Biotechnol. Prog. 1995, 11, 596-600
596
Fluctuating Shear Stress Effects on Stress Fiber Architecture and Energy Metabolism of Cultured Renal Cells Vinayak D. Bhat,* Paula A. Windridge, Robert S. Cherry3 and Lazaro J. Mandel* Center for Biochemical Engineering, Duke University, and Department of Cell Biology, Duke University Medical Center, Durham, North Carolina 27708-0276
The project investigates the relationship between the external shear force and the actin cytoskeleton along with the metabolic changes occurring inside the cells due to this force. Anchorage-dependent Madin Darby canine kidney (MDCK) cells were placed in spinner flasks with paddle-type stirrers agitated at 20 rpm, where they experienced shear stress fluctuations from 0.02to 0.27 dyn/cm2 in magnitude. Following furation, permeabilization, and staining with rhodamine-phalloidin, the relative amounts and distribution of F-actin stress fibers in the 1 pm basal layer of the cells were visualized by confocal microscopy. These structures disappeared after 12-15 h of exposure to shear stress. Previous results showed that the stress fibers disappear, leading to loss of epithelial attachment, after only 1 h of starvation-induced energy depletion. Therefore, in this study, the energy metabolism of the cells was established by measuring adenosine triphosphate (ATP) levels at different time intervals. No statistical difference in ATP content was found between the shear-stressed cells and the controls, showing that shear stresses cause cytoskeletal reorganization by a mechanism other than ATP depletion.
Introduction and Background Hydrodynamic effects on cells in agitated reactors and viscometers were f i s t reported by Midler and Finn (1966) for protozoa cells. Since then there have been many reports on the effect of well-defined flows on freely suspended and anchorage-dependent cells. It has been established that laminar shear stresses affect the shape, physiology, cytoskeletal structure, and formation of certain metabolites and proteins in certain cells (mostly endothelial). Confluent monolayers of endothelial cells subjected to steady shear stress elongate in the direction of shear (Levesque and Nerem, 1985). The rate and amount of elongation is an increasing function of shear stress, appearing after 1 h at stresses as low as 4 dyn/cm2. Changes in the cytoskeleton architecture occur, the most visible ones being alignment of microtubules (Ives et al., 1986)and the appearance of stress fibers (Franke et al., 1984). Disruption of the actin network with cytochalasin in hybridoma cells (Peterson, 1989)and endothelial cells (Wang et al., 1993)leads to reduced shear resistance of the cells to external forces. Microtubules and intermediate filaments have also been shown to contribute toward the shear resistance of cells. Furthermore, while exposure to 1.5 dyn/cm2of laminar shear stress has no effect on endothelial cells, turbulent flow at the same mean shear stress causes an increase in DNA synthesis (Davies et al., 1986). Cells grown in mixed bioreactors experience turbulent shear stress which might elicit diminished cell.growth and viability. Such a situation may preclude growing certain types of cells under bulk conditions, significantly
* Corresponding
author: Vinayak D. Bhat, Department of Biomedical Engineering, Box 90281,Duke University, Durham, NC 27708. Present address: Idaho National Engineering Laboratory, P.O. Box 1625,Idaho Falls, ID 83415. Duke University Medical Center.
reducing their biotechnological potential. Although the actin cytoskeleton appears to be involved in shear resistance, little is known about the underlying mechanisms. It could be that the shear stresses injure the cells, perhaps imposing a strong metabolic demand and making the cells energy-limited. Under this scenario, shear resistance may be a function of energy availability. The overall objective of this project was to gain a better understanding of the cellular mechanisms underlying stress sensitivity and stress resistance in cultured epithelial cells. The main goals were to document the effect of fluctuating shear stress on the actin cytoskeleton over a period of time and to explore the relation between this disruption and the energy content of the cells. All experiments were carried out with Madin Darby canine kidney (MDCK) cells. This cell line was selected because its three-dimensional cytoskeletal structure and many of its transport properties have been well-characterized (Bacallao et al., 1989). The cells were grown to confluency on glass slips in six-well plates before being transferred to the spinner flask for shear stress exposure. They were then removed a t different time intervals for comparison with controls from a no-stress environment. The F-actin in the cells under all experimental conditions was studied by confocal microscopy. Cellular ATP and protein content were also measured. The shear stress experienced by the cells inside the spinner flask was measured using hot-film anemometry.
Materials and Methods Spinner Flask Apparatus. The spinner flask was selected as it produced turbulence with low mean shear stress. A Corning spinner flask (125 mL) was fitted with two Teflon annular rings at the bottom (Figure 1). The external ring secured the slotted internal ring which held the cell-seeded glass cover slips (22 x 22 mm). All the experiments were carried out at an impeller speed of 20 rpm.
8756-7938/95/3011-0596$09.O0/00 1995 American Chemical Society and American Institute of Chemical Engineers
Biotechnol. Prog., 1995, Vol. 11, No. 5
597
power input required to maintain the sensor’s heated resistance, which depends on temperature (Geremia, 1972). It was of the form (shear
Glass cover dips with confluent layer of cells fvcinp the impeller Outer rins
-n iner
Outer rlny ring with slots for the glass cover slips
Figure 1. Spinner flask apparatus: impellar diameter = d = 4 cm, medium density = Q = 1 g/cm3, medium viscosity = p = 0.0075 g/cms, impeller speed = n = 20 rpm = 0.33 rps, Reynolds number = Re = d2n@(u= 700 [Re < 10 (laminar), 10 < Re < 1000 (transitional),Re > 1000 (turbulent)].
Cell Culture. Madin Darby canine kidney (MDCK) cells were cultured and grown on glass cover slips as described by Bhat et al. (1995). Once confluency was achieved the cover slips were placed inside the spinner flask with the cells facing the impeller. One cover slip from each batch was kept in a six-well plate, where it faced no stress and served as a control. The spinner flask and control plate were incubated and maintained in sterile conditions. The cover slips were removed from the spinner flasks at different time intervals for F-actin staining or energy measurements. Fixation and Staining. Cells were fured and stained by following the procedures described by Bacallao and Stelzer (1989).The cover slips with stained cells were placed cell-side down onto microscope slides on support mounts ( 4000. As the shear stress probe had a frequency response of 1 kHz, it could easily measure stresses varying at 1-100 Hz. Hence this probe was used to measure the shear stress within the quasirandom turbulent system of the spinner flask. The mean voltage recorded when the probe was placed on the glass slide inside the spinner flask was 2.26 V. The fluctuations were added to this mean, and the shear rate was obtained using the calibration curve. The plot of shear stress experienced by the probe on the slide inside the spinner flask at 20 rpm over a representative segment of time is shown in Figure 2a. A total of 4096 data points were collected over 8 s to produce this profile. The shear stress fluctuated from as low as 0.02 to 0.27 dyn/cm2, and the mean shear stress was 0.092dyn/cm2. This range of fluctuating shear stresses seen by the cells inside the spinner flask suggested t4e presence of turbulence despite the low mixing speed. The shear stress peaks correspond to the passage of the impeller tip and occur every 1.5 s as expected from the 20 rpm impeller speed. The shear stress files were run through autocorrelation and fast Fourier transform software (Turbo Pascal Numerical Methods Toolbox) to produce the power spectra. Figure 2b shows the shear stress profile as a
Biotechnol. frog., 1995, Vol. 11, No. 5
598
a
a N
E
..
,,
,.
..
,,
.,
,.
. ...
, . . . . . . . ................................................................. . . . .
.
h
.
..
..
..
..
0
V
0 I
1 '
2
3 4 5 6 Time, seconds '
"""'I
"""'I
'
' ' ' 1 ' ' ' I
7
8
- -
- -
. . . . . . . . . . . . . . . . . . . . .
0.5
1 1.5 Time, seconds
0
E ' ""I
'
"""'!
'
...................................
2
q'""l *.............. ;........................ a i ' ' ' ' ' ' !
'
d
a
10'
Y
. C . . I I
a 6 0.1
1
10 100 Frequency, Hz
lo00
Figure 2. (a) Shear stresses experienced by the cells inside the spinner flask a t 20 rpm. (b)Power spectrum for the shear stresses in the spinner flask at 20 rpm. Amplitude refers to the Fourier transform of the autocorrelation of the shear stress values plotted in a.
power spectrum showing the frequency distribution of different shear stress magnitudes measured by the sensor. The highest shear stress amplitude values, to loT5(dyn/cm2)2,appear to be around ranging from a frequency of 1/(1.5 s) or 0.67 Hz. For comparison purposes, the shear stress fluctuations were observed for higher impeller speeds of 60 and 150 rpm (Figure 3a). The mean shear stress at 60 rpm was 0.7 dyn/cm2,with peak fluctuations of 1.2 dyn/cm2. The mean shear stress at 150 rpm was 2.0 dyn/cm2,and the fluctuations went up to 8 dyn/cm2. The power spectra show maximal shear stress amplitudes ranging from to (dyn/cm2)2,at about 1Hz (Figure 3b). These two power spectra show a 1 order of magnitude increase in energy input over that observed at 20 rpm. The Cytoskeleton. Rhodamine-phalloidin was used to label F-actin filaments as it is very specific for F-actin and will not bind to G-actin or produce any detectable nonspecific staining of other components (Barak et ai., 1981). The stained cells were observed under confocal microscopy and stress fibers in the basolateral layer were photographed. The control showed brightly stained well-developed stress fibers with a random orientation (Figure 4a). They also showed F-actin filaments localized at the periphery of the cells which are referred to as cortical actin (Bacallao et al., 1994). Within 3 h of shear stress exposure the stress fibers started to disappear (Figure 4b). The F-actin Qppeared to start depolymerizing, and there was a large loss of anchored cells. After 6 h of stress, the cells showed very few F-actin filaments and the staining was very dim (Figure 44. The peripheral F-actin also started to disappear, and there was further loss of cells as they fell off. There were very few cells remaining on the slide after 12 h insidk the spinner flask, and there was a negligible amount of F-actin remaining
-
.......?....... IU
lo-'
..............i.0.Q ..........
1oo 10' Frequency, Hz
1o2
Figure 3. (a) Shear stresses on a coverslip inside the spinner flask at 60 and 150 rpm. (b) Power spectra for the shear stresses in a.
in the cells (Figure 4d). These figures show that the F-actin appeared to depolymerize and disappeared due to the external hydrodynamic shear forces. Cell Count and Viability. Each confluent slide held approximately a million cells. The cells which fell off the slide inside the spinner flask were seeded onto new slides under optimum conditions and fresh medium, but they failed to grow. They were later treated with trypan blue and found to be dead, confirming the cells remaining on the slide to be the viable ones. The decrease in the viable cells under stressed conditions over a period of time is shown in Figure 5. The data for the plot were taken from eight sets of experiments. Metabolic Energy Content. Bacallao et al. (1994) have shown that there is a consistent decrease in the stress fiber density at the basolateral substrate attachment sites with increasing time of energy depletion, leading to complete loss of attachment a h r about 1h. Kroshian et al. (1994) have shown that cells subjected to chemical anoxia lose all the stress fibers and the F-actin depolymerizes, also showing that cellular ATP depletion causes cytoskeleton breakdown. Since the stress fiber density decreased with time in the spinner flask, the metabolic energy content of the control and stressed cells was measured to check for any depletion due to the stressed environment. The hypothesis to be tested was that fluctuating shear stresses injured the cells, imposing such a strong metabolic demand that the cells became energy-starved, undergoing cytoskeleton breakdown and subsequent cell death. The total amount of protein on each slide decreased with increasing time in the stressed environment, which was due to an increasing loss of cells from the slide. The protein content per cell was concluded to be constant and not dependent on shear stress since the total amount of protein matched well with the number of cells on the slides in all the experiments (1.096 pg/cell). The energy measurements for each set were performed by calculating the amount of ATP per milligram of
Biotechnol. Prog., 1995,Vol. 11, No. 5
599
r
r
II
a
i
I1
I
Figure 4. Images of rhodamine-phalloidin stained F-actin stress fibers of MDCK cells observed using confocal microscopy. (A) The stress fibers are clearly observed in the basolateral layer under static conditions (controls). (B) The cells start rounding up and the F-actin becomes partially internalized after 3 h in the spinner flask under stress. (C)The cells start losing contact and F-actin starts disappearing after 6 h of stress. (D)Very few cells remain on the slide after 12 h of stress. (Scale bar = 15 pm).
'*
___
A
U
6 9 12 15 3 Time under stress (h) Figure 5. Decrease in the number of viable cells under stress.
-r
0
protein, thus taking the cell loss into consideration. For comparison purposes, all the calculations were standardized using paired experiments from the same run. Results demonstrated no statistical different between the amounts of ATP per milligram of protein for the controls and the stressed condition, showing that the ATP content of the cells did not change due to externally applied stresses (Figure 6). Flow in Kidney. Epithelial cells in renal tubules are normally subjected to a continuous flow of fluid. A simple calculation made to determine the Reynolds number for flow in human proximal tubules results in a number very much less than one. Since most fluid reabsorption occurs in proximal tubules, the flow rate and velocity decrease as the fluid flows in the distal tubules, where epithelial cells experience flow a t even smaller Reynold number. Thus, MDCK cells, which are good representatives of the distal tubules, have evolved in an environment where they have not seen any turbulence; this might be one of the reasons why they were unable to adapt to the stressed conditions. On the other hand, vascular endothelial cells are subjected to changing levels of hemodynamic shear stress in vivo and are much more shearresistant. Bovine aortic endothelial cells (BAEC), for example, can withstand shear stresses up to 5-10 dyn/ cm2 before showing morphological changes (Mo et al., 1991). This could mean that the origin of cell cultures
Time under stress (h) Figure 6. Normalized ATP/mg of protein values for control and stressed cells. The values show no statistical significant difference among any of these conditions.
may be an important factor to consider when selecting a cell line for large-scale production.
Conclusions The primary objectives were to study the relationship between shear stress and its effect on the cytoskeleton of the cells and to check its effect on the cellular energy metabolism. The MDCK cells were found to be extremely shear sensitive. Under turbulent conditions, the actin cytoskeleton was seen to depolymerize within 12 h a t a mean shear stress of 0.09 dyn/cm2. The cells would then lose adherence to the slides, fall off, and eventually die. The metabolic state of the cells was measured to determine whether the cells became energy-starved in a stressed environment. The ATP content of the cells was found to be practically unchanged, suggesting that the cytoskeleton breakdown was not due to lack of metabolic energy. So, although others found that ATP depletion of cultured cells lead to a breakdown of the actin cytoskeleton (Bacallao et al., 1994; Doctor et al., 1994; Kroshian et al., 1994), the cytoskeleton breakdown due to shear stress seems to occur through other mechanisms. Acknowledgment The authors acknowledge the assistance of Dr. George Truskey for providing the anemometer and for his helpful
Biotechnol, Prog., 1995,Vot. 11, No. 5
600 suggestions. Partial support for this work was provided by North Carolina Biotechnology Center. PAW.and V.D.B. were supported by the National Science Foundation and a Duke University Graduate Fellowship, respectively.
Literature Cited Bacallao, R.; Stelzer, E. H. K. Preservation of biological specimens for observation in a confocal fluorescence microscope and operational principles of confocal flurescence microscopy. Methods Cell Biol. 1989, 31, 437-451. Bacallao, R.; Antony, C.; Dotti, C.; Karsenti, IC; Stelzer, H. K.; Simons, K. The subcellular organization of Madin-Darby canine kidney cells during the formation of a polarized epithelium. J . Cell B i d . 1989, 109, 2817-2832. Bacallao, R.; Garfinkel, A.; Monke, S.; Zampighi, G.; Mandel, L. J. ATP depletion: A novel method to study junctional properties in epithelial tissues. 1. Rearrangement of the actin cytoskeleton. J.Cell. h i . 1994, 107, 3301-3313. Barak, L. S.; Nothnagel, R. E.; DeMarco; Webb, W.W. Differential staining of actin in metaphase spindles with 7-nitrobenz-2-oxa-l,3-diazole-phallacidin and fluorescent DNase: Is actin involved in chromosomal movement? Proc. Natl. Acad. Sei. U S A . 1981, 78, 3034-3038. Bhat, V. D.; Cherry, R. S.; Mandel, L. J. Hydrodynamic shear stress effects on the actin cytoskelton and energy status of cultured epithelial cells. In Industrial Mixing Fundamentals; AICHENAMF Symposium Series, Vol. 91, Tatterson, G. B., Ed.; American Institute of Chemical Engineers: New York, 1995. Davies, P. F.; Remuzzi, A,; Gordon, E.; Dewey, F. C., Jr.; Gimbrone, M. A., Jr. Turbulent fluid shear stress induces vascular endothelial cell turnover in vivo. Proc. Nutl. Acad. Sei. U S A . 1988,83,2114-2117. Doctor, B. R.; Bacallao, R.; Mandel, L. J. Method for recovering ATP content and mitochondrialfunction after chemical anoxia in renal cell culture. Am. J . Physwl.: Cell Physwl. 1994,266, C1803-Cl8ll.
Franke, R. P.; Graefe, M.; Schnittler, H.; Seiffge, D.; Drecnkhahn, D. Induction of human endothelial stress fibers by fluid stress. Nature 1984,307, 648-649. Geremia, J. 0.Experiments in the calibration of flush mounted film sensors. DISA Infi 1972,13,5-10. Ives, C. L.; Eskin, 5.G.; McIntire, L. V. Mechanical effects on endothelial cell morphology: in vitro assessment. In Vitro Cell. Dev. Bwl. 1986,22,500-507. h h i a n , V. M.; Sheridan, A. M.; Lieberthal, W. Functional and cytoekeletal changes induced by sublethal injury in proximal tubular epithelial cells. Am. J . Physiol.: Renal Fluid Elect d y t e Physiol. 1994,266, F21-F30. Levesque, M. J.; Nerem, R. M. The elongation and orientation of cultured endothelial cells in response to shear stress. J . Biomch. EM. 1985,107,341-347. Lundgren, T. 5.; Sparrow, E. M.; Starr, J. B. Pressure drop due to the entrance region in ducts of arbitrary cross section. J . B a i t Eng. 1964,620-626. Midler, M., Jr.; Finn,R. K. A model system for evaluating shear in the design of stirred fermentors Biotechnol Bioeng. 1966, 8, 71-84. Mo, M.; Eskin, E. S.; Schlling, W. P. Flow-induced changes in ea2+signalling of vascular endothelial cells: effect of shear stress and ATP. Am. J . Physwl. 1991,260, H1698-Hl707. Peterson, J. F. Shear stress effeds on cultured hybridoma cells in a rotational Couette viscometer. Ph.D. Thesis, Rice University, Houston, TX, 1989. Stanley, P. E.; Williams, S. G. Use of the liquid scintillation spectrometer for determining Adenosine Triphosphate by the luciferase enzyme. Anal. Biochem. 1969,29, 381-392. Wang, N.; Butler, J. P.; Ingber, D. E. Mechanotransduction across the cell surface and through the cytoskeleton. Science 1993,260, 1124-1127. Accepted March 31, 1995.@
BP950014M @
Abstract published in Advance ACS Abstracts, June 1, 1995.
