In this study we found that microtubules play a major role in the alignment of HASM cells subjected to cyclic uniaxial stretch. A population of randomly oriented HASM cells showed statistically significant alignment by 1 h but this was preceded by significant cell spreading at 0.5 h. Since it has been established that cell spreading involves microtubules 
, the above finding indicates that microtubule dynamics should contribute to the stretch-induced cell reorientation in HASM cells. Furthermore, when the microtubule assembly was inhibited by colchicine or nocodazole treatment, and also when microtubules were stabilized by paclitaxel treatment, the HASM cells showed no tendency to align since cell orientation remained uniform after 2 h of stretch. To quantitatively understand individual cell dynamics that lead to these experimental observations, we developed a model in which microtubule polymerization and attachment were regulated by mechanical stretch. Together with our experimental findings, these results demonstrate, for the first time, that microtubules play a significant role in the cyclic stretch-induced cytoskeletal remodeling and subsequent cell reorientation in HASM cells.
Contradictory results have recently been reported in NIH3T3 fibroblasts by Goldyn et al 
. They observed that NIH3T3 cells can align in response to stretch even after knocking out microtubules. Similarly, Hayakawa et al. reported that the cyclic stretch-induced cell alignment was not affected by pharmacological disruption of microtubule polymerization in A10 smooth muscle cell line 
. These findings suggest that there may be cell type specific differences in the way cells remodel their cytoskeleton in response to cyclic uniaxial stretching. Even though the conclusions by Goldyn et al. 
indicate that cell re-orientation is purely a consequence of actin remodeling, they report several observations which are similar to ours. In their study, upon stretching non-treated, unstretched cells, both microtubules and F-actin re-align significantly. Further, when treated with cytochalasin D, microtubules no longer align with stretch like they do in untreated cells. More importantly, they found that microtubules can modulate the kinetics of the stretch-induced alignment of NIH3T3 fibroblasts 
. These observations support our hypothesis that the remodeling of actin in response to stretch is coupled to that of the microtubules.
In this study, we used nocodazole and colchicine to disrupt microtubules. Both drugs are also known to cause an increase in myosin phosphorylation, which may lead to the contractility of actin stress fibers 
. However, treatment of cells with nocodazole did not induce apparent F-actin formation in HASM cells (). Further, we have demonstrated that upon inhibiting microtubule dynamics by treatment with paclitaxel HASM cells fail to align (). This implies that it is not just the presence or absence of microtubules, but the dynamic nature of the polymerization process of microtubules, in response to stretch that is critical for the cell to be able to adapt to cyclically varying uniaxial stretch. Further evidence for the involvement of microtubule dynamics may be found in our measurements of projected cell area which is an indicator of how well cells spread 
. Our results indicate a highly significant increase in cell spreading after 0.5 h of stretch (). Spreading progressively decreased until 2 h at which point the projected areas were statistically the same as those of unstretched cells (). These results suggest that the reorientation process is preceded by a period of increased microtubule polymerization-depolymerization turnover and cell spreading.
It is known that microtubules can bear compressional stress in cells 
. On the extracellular side, the focal adhesions bind to extracellular matrix proteins through integrins and on the cytoplasmic side they link to actin stress fibers through various adaptor proteins. It has been shown that microtubules actively target and bind to focal adhesions 
. Moreover, microtubules and stress fibers are linked by structural and biochemical processes 
. When a local tensile strain was applied to a focal adhesion, microtubules polymerized toward the focal adhesion that was strained 
. Putnam et al. 
demonstrated that 10% static stretch of rat aortic smooth muscle cells significantly increased tubulin polymerization. Nevertheless, the relationship between the dynamics of microtubule polymerization and cyclic stretch has not been fully explained. It has been reported that the speed of tubulin polymerization within a cell is below 10 µm/min 
. In contrast, the speed of tubulin rapid de-polymerization, called a catastrophe, is much faster than polymerization speed 
. In the present study, 20% cyclic stretch was applied to the ASM cells at 30 cycle/min. Thus, it is likely that the stretching/unloading speed was sufficiently fast so that polymerization/de-polymerization of tubulin could not complete within one stretching-unloading cycle. Instead, cell shape slowly changed towards cell alignment due to cytoskeletal reorganization and focal adhesion remodeling during repeated cyclic stretching 
. As shown in , cyclic stretch induced microtubule reorientation which was accompanied by actin stress fiber formation. Similar observations of stretch-induced microtubule reorientation have been reported previously in various cell types 
. Oakley and Brunette 
demonstrated that microtubules are the first to align prior to cell alignment in cultured fibroblasts. These findings also indicate an important role of microtubule dynamics in regulation of cyclic stretch-induced cell alignment in HASM cells. Furthermore, there is a link between dynamics of microtubules and the fluctuations in mechanical forces on the stress fiber/focal adhesion assembly due to applied cyclic stretch.
Previous studies have demonstrated an important role of experimental conditions, specifically strain, frequency, duration, and cell types, in the cell reorientation induced by cyclic stretch 
. We previously reported that cyclic stretching primary cultured human pulmonary microvascular endothelial cells (20%, 12 h) induced cell alignment perpendicular to stretch axis 
. In the present study, we used relatively shorter stretch duration (within 2 h) in order to minimize genomic effects such as protein synthesis. In our preliminary results, stretch-induced alignment behavior of HASM cells was not different between 2 h and 3 h stretch, indicating that 2 h stretch was long enough for primary HASM cells to align. We also found that HASM cells still aligned when the cells were cyclically stretched at a lower (10%) strain for 2 h. Kaunas et al. 
examined the strain-dependency of stress fiber alignment induced by cyclic stretch (10%, 60 cycles/min, 6 h) in bovine aortic endothelial cells. They found that the stress fiber aligned at 3% strain but not at 1% strain. They developed a model which was capable of describing stress fiber reorganization in response to diverse temporal and spatial patterns of cyclic stretch in bovine aortic endothelial cells 
. However, the threshold of strain levels which induces cell alignment remains unclear in HASM cells and examining the threshold stain is beyond of the focus of the present study.
Mathematical models have often been employed to help elucidate the mechanism of cell alignment in response to cyclic stretch 
. Hsu et al. 
developed a dynamic stochastic model of frequency-dependent stress fiber alignment induced by cyclic stretch in endothelial cells. They concluded that when the rate of stretch is faster than the rate of stress fiber self-adjustment, the stress fibers gradually accumulate and align in a direction that require the smallest deviation form equilibrium. This is the reason why the cell aligns perpendicular to the stretch axis in response to uniaxial stretch at relatively high rate (>0.1 Hz). Thus, if the rate of stretch is faster than the rate of tubulin self-adjustment as well as stress fiber adjustment, cells align perpendicular to the stretch axis in HASM cells.
In constructing the model, we used the experimental observation that microtubules polymerize in response to stretch and depolymerize when tension on actin/myosin is inhibited to form a hypothesis that microtubule dynamics are regulated by the strain on the stress fibers. An alternative hypothesis, based on the above experimental evidence, is that they are responding to stress/strain on focal adhesions. However, we choose the former interpretation as this would imply that the cell is attempting to reestablish the homeostatic pre-stress levels that existed before stretch. This is more likely as the pre-stress in the cell is an important regulator of a variety of cellular functions 
. The model based on this assumption was able to reproduce the observed dynamics of reorientation (). Specifically, our model predicts that HASM cells parallel to the direction of stretch are the first to align, with cells closer to the equilibrium angle taking more time to re-align (). Further, we show that the process of alignment is accompanied by shape changes in the cell. The extent of this shape change also depends on the initial orientation of the cell with cells parallel to the direction of stretch becoming almost circular and then re-elongating in a new equilibrium direction of stretch 
Apart from their traditional role in an unstretched cell as compression bearing elements, in our model, microtubules under conditions of cyclic stretch are capable of bearing both compression and tension. This is a plausible assumption as microtubules are polymers that can buckle 
. Further, live imaging of cyclically stretched cells shows that microtubules straighten out under application of tensile stretch 
. However, even the simpler Poisson statistics used here are able to reproduce the statistics of cell alignment well.
In summary, our experiments and modeling results demonstrate that microtubules play a major regulatory role in the way HASM cells respond to uniaxial, cyclically varying stretch patterns. Specifically, we find that cell alignment results from the interplay of microtubules and actin stress fibers acting in tandem to maintain a homeostatic pre-strain in the cytoskeleton. These results can help better understand how ASM cells respond to pathological alterations in the mechanical properties of the underlying extracellular matrix induced by the onset of diseases such as asthma.