This study provides a computational and experimental platform for the study of fluid flow properties on cell alignment and retention within grooved microfluidic channels. Computational fluid dynamic modeling was used to predict cell positioning within an experimental setup of a microfluidic device. In particular, this work shows agreement between experimental cell alignment on the groove corners and magnitudes and directions of computationally predicted shear stresses within the grooves. It was demonstrated that the localized shear stress direction and micro-circulation significantly controlled cell alignment within grooved channels. In addition, shear stress profiles were influenced by the groove widths. Small grooves (25 and 50 μm in width) resulted in the formation of micro-circulation areas, which inverted the direction of the shear stress near the base of the grooves. In 50 μm wide grooves, where the shear stress was high enough in magnitude to move the cells, the cells aligned on the upstream corner of the grooves, while cells resulted in being randomly distributed in the 25 μm wide grooves where the shear stress was extremely low. In large grooves (75, 100 μm wide), there was higher penetration of the mainstream flow and small micro-circulation areas were present only near the corners. The wall shear stress at the center of the base of the grooves had the same direction of the mainstream flow, aligning the cells on the downstream corner. These results were achieved by two different cell types (i.e. cardiomyocytes and fibroblasts) with an inlet velocity of 5.2×10−4 m sec−1.
The presence of micro-circulation areas in 50 μm wide microgrooves was in agreement with previous findings from Horner et al.30
, whom studied oxygen transport within a microgrooved bioreactor. In this work, the computational modeling was carried out for a single inlet velocity, comparable to the one applied in our study. However, as compared to our study, micro-circulation areas were studied only in 50 μm wide microgrooves. We found that micro-circulation areas also occur in smaller grooves (25 μm in width), but not in wider ones (75 and 100 μm in width), and build on their work to demonstrate the effects on wall shear stresses and subsequent alignment of cells in microchannels.
In our study, the computational model was validated by comparing the computed shear stresses at the channel walls, outside the grooves, with the corresponding analytical values. Indeed, when the Reynolds number is low (<< 1), the flow into channels can be approximated to a Stokes flow and the average wall shear stress (τfre-cell
) is usually estimated by solving Equation (2)
. Considering the channel geometry and an inlet velocity of 5.2×10−4
(5 μL min−1
flow rate), the analytically calculated wall shear stress on the channels is 0.63 dyne cm−2
, comparable to the results found by the computational model (0.65 dyne cm−2
). Computationally predicted wall shear stresses within the microgrooves were found to be significantly lower than in the main channel as reported in previous studies29
, in particular when employing comparable set-up conditions15
. Further computational models, simulating several different groove widths within the 50−75 μm range should be carried out and eventually validated through experimental tests. This work could be addressed as further development of the present research.
One of the main limitations of our computational model is that it does not take into account the actual geometry of the cells and their distribution within the microgrooves: the wall shear stresses are considered to be an index of the shear stresses acting on the cell membrane, although the cell presence is not simulated. However, for the low magnitude of Reynolds numbers, the cell presence does not strongly change the flow. It has been already calculated by previous studies that the average shear stress on the cell membrane, when the cell adheres to a channel, is similar to the average wall shear stress without the cell27
. This approximation may be also applied to the larger microgrooves (50, 75, 100 μm wide), where the flow near the base of the grooves can be approximated to a stationary flow into the channel (Poiseuille flow) and the cell size (15 μm diameter35
) is relatively small compared to the groove. On the other hand, the cell size in the small grooves (25 μm wide) becomes more relevant as compared to the groove dimensions. Thus, actual shear stress experienced by the cell can not be analytically derived and was not taken into consideration by the computational model. A reference cell diameter of 15μm was considered in the simulations because it was in agreement with the average cell size experimentally observed for the fibroblasts used in the retention tests.
We used the inlet velocity of 5.2×10−4
, since it resulted in cell alignment with low shear stresses in all the grooves (~ 0.045 dyne cm−2
). These shear stress levels were not considered to induce significant changes in cell phenotype for the short durations required for the alignment (~ 10 minutes). After cell alignment, the inlet velocity should be accurately set to optimize the fluid dynamic environment and shear stress stimulation for the particular cell type. For instance, it is known that steady shear stresses in the range of 10−15 dyne cm−2
stimulate vascular endothelial cellular responses that are essential for endothelial cell function36
. Different results were found for human chondrocytes, in which high shear stress (16.4 dyne cm−2
) was found to down-regulate the expression for extracellular matrix production37
, while low shear stress regimes (below 0.1 dyne cm−2
) increase extracellular matrix synthesis in engineered cartilage constructs38
Moreover, the inlet velocity should be optimized to avoid cells from being washed off in the device. After cell alignment, cell retention was tested in response to different flow rates. The cells positioned within the grooves were progressively washed off with increasing the shear stresses. The cell retention in the grooves was significantly influenced by the magnitude of shear stresses near the walls (where cells ware located) but not by the groove widths. To explain this finding, the average shear stress values experienced by the aligned cells near the groove corners (~ 15 μm from the corners) was calculated in 50, 75, and 100 μm wide grooves for the tested inlet velocities (). For the 25 μm wide grooves, the average shear stress was calculated on the entire base of the grooves, because cells were not significantly aligned on the corners. For the same inlet velocities, the amplitude of the considered shear stresses were similar for the 50, 75, and 100 μm wide grooves (0.1 ± 0.01 dyne cm−2 at 10.4×10−4 m sec−1), but lower for the 25 μm wide groove (7.6×10−4 dyne cm−2 at 10.4×10−4 m sec−1). This could be explained by the exclusion of cells from the CFD simulations. As expected, an increase in the calculated shear stresses as a result of the increasing flow rate resulted in increased shear stresses and lower cell retainment in the channels.
It is interesting to note that previous work18
that focused on the effect of the shear stress on cell retention have found higher percentage of cell adhesion (> 80%) after 10 minutes, when the cells were exposed to higher shear stresses than those applied in the present study. This difference is likely caused by the different levels of cell adhesion to the surfaces. In the present study, the flow was applied to channels after 10 minutes of cell seeding, not allowing cell attachment on the surface. On the contrary, in the previous study, cells were adhered on fibronectin coated microchannels.
A potential limitation of the cell retention test was the device design, since the microchannels had grooves with different widths in series (25~100 μm), allowing for possible cross-contamination. Thus, theoretically, a cell washed by a small groove may be trapped into the larger ones. However, additional experiments performed by inverting the direction of the medium flow were not significantly different from those presented in (Data not shown). Cell positioning changed in the sense that the cells aligned in the opposite corner as compared with the experiments performed with the reference flow direction, as expected (i.e. counter flow in 50 μm wide grooves and in the same direction of the flow for 75 and 100 μm wide grooves).
This simple integrated computational and experimental approach can be a powerful tool for the design of microfluidic devices with controlled fluid dynamic environments. Similar integrated approaches are currently applied for the development of microdevices for different applications such as polymerase chain reaction (PCR) analyses39
or dynamic cultures in bioreactors40,41
. This paper shows the feasibility of designing microfluidic devices which enables controlled cell alignment. For example, the predicted fluid dynamic fields and subsequent cell patterning within microgrooves may be useful for studying co-cultures. The control of co-cultures of different cell types can be performed by using a simple setup (i.e. different groove channels and flow direction of medium perfusion) as compared to previous methods42,43,44,45
. Different cell types could be aligned on opposite corners of the same microgrooves by controlling flow direction and groove width. Therefore, computational and experimental approaches for controlling localized shear stress and micro-circulation could be useful tools to understand cellular docking and alignment in a microfluidic device.