A microscale anisotropic biaxial cell stretching device for applications in mechanobiology
- First Online:
A multi-layered polydimethylsiloxane microfluidic device with an integrated suspended membrane has been fabricated that allows dynamic and multi-axial mechanical deformation and simultaneous live-cell microscopy imaging. The transparent membrane’s strain field can be controlled independently along two orthogonal directions. Human foreskin fibroblasts were immobilized on the membrane’s surface and stretched along two orthogonal directions sequentially while performing live-cell imaging. Cyclic deformation of the cells induced a reversible reorientation perpendicular to the direction of the applied strain. Cells remained viable in the microdevice for several days. As opposed to existing microfluidic or macroscale stretching devices, this device can impose changing, anisotropic and time-varying strain fields in order to more closely mimic the complexities of strains occurring in vivo.
KeywordsAnisotropic deformation Cell stretching Live-cell imaging Mechanobiology Microfabrication Microfluidics
Mechanical forces play an important role in the development, homeostasis and repair of tissues. This is mainly the result of the mechanosensitivity of many biological remodeling processes at the cellular level such as proliferation, migration, differentiation, apoptosis and extracellular matrix synthesis (Ingber 2006). Cell stretching devices have demonstrated their potential to contribute to our fundamental understanding of pathways in cellular mechanotransduction and mechanosensitivity (Huh et al. 2010; Huang and Nguyen 2013; Kim et al. 2012; Moraes et al. 2010, 2013). When integrated on a microfluidic platform, these devices offer significant improvements over their macroscale counterparts (Wang et al. 2010; Huang et al. 2010), mainly given their potential for high throughput processing as well as their ability to be combined with other on-chip functions. Recently, organ-on-a-chip systems have attracted attention by highlighting the ability to better understand the effects of mechanical forces at the cellular level for different organs (Huh et al. 2010; Kim et al. 2012). In vivo, tissue-embedded cells undergo mechanical strains that often vary spatially and temporally. It is the case in vascular tissues where the combination of the local hemodynamic forces (Frydrychowicz et al. 2008) with the anisotropic mechanical properties of vascular tissues (Duprey et al. 2010; Tremblay et al. 2010) exposed endothelial and smooth muscle cells to complex multi-axial and cyclical deformations. Moreover, these strain fields can induce significant sub-cellular, cellular- and multi-cellular remodeling responses in a frequency and magnitude dependent manner (Balachandran et al. 2011; Goldyn et al. 2009; Jungbauer et al. 2008).
Microfluidic stretching devices have been developed to study single cell response to mechanical deformation or to observe multi-culture cell system mimicking organ-level functions under mechanical stimuli. The elegant work by Huh et al. (2010) demonstrated the ability to mimic organ-level functions in a microfabricated stretching device. They were able to uniaxially stretch a co-culture of alveolar epithelial cells and endothelial cells to examine cellular responses to mechanical deformation in a model of the lung. Using a similar device Kim et al. (2012) demonstrated that human intestinal epithelial cells exhibit changes in cell morphology and increased aminopeptidase activity under cyclic uniaxial stretching. Several groups have now integrated microfabricated stretching devices into microfluidic networks in order to allow for high throughput screening. Huang and Nguyen (2013) have integrated microfabricated uniaxial devices in a high throughput platform allowing the investigation of the effect of various uniaxial stretching conditions on cell response within the same experiment. Other systems have used piston-like structures to deform a membrane on which cells are firmly attached to perform high throughput screening. Kamotani et al. (2008) employed microwells with flexible bottom membranes placed over computer-controlled, piezoelectrically actuated pins inducing a broad range of biaxial strain fields in the same microwells. Similar high throughput devices have also been used to monitor the influence of mechanical substrate strain on β-catenin accumulation in the nucleus or myofibroblast differentiation (Moraes et al. 2010, 2013). Taken together, these devices have strongly contributed to the development of a new class of microfabricated devices capable of studying cellular biological processes under mechanical strain.
Although existing devices have clear utility, an area of improvement would be the integration of full and independent biaxial control of the strain field. While idealized strain fields have provided important insights into strain-induced cellular remodelling processes, imposing more complex strain fields in the future would better mimic in vivo cellular systems. In this study, we build upon existing microfluidic stretcher designs and present a complementary device capable of imposing dynamic anisotropic biaxial strains on cells. In addition, our device can also maintain microfluidic control over the introduction of samples and allowing simultaneous imaging by optical microscopy. This device allows the independent and dynamic control of the strain magnitude and waveform frequency (milliseconds to days) in two orthogonal directions during the same stretching experiment, leading to better replication of complex multi-axial cyclic strains common to in vivo systems. We chose human foreskin fibroblast (HFF) cells as a model system for this study as fibroblasts are well known to sense and respond to strain (Wang et al. 2004). We show that the device can maintain cell viability over several days and allows the study of the same group of cells in response to a changing biaxial strain field.
Materials and methods
Working principle of the device
Before introducing cells, the device’s top and bottom fluidic channels are first wetted and sterilized with 95 % ethanol for 5 min prior to being flushed with autoclaved deionized water for another 5 min. Water is then replaced by a fibronectin solution at 10 μg/ml of HEPES-buffered salt solution (HBSS; 20 mM HEPES at pH 7.4, 120 mM NaCl, 5.3 mM KCl, 0.8 mM MgSO4, 1.8 mM CaCl2 and 11.1 mM glucose). Once the microfluidic channels are filled with the fibronectin solution, the ends of all tubing leading to the device are placed in a single solution-filled vial. This equilibrates all pressures and completely stops flow within the device, promoting fibronectin functionalization of the membrane. Fibronectin is incubated for 2 h at 5 % (v/v) CO2 and 37 °C. Subsequently, the fibronectin solution is replaced with culture medium (DMEM) supplemented with 10 % (v/v) fetal bovine serum and 1 % penicillin/streptomycin at a flowrate of ~10 μl/min. In the mean time, cells cultured in a standard incubator (5 % v/v CO2 and 37 °C) are trypsinized and resuspended in culture medium at 2 × 106 cells/ml. The top microchannel is then filled with the culture medium supplemented with cells, whereas the bottom channel is further flushed with fresh culture medium. Individual cells quickly adhere to the fibronectin-coated membrane surface within 10 s under no flow conditions. After 10 s, more cells were carried in the device’s chamber while the cells already present in the chamber remained attached to the membrane. Cells are thus immobilized to the membrane, one by one, until about 70 cells are present in the stretching chamber. Once the cells are adhered to the membrane, flow is again completely stopped by placing all tubing in the same media-filled vial. The cells are left to firmly attach to the fibronectin-coated PDMS membrane overnight. Supplementary Fig. 2 shows the speed at which cells attach to the fibronectin-coated PDMS membrane. The deposited cells are initially somewhat lined up with the fluid flow direction. However, HFFs are very motile and quickly cover the entire surface of the membrane after overnight incubation.
Image analysis and cell orientation
Cell orientations were quantified using filtered and thresholded phase-contrast images of the cells. A FFT band-pass filter was first applied on the phase-contrast images using ImageJ (http://rsbweb.nih.gov/ij/) to smooth background and isolate cell features. Thresholding was applied to create binary images of the cell features. The orientation of each of the features was computed and record to produce a histogram for each of the stretching conditions.
Cellular responses to dynamic and complex strain fields
The recent development of microscale stretching devices has provided numerous insights into the kinetics of cellular responses to mechanical strain, at various time scales (milliseconds to hours) (Huh et al. 2010; Huang and Nguyen 2013; Kim et al. 2012; Moraes et al. 2010; Jungbauer et al. 2008; Moraes et al. 2013; Wang et al. 2001; Chen et al. 2013). Here, we build upon existing designs and present a microfabricated device that allows cells to be exposed to a strain field that can be controlled in two orthogonal directions independently. Existing approaches typically employ only uniaxial strain and do not possess the ability to dynamically change its direction. This provides the ability to change strain directions on the fly and also to create dynamic, complex and anisotropic strain fields. This approach provides a method for studying cellular reorientation resulting from complex and dynamic strains that better mimic what happens in vivo. Similarly, the work of Moraes et al. (2010, 2013) demonstrated a device that could independently change the radial and circumferential strain components, albeit with a maximum strain magnitude of 6 %. Our device is able to independently change both of the strain-field components dynamically with a maximum strain magnitude of 20 %. Importantly, the device allows the investigation of the effects of pure uniaxial or standard uniaxial stretching on cellular responses. Indeed, the effect of deforming cells perpendicularly to their orientation axis can induce severe disruption of microarchitecture of valve endothelial cells (Balachandran et al. 2011). Consequently, precise control of the strain field (pure uniaxial, standard uniaxial, biaxial, equibiaxial) as well as its direction and magnitude, will facilitate a systematic understanding of how cells respond to the complex, anisotropic and time-varying strain fields they encounter in vivo.
HFF cells respond to cyclic substrate deformations by changing their morphology and orientation. Indeed, cells undergo morphological changes under uniaxial stretching by orienting themselves almost perpendicularly to the stretching direction. The orientation of the cells reflects the slight non-uniformity of the applied strain field, as evident from the inset in Fig. 4b, c. As revealed by the strain maps obtained experimentally and from finite element simulations (Fig. 2d, f), the magnitude of the vertical deformation of the membrane is non-uniform and follows a curved shape, the gradient of which is estimated by the white dashed lines. This arrangement suggests that individual cells are sensitive to local strain variation. It is still not clear what mechanisms are responsible for this behavior found in many cell types, but it is hypothesized that cells position themselves to experience the least amount of deformation (Wang et al. 2001; Faust et al. 2011). To our knowledge, cell response to non-uniform strain fields has never been investigated before. Given the microscale dimensions of our device, it is possible to investigate the effect of strain field gradients across the same cell while monitoring cellular remodeling and migration. In other applications, up-sizing the device dimensions would provide for larger areas with uniform strain, where a greater number of cells could be exposed to similar deformations.
The integration of independent biaxial stretching capabilities on a microfluidic device provides precise control over the biochemical and mechanical environments experienced by cells. We have demonstrated that cells are able to proliferate in the device and reorient themselves in response to applied strain. The ability to induce deformation along two orthogonal directions allows the investigation of how anisotropic strain modulates the mechanisms governing cellular proliferation, organization and cytoskeletal remodeling in response to cyclic stretch (Goldyn et al. 2009; Chen et al. 2013; Jaalouk and Lammerding 2009). This may contribute to our understanding of how complex and anisotropic mechanical forces and strain originating in the extra-cellular matrix couple to the cytoarchitecture. Building upon the designs of previous microfluidic or macroscale stretching devices, we present an approach that provides the user with a unique ability to generate changing, anisotropic and time-varying strain fields in order to more closely mimic the complexities of strains occurring in vivo.
DT thanks the Fond de recherche du Québec: Nature et Technologie (FQRNT) and Mitacs Elevate Program. AEP. acknowledges generous support from a Province of Ontario Early Research Award, a Canada Research Chair (CRC), a NSERC Discovery Grant and a NSERC Discovery Accelerator Supplement. MG acknowledges a NSERC Discovery Grant and a CFI grant.
- Frydrychowicz A, Berger A, Russe MF, Stalder AF, Harloff A, Dittrich S, Hennig J, Langer M, Markl M (2008) Time-resolved magnetic resonance angiography and flow-sensitive 4-dimensional magnetic resonance imaging at 3 Tesla for blood flow and wall shear stress analysis. J Thorac Cardiovasc Surg 136:400–407PubMedCrossRefGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.