns of matrigel and facilitating attachment of hESCs (Figure 1B and Figure 2), though it may be possible to achieve similar results with lower concentrations. On the other hand, it may be necessary to further increase the concentration of ligand for cell types that are less adherent or for applications that require shorter incubation times. Increased concentration of ligand may also be required for different ECM ligands, such as fibronectin or collagen, which are less hydrophobic than matrigel and therefore may not adsorb to the hydrogel as strongly. Because the transfer of ligand from coverslip to hydrogel occurs during polymerization of the hydrogel, commonly used techniques for robustly cross-linking the ECM ligand to the hydrogel surface (such as sulfo-SANPAH treatment) are impossible. Using fluorescently-labelled ECM ligands to enable visualization of the patterns at each step of the protocol is extremely helpful when optimizing and troubleshooting these parameters.
Additionally, the protocol for seeding cells may require optimization depending on the cell type and media conditions used. For cells that adhere more rapidly or efficiently, a lower seeding density may be required to prevent cell-cell adhesions that span between patterns and result in aggregates rather than patterned monolayer colonies. For cells that display very poor attachment, a larger seeding density or longer length of time before the initial media swap may be required to facilitate complete formation of confined colonies (Figure 4B).
The key limitation of this method is its technical complexity, which results in relatively low-throughput results compared to similar methods that involve culturing cells on patterned glass substrates8. However, this drawback is far outweighed by the physiological relevance achieved by culturing confined hESC colonies on compliant substrates. By effectively recapitulating the mechanical properties of the early embryo, we are able to better model and understand the processes that lead to self-organization of the primary germ layers.
An additional benefit of confining hESC colonies on polyacrylamide hydrogels is that it enables the use of TFM to examine the link between the organization of an hESC colony, as a model of the early embryo, and the distribution of cell-generated forces that may underlie morphogenesis and cell fate specification. Confining hESC colonies results in distributions of traction stresses that are dependent on colony geometry (Figure 3B). Despite the non-uniformity of these traction stresses, cells throughout the colonies remain pluripotent in maintenance conditions (Figure 3C). However, we hypothesize that the traction stress distributions may be involved in regulating patterns of cell fate specification by tuning the response to induction cues, such as soluble morphogens. We anticipate that this method will allow us and other groups to better model the early human embryo with hESCs, leading to a more complete understanding of the fundamental processes that underlie human embryogenesis.
The authors have nothing to disclose.
We would like to acknowledge funding from CIRM grant RB5-07409. J.M.M. would like to thank FuiBoon Kai, Dhruv Thakar, and Roger Oria for various discussions that guided the generation and troubleshooting of this method. J.M.M. also thanks the UCSF Discovery Fellowship for the ongoing support of his work.
0.05% Trypsin | Gibco | 25300054 | |
100 mm glass petri dish | Fisher Scientific | 08-747B | |
100 mm plastic petri dish | Fisher Scientific | FB0875712 | |
15 mL conical-bottom tubes | Corning | 352095 | |
150 mm plastic petri dish | Fisher Scientific | FB0875714 | |
18 mm diameter #1 coverslips | Thermo Scientific | 18CIR-1 | |
2% bisacrylamide | Bio-Rad | 161-0142 | |
3-aminopropyltrimethoxysilane | ACROS Organics | 313251000 | |
40% acrylamide | Bio-Rad | 161-0140 | |
Aluminum foil | Fisher Scientific | 01-213-100 | |
Basic fibroblast growth factor | Sigma-Aldrich | F0291 | |
Bleach | Clorox | N/A | |
Centrifuge with swing-buckets | Eppendorf | 22623508 | Model: 5804 R |
Collagen | Corning | 354236 | |
Dessicator | Fisher Scientific | 08-642-7 | |
Ethanol | Fisher Scientific | AC615095000 | |
Fetal bovine serum | Gibco | 16000044 | |
Fluorescent microspheres | Thermo Scientific | F8821 | |
Forceps (for coverslips) | Fisher Scientific | 16-100-122 | |
Forceps (for wafers) | Fisher Scientific | 17-467-328 | |
Gel holders | N/A | N/A | Gel holders are custom 3D-printed, CAD drawing available on request |
Glutaraldehyde | Fisher Scientific | 50-261-94 | |
HEPES | Thermo Scientific | J16926A1 | |
Hot plate | Fisher Scientific | HP88854100 | |
Hydrochloric acid | Fisher Scientific | A144S-500 | |
Isopropyl alcohol | Fisher Scientific | A416-500 | |
Kimwipes (delicate task wipes) | Kimberly-Clark Professional | 34120 | |
Knockout serum replacement | Gibco | 10828028 | |
Knockout-DMEM | Gibco | 10829018 | |
Mask aligner (for photolithography) | Karl Suss America, Inc. | Karl Suss MJB3 Mask Aligner | |
Matrigel | Corning | 354277 | |
Microscope for traction force | Nikon | N/A | Model: Eclipse TE200 U |
Motorized positioning stage | Prior Scientific | N/A | Model: HLD117 |
Nitrogen gas | Airgas | NI 250 | |
Norland optical adhesive 74 (UV-curable polymer) | Norland Products | NOA 74 | |
Oven | Thermo Scientific | PR305225G | |
Parafilm (laboratory film) | Fisher Scientific | 13-374-12 | |
PDMS (Sylgard 184) | Fisher Scientific | NC9285739 | |
Photomask | CAD/Art Services, Inc. | N/A | Photomasks are custom made. CAD drawing for our designs available upon request |
Plasma cleaner | Fisher Scientific | NC9332171 | |
Plastic for gasket | Marian Chicago | HT6135 | |
Plastic for spacer | TAP Plastics | N/A | Polycarbonate sheet, .01 inch thickness |
Potassium chloride (for making PBS) | Fisher Scientific | P217-500 | |
Potassium phosphate monobasic (for making PBS) | Fisher Scientific | P285-500 | |
Pottassium persulfate | ACROS Organics | 424185000 | |
Scalpel | Fisher Scientific | 14-840-00 | |
Silicon wafer | Electron Microscopy Sciences | 71893-06 | Type P, 3 inch, silicon wafers |
Sodium chloride (for making PBS) | Fisher Scientific | S271-1 | |
Sodium hydroxide | Fisher Scientific | S318-100 | |
Sodium phosphate dibasic dihydrate (for making PBS) | Fisher Scientific | S472-500 | |
SU8-3050 Photoresist | MicroChem | SU8-3000 | |
SU8-Developer | MicroChem | Y020100 | |
TEMED | Bio-Rad | 161-0800 | |
UV-sterilization box | Bio-Rad | N/A | Bio-Rad GS Gene Linker UV Chamber |
Y27632 (Rho kinase inhibitor) | StemCell Technologies | 72304 |