Sie sind auf Seite 1von 9

REVIEW OF SCIENTIFIC INSTRUMENTS 77, 104301 2006

Development of a novel bioreactor to apply shear stress and tensile strain simultaneously to cell monolayers
Liam T. Breena and Peter E. McHughb
National Centre for Biomedical Engineering Science, National University of Ireland, Galway, University Road, Galway, Ireland and Department of Mechanical and Biomedical Engineering, National University of Ireland, Galway, University Road, Galway, Ireland

Brendan A. McCormackc and Gordon Muird


School of Engineering, Institute of Technology, Sligo, Ballinode, Sligo, Ireland

Nathan J. Quinlane and Kevin B. Heratyf


National Centre for Biomedical Engineering Science, National University of Ireland, Galway, University Road, Galway, Ireland and Department of Mechanical and Biomedical Engineering, National University of Ireland, Galway, University Road, Galway, Ireland

Bruce P. Murphyg
National Centre for Biomedical Engineering Science, National University of Ireland, Galway, University Road, Galway, Ireland

Received 24 May 2006; accepted 27 August 2006; published online 9 October 2006 To date many bioreactor experiments have investigated the cellular response to isolated in vitro forces. However, in vivo, wall shear stress WSS and tensile hoop strain THS coexist. This article describes the techniques used to build and validate a novel vascular tissue bioreactor, which is capable of applying simultaneous wall shear stress and tensile stretch to multiple cellular substrates. The bioreactor design presented here combines a cone and plate rheometer with exible substrates. Using such a combination, the bioreactor is capable of applying a large range of pulsatile wall shear stress 30 to + 30 dyn/ cm2 and tensile hoop strain 0%12%. The WSS and THS applied to the cellular substrates were validated and calibrated. In particular, curves were produced that related the desired WSS to the bioreactor control parameters. The bioreactor was shown to be biocompatible and noncytotoxic and suitable for cellular mechanical loading studies in physiological condition, i.e., under simultaneous WSS and THS conditions. 2006 American Institute of Physics. DOI: 10.1063/1.2356857

I. INTRODUCTION

Endothelial cells are one of the key components that control the health of a human artery wall, and in vitro assessment of the performance of human endothelial cells ECs is central to understanding the pathophysiology of diseases of the circulatory system. However, in the in vitro environment experiments that investigate the response of ECs to biochemical stimuli are usually conducted in a static environment, whereas in the in vivo environment a number of dynamic mechanical stimuli are present: wall shear stress WSS, tensile hoop strain THS, and pressure. The importance of mechanical stresses was rst highlighted by Fry,1 whereby the author hypothesized that a high shear stress was detrimental to the endotheliums integrity. Since this early report there have been a number of studies that have identia

Electronic mail: liam.breen@nuigalway.ie Electronic mail: peter.mchugh@nuigalway.ie c Electronic mail: mccormack.brendan@itsligo.ie d Electronic mail: muir.gordon@itsligo.ie e Electronic mail: nathan.quinlan@nuigalway.ie f Electronic mail: kevin.heraty@nuigalway.ie g Author to whom correspondence should be addressed; FAX: 353 091 494596; electronic mail: bruce.murphy@nuigalway.ie
b

ed unfavorable hemodynamic conditions that correlate with EC dysfunction and subsequent atherosclerotic lesion formation.214 The evidence in the literature demonstrates that not only is the biochemical environment important in the development of atherosclerosis, but also the mechanical environment is equally as important and necessitates inclusion in any in vitro experimentation.1520 The focus of most EC mechanotransduction studies has been WSS; however, the tensile hoop strain is also known to inuence the performance of endothelial cells. Arterial regions with a sudden increase in cross sectional area and a given internal blood pressure are most susceptible to elevated hoop stress, e.g., artery bifurcations, the internal carotid artery sinus region,21 and the abdominal aorta; all these areas are prone to atherosclerotic lesion formation. If one assumes a constant arterial wall thickness, then these areas of elevated arterial hoop stress will generate regions of elevated arterial tensile hoop strain and elevated stretch applied to the endothelium, leading to potential endothelial damage. A morphological response to cyclic stretch has been observed, whereby ECs elongate and align almost perpendicular to the direction of substrate deformation.2231 The EC actin laments stress bres were also reported to align in bundles parallel to the ECs long axis transverse to the stretch
2006 American Institute of Physics

0034-6748/2006/7710/104301/9/$23.00

77, 104301-1

Downloaded 18 Oct 2006 to 140.203.2.9. Redistribution subject to AIP license or copyright, see http://rsi.aip.org/rsi/copyright.jsp

104301-2

Breen et al.

Rev. Sci. Instrum. 77, 104301 2006

direction.2830 Furthermore Cheng et al.32 found a functional response to mechanical stretch; they showed that EC ICAM-1 expression increased with an increase in strain. These studies indicate that not only does WSS initiate a morphological response but THS can also inuence the function of an EC. This study is concerned with the response to a pulsatile THS as a result of the cardiac cycle, as opposed to any vasoactive contractions or dilations. A number of studies combine both WSS and THS; Benbrahim et al.33 and Peng et al.34 have evaluated the in vitro response of human umbilical vein endothelial cells HUVECs to a combination of both arterial strain and uid shear in a tubular model. Results show that pulsatile stretch and ow produce cellular alignment and actin cytoskeleton rearrangement as observed in stretch alone models35 and WSS morphological studies.36 Qiu and Tarbell37 investigated the EC response to a number of different ow conditions. It was found that an oscillating WSS with no THS was the most harmful ow condition, resulting with the greatest production of a vasoconstrictor ET-1, and that a steady WSS was atherosclerosis protective ow condition, resulting with the greatest production of two vasodilators NO and PGI2. Zarins et al.4 and Zhao et al.38 demonstrated in a nite element model that regions in the vascular system with low WSS, ow separation, ow reversal, and high THS correlated with in vivo areas where atherosclerotic plaques and intimal thickening occurs in the carotid arteries. Numerous studies have reported important evidence of EC dysfunction, as a result of individual in vitro forces involving WSS Refs. 3,3942 or THS.23,26,27,2931,35,4345 However, applying isolated hemodynamic stresses will not accurately mimic the in vivo vascular environment. An experimental combination of both pulsatile WSS and THS would create a better physiological approximation of the in vivo mechanical environment. The aim of this study is to design and validate a bioreactor that is capable of applying both pulsatile WSS and pulsatile THS of physiological proportions and that will allow much greater exibility in the range of combinations of WSS and THS that can be applied, in comparison to tubular devices previously used.33,34
II. BIOREACTOR DESIGN

FIG. 1. A cross sectional view of a cone and plate rheometer. The cone rotates about the Z axis and applies a WSS to the plate surface cultured ECs via the rotating uid between the cone and plate.

stress being applied to the upper surface of the seeded cells. Using the Navier-Stokes equations assuming that the endothelial growth media EGM-2 Ref. 58 are Newtonian, it is possible to determine an expression for the plate surface WSS,47 assuming the cone apex touches the plate surface, i.e., H = h in Fig. 1, as follows:

V vz = , z H

where is the WSS, is the dynamic viscosity, vz is the tangential velocity distribution in the Z direction at a given radial position r, and V and H are the cone linear velocity and height, respectively, at a radial distance r. Written in terms of rotational quantities:

The bioreactor achieves its design goals by combining two previous bioreactor techniques: a cone and plate rheometer40 and multiple exible silicone substrates.46 In summary, the cone and plate rheometer controls the applied WSS, and the applied THS is controlled by the displacement applied to silicone substrates upon which the cells are seeded. The following sections describe the application of mechanical stresses to a cell seeded substrate and the setup and operation of the bioreactor.
A. Application of WSS to cultured cells

where is the cone angular velocity and is the angle between the cone and plate surfaces. In order to achieve an accurate desired WSS, uniformly applied to the test regions, the cone must be positioned precisely in the vertical direction. The ideal cone position would be achieved when the gap distance GD is zero see Fig. 1. In practice, however, in order to guarantee that the WSS applied to the plates surface in the bioreactor has an accuracy of 95% or greater, the GD must be less than or equal to a gap tolerance GT. Modied versions of Eqs. 1 and 2 can be used to determine the gap tolerance. In Eq. 1 H is made equal to the actual cone height h at a radial position r written in terms of r and the cone angle plus the GT. To achieve a WSS accuracy of 5%, equating modied versions of 1 and 2 generates the following equation that can be solved for GT: 0.95 V . = r tan + GT 3

A cone and plate rheometer is used to apply the WSS to the cells; in this situation, a cone rotates in a bath of media and below the cone is a plate upon which cells are grown see Fig. 1. When the cone rotates, the media is set in motion and the ow of media over the plate results in a shear

This method of determining the appropriate gap tolerance has been previously proposed by OKeefe et al..48 For the specications of the current bioreactor cone angle is 1.0 0.017 45 rad and the mean cellular sample radial position r is 35 mm, the maximum gap tolerance is determined to be 32.15 m. To achieve this accurate cone placement, a coordinate measuring system Eley Metrology Ltd., UK with an accuracy of 0.5 m and a specically machined gap

Downloaded 18 Oct 2006 to 140.203.2.9. Redistribution subject to AIP license or copyright, see http://rsi.aip.org/rsi/copyright.jsp

104301-3

Novel shear and tensile stress bioreactor

Rev. Sci. Instrum. 77, 104301 2006 TABLE I. Mechanical stimulus applied to each substrate see Fig. 2. Mechanical stimulus Substrate number 1 and 5 2 and 6 3 and 7 4 and 8 9 and 10 WSS dyn/ cm2 030 030 030 030 0 THS % 011.56 08.02 04.07 0 0

FIG. 2. A schematic of the ten cellular substrates. Eight of the substrates 18 are exposed to WSS and a range of THS. The six double sided arrows indicate the substrate stretch directions. Two control substrates 9 and 10 are contained in two compartments in the main container of the bioreactor submerged in the same media as all the other cellular samples, but substrates 9 and 10 are subjected to no mechanical stimuli.

gauge is used for cone positioning. For practical purposes the cone tip was machined at with a diameter of 20 mm. The cone apex was machined at to prevent rubbing between the cone and plate surface. This is common in the design and construction of cone and plate rheometers.49

B. Application of tensile strain

In the bioreactor, the cells are not seeded simply on the plates surface, as they would be in a standard cone and plate rheometer. In this case, the cells are seeded within a 14 14 mm2 cellular region on each exible substrate, which is centered in each well 16 16 mm2 square hole in the thin rheometer plate see Fig. 2. The bioreactor was designed to incorporate ten of these silicone substrates. Six of these substrates are subjected to varying levels of THS and constant WSS while two substrates are subjected to WSS with no THS, and the nal two substrates are not subjected to any mechanical stimulus and act as control samples see Fig. 2 for a schematic of the substrate locations. The range of mechanical strains is applied to the exible substrates by xing the inner end of each substrate with a central clamping device beneath the thin rheometer plate and attaching the outer end free end of each substrate to one of the six movable stainless steel clamps, which apply a tensile stretch via a cable and pulley system. The substrates clamped beneath the rheometer plate are submerged in cell media during testing, which results in a negligible frictional force between the plate and the substrate surfaces. A programmable stepper motor connected to the cable system controls the displacements applied to the silicone substrates. Six of the ten substrates are connected to the stepper motor and are axially stretched within a realistic range of physiological THS, similar to previous studies.23,26,27,29,30,32,43,44

C. Simultaneous application of WSS and THS

In this bioreactor design, both the WSS and THS are applied simultaneously to monolayers of cells. Careful consideration was required when applying the WSS to the cells within the rheometer, as they are not seeded directly on the plates surface. In this design, the cellular substrates are movable exible silicone substrates, and a modication had to be made to the rheometers design to include a tensile stretch of the substrates. In this case, the substrates are lo-

cated 100 m below the rheometer plate surface and exposed to the rotating uid by windows that were laser cut into the plate, as shown in Fig. 2. The motion of the cone and the displacement of the silicone substrates are controlled by separate stepper motors. The rst stepper motor is mounted directly to the cone axle, and the angular velocity of the motor shaft is equivalent to the angular velocity of the cone. The second stepper motor controls the strain magnitude applied to the set of six cellular substrates. This motor is attached to a pulley and cable system. This cable and pulley system can apply three different magnitudes of strain to the set of six different substrates. Substrates that are opposite pairs are stretched at the same strain magnitudes, i.e., high strain applied to substrates 1 and 5, medium strain applied to substrates 2 and 6, and nally low strain applied to substrates 3 and 7, as shown in Table I and Fig. 2. The pulley and cable system incorporates a three tier pulley that is xed to the shaft of the second stepper motor, as shown in Fig. 3. For a given rotation of the second stepper motor, the three tier pulley turns through the same angle of rotation and the cables attached to each level of the three tier pulley are linearly displaced, i.e., the two cables attached to the top level smallest shaft diameter of the three tier pulley are pulled by the smallest displacement and the two cables attached to the bottom level largest diameter of the three tier pulley are pulled by the largest displacement. The variation in the tier diameters predetermines the range of linear displacement magnitudes achieved by the cables. The cable tension is controlled by a turnbuckle; in this case six stainless steel turnbuckles were attached in line on each of the exible substrates cable system, beneath the main container, as shown in Fig. 3. The strain applied to each substrate can be ne tuned by the turnbuckle tension. A turnbuckle support was designed and built to prevent any cable sagging due to the additional weight of a metallic turnbuckle on each cable, it was made from polytetrauoro ethylene PTFE Teon to take advantage of its material properties: biocompatibility, autoclavablity, and low surface coefcient of friction. The turnbuckle support component was machined in two halves that lock together via four alignment pins and corresponding holes, which allows for easy removal and installation. During testing all ten substrates including two control samples are submerged in a common pool of media typically 300 ml of EGM-2 cell media56 and a retaining ring 316L stainless steel is bolted in position over the perimeter

Downloaded 18 Oct 2006 to 140.203.2.9. Redistribution subject to AIP license or copyright, see http://rsi.aip.org/rsi/copyright.jsp

104301-4

Breen et al.

Rev. Sci. Instrum. 77, 104301 2006 TABLE II. A list of materials used in the bioreactor. All the materials are inert and nontoxic. Bioreactor components Frame Central clamping device Pulley and cable system, substrate radial clamps, and inline turnbuckles Thin plate, plate retaining ring, and plate supporting device Medium container Turnbuckle support piece and ller device Bioreactor cover Cone Cellular substrates Material used 316L stainless steel, 316L stainless steel 316L stainless steel

316L stainless steel PTFE Teon PTFE Teon Polycarbonate Polycarbonate Silicone elastomer lm

57 which generates wave signals. These signals are amplied by two motor drives Stebon Ltd. UK and the amplied signals are sent to the corresponding stepper motors Stebon Ltd. UK. This process enables the user to program a fully adjustable input data array that generates the desired WSS and THS wave forms.
F. Substrate and bioreactor uid preparation

FIG. 3. A A schematic of the bioreactor design. A cable and pulley system controls the strain applied to each of the six exible silicone substrates. The variance in the pulleys tier diameters creates three different magnitudes of cable linear displacement and cellular substrate strain magnitude. B A picture of the bioreactor placed inside an incubator. The incubator controls the testing temperature, humidity, and atmospheres CO2 levels.

of the thin plate, which holds the plate in place, as shown in Fig. 3.
D. Bioreactor materials and environment control

The entire bioreactor is made from inert and nontoxic materials see Table II. Both stepper motors are enclosed by polycarbonate boxes to prevent any potential contamination risks caused by the motors internal components. The assembled bioreactor was designed to t inside a standard incubator, ensuring a controlled environment with a 37 C temperature and 5% CO2 in air.
E. Stepper motor motion control

In developing the bioreactor it was considered important to have a medium viscosity similar to that of adult human blood 0.003 N s / m2.50 For the EGM-2 media this was achieved by adding sterile dextran powder Sigma Aldrich Inc. Ireland, similar to the methods employed by Blackman et al.39 The viscosities of various concentrations of dextran and medium solutions were measured using Poiseuilles law of ow and a vertical Ostwald U-bend viscometer Poulten Selfe & Lee Ltd. UK. The densities of the medium and dextran solutions were calculated before testing using a laboratory balance and a known volume of solution at 37 C, and all solution weighting and mixing were conducted in a sterile laminar ow hood. Experiments showed that a 2.586% w/v dextran and endothelial general medium EGM-2 solution was required to achieve the desired dynamic viscosity of 0.003 N s / m2 3.0 cP. The dynamic viscosity of the dextran and EGM-2 solution was measured three times for each dextran concentration. In this experiment, the Oswald viscometer was placed vertically in a water bath, which was maintained at 37 C.
G. Substrate preparation

A custom LABVIEW program version 6.1i was written to control the input wave forms for the two stepper motors. The program was split into two sections: the rst section generates a velocity versus time wave form, controlling the rst stepper motor motion ECs WSS wave form, and the second section generates a motor angular position versus time wave form, controlling the second stepper motor ECs THS waveform. The program controls the stepper motors via a control card National Instruments PCI-7342 Ref.

The cellular substrates were cut from silicone Goodfellow Cambridge Limited sheets 100 100 0.1 mm3; each substrate was cut into 65 20 mm2 strips. The substrates were prepared in the following way: rstly cleaned with 70% ethanol and then placed in beaker of distilled water in the ultrasonic bath Bransonic for 15 min. They were subsequently rinsed with distilled water and allowed to dry in a laboratory oven at 50 C for 30 min. The silicone substrates were wet etched with 70% sulfuric acid to improve the silicone surface hydrophilicity. An Alamar Blue assay found

Downloaded 18 Oct 2006 to 140.203.2.9. Redistribution subject to AIP license or copyright, see http://rsi.aip.org/rsi/copyright.jsp

104301-5

Novel shear and tensile stress bioreactor

Rev. Sci. Instrum. 77, 104301 2006

that 2 min etching resulted in a substrate that was mechanically sound and allowed maximum cellular adhesion results not shown. Finally, a bronectin coating was used to further increase cellular attachment. This method of coating silicone was also optimized using an Alamar Blue assay, and the optimal concentration was found to be 15.36 g / ml 6 g / cm2, similar to that of Clark et al.43
III. BIOREACTOR PERFORMANCE ASSESSMENT

As regards the analysis of the performance of the bioreactor, the cell toxicity and magnitude of THS applied were directly validated by experimental measurement. In addition, computational uid dynamics modeling was used to determine the ow patterns over the indents and to calibrate the bioreactor to produce a desired range of WSS values.
A. Substrate strain validation

FIG. 4. Schematic diagram of the uid region present between the plate and cone surfaces. The cone has a attened tip with a diameter of 20 mm. The volume also includes the two sets of four indents; a set at radial position R1 35 mm from the plate center point and the second set placed at a radial position R2 25 mm from the center point. Model B, with the indent at R1, and model C, with the indent at R2, are 45 slices of the entire model.

r 2 2 R= , 12

The strain applied to each of the silicone substrates was measured using a video-extensometer system Messphysik Material Testing Video-Extensometer. Two reference markings black lines were placed on each substrate. The camera then recorded the displacement of these reference markings. The software converts this displacement into strain based on an initial inputted reference distance, with an accuracy of 0.01% strain. A sawtooth strain waveform, with a dwell period at each peak and trough, was used to calibrate each substrate. This wave form oscillated the second stepper motor from 0 to 40.04 0455 steps, 4000 steps per revolution at a frequency of 0.1 Hz. A 5 s dwell period provided sufcient time to acquire a substrate strain reading. The substrate strains were ne tuned and calibrated by adjusting the appropriate in-line cable turnbuckles. This procedure resulted in the nal range of strains applied to the individual substrates, as shown in Table I to a strain magnitude error of 0.01%.
B. Determination of applied WSS

where is the kinematic viscosity. Notwithstanding the aforementioned point about the indent disturbing the ow, this formula was used to make a simple estimate of an upper limit angular velocity for the bioreactor to ensure that the ow remained laminar. R values were analytically calculated for a variety of different cone angular velocities and plate radial positions. The uid was given a dynamic viscosity of 0.003 N s / m2, a density of 1057 kg/ m3, and a 1 cone angle. Using the maximum bioreactor plate radius of 65 mm, a maximum possible ow remaining laminar, R 1 cone angular velocity is calculated to be approximately 250 rpm. This value of 250 rpm is taken as the maximum allowable value that can be used during any experiments.
2. Computational uid dynamics calibration
CFD Ansys-CFX Inc., Canonsburg, PA models were used to calibrate the WSS generated within the bioreactor. The geometry of the models was reduced by using axisymmetry about the cones central axis, reducing the initial geometry to three smaller pizzalike 45 slice sector models, as shown in Fig. 4. Model B is a slice model with a square indent 16 16 0.1 mm3 deep at a distance of 35 mm R1 = 35 mm from plate center point, and model C is a slice model with a similar indent at a distance of 25 mm R2 = 25 mm from the plate center point. Model A is a slice model similar to model B without an indent. Models B and C were further reduced by applying two curved circumferential free slip surfaces at a distance of 10 mm from the corresponding edge of the indent, as shown in Fig. 5. In model A, the free slip walls were placed in the same positions as in model B. The free slip surfaces were assumed to have a negligible effect on the WSS applied to the region of interest. In all three models, the regions adjacent to the plate and indent surfaces were meshed with ne prismatic elements and the remaining volume was meshed with tetrahedral elements. CFXs ination boundary meshing tool was used to efciently mesh the volume. This is done by using a ne mesh rst layer height of 0.07 mm in the regions of interest and a coarser mesh for the less critical regions. The ination boundary parameters ve layers, 1.2

The bioreactor plate has eight square shallow indents 16 16 0.1 mm3 where silicone substrates are positioned with cells seeded upon them. These shallow indents disturb the ow, which means that the use of a simple cone and plate analytical solution to determine the shear stress over the cells is not sufcient. A computational uid dynamics CFD, Ansys-CFX version 5 Ref. 58 model was used to calibrate the resultant WSS in this altered rheometer geometry. The CFD WSS results were compared to the expected analytical solution results for simpler geometries cone and plate without indents. After conrmation that the CFD generated accurate results for simpler geometries, the CFD was used to calculate the WSS in more complex models with indents.
1. Reynolds number

The desired ow state in the bioreactor is laminar. Laminar to turbulent transition is indicated by the Reynolds numR 1 is laminar and R 1 is turbulent.51 Assuming ber R: an idealized cone and plate conguration, the Reynolds number is given as

Downloaded 18 Oct 2006 to 140.203.2.9. Redistribution subject to AIP license or copyright, see http://rsi.aip.org/rsi/copyright.jsp

104301-6

Breen et al.

Rev. Sci. Instrum. 77, 104301 2006

FIG. 5. This image shows the control volume the region in which the ow is modeled to mesh all the CFD models. This control volume is represented as an ination boundary layer. In all three models, the regions adjacent to the plate and indent surfaces were meshed with ne prismatic elements and the remaining volume was meshed with tetrahedral elements. An ination boundary ve layers, 1.2 expansion factor, and a rst layer height of 0.07 mm was used to generate this mesh. Periodic boundary conditions axisymmetry about the Y axis were also applied to the inlet and outlet surfaces.

FIG. 6. a The computed WSS present on the plate and inner plane 16 16 mm2 square of model B. The cone surface was given a constant angular velocity of 100 rpm about the Y axis. By reducing the observation region 14 14 mm2, the ow disturbances edge effects have been removed and the WSS is more consistent throughout the remaining area, as shown in b.

expansion factor determine the gradual increase in the ne mesh size, which interconnects the ne and coarse mesh regions. Approximately 135 000 tetrahedral ne mesh elements and prismatic elements coarse mesh elements were used to mesh each model. Periodic boundary conditions about the Y axis of the computational model were applied to the inlet and outlet surfaces of all models, as shown in Fig. 5. The plate surface and all the indent surfaces were given noslip boundary conditions. The cone surface top surface was given a no-slip boundary condition with a constant angular velocity about the Y axis see Fig. 5. The ow was assumed laminar. The uid used in the CFD analysis was given the following properties: density of 1057 kg/ m3 and a dynamic viscosity of 0.003 N s / m2, which are representative of the uid used in the bioreactor. The uid was assumed to be Newtonian, homogeneous, and incompressible.50,52 In large arteries where high shear rates occur, a Newtonian approximation is acceptable.52
3. CFD calibration results and master curve correlation

All three models were solved for cone angular velocities of 10, 20, 40 60, 80, 100, and 200 rpm. In all three cases, an edge effect are area of nonuniform WSS was observed around the free slip curved walls. The edge effects created were small and localized see Fig. 6 for the example of case B. The CFD solutions for models B and C showed minimal edge effects around the fringes of each indent. The edge effects within the indent regions were avoided by reducing the region 16 16 mm2 of observation for both models B and C to a smaller square 14 14 mm2 within the indent as shown in Fig. 6. Based on this, in the operation of the bioreactor, cells within a 14 14 mm2 square region only are harvested for analysis. The CFD simulations show a relatively consistent WSS within this inner plane 14 14 mm2 cellular region, where 93% of the surface area is exposed to a WSS within a 10% tolerance of the desired area average WSS. The results are based on CFD simulation of model B geometry with a constant cone angular velocity of 10 rpm.

The area average WSS was recorded for the plate surface for the CFD model A model B excluding an indent for angular velocities of 10, 20, 40, 60, 80, 100, and 200 rpm. The results for the area average WSS on the plate surface of model A compare favorably with the calculated analytical solution see Fig. 7. This correlation between the computational and the analytical solutions conrmed that the computational models were producing accurate results. This result generates condence in the accuracy of the results for the more complex models that predicted the WSS on the substrate surface. The CFD area average WSS solutions for the indent inner plane surface for both model B and model C are also shown in Fig. 7. It is clear that the WSS values for both models are very similar and signicantly less than the WSS on the entire plate surface of model A. This decrease in area average WSS is due to the increase in the distance between the cone surface and indent surface as would be suggested by Eq. 1. The WSS versus angular velocity curves, as shown in Fig. 7 for both indented geometries, are sufciently close so that for all practical purposes they can be represented by a single curve that can be referred to as the master curve. Using this curve, the bioreactor user can calculate the angular velocity required to achieve a desired WSS indent inner plane WSS. The difference between the analytical solution curve and the master curve veries that an analytical WSS calibration alone could not have been used

FIG. 7. Plot of WSS values vs angular velocity for model A plate surface WSS, the indent inner plane area averaged WSS for both models B and C, and the analytical WSS solution for a range of cone angular velocities.

Downloaded 18 Oct 2006 to 140.203.2.9. Redistribution subject to AIP license or copyright, see http://rsi.aip.org/rsi/copyright.jsp

104301-7

Novel shear and tensile stress bioreactor

Rev. Sci. Instrum. 77, 104301 2006 TABLE IV. Measurement of growth medium pH levels during the strain experiment. Time in bioreactor h 0 3 18

TABLE III. Alamar Blue assay results for cellular samples placed in the bioreactor for time periods up to 24 h; the data are presented as a mean percentage of three tests and the standard deviation of these tests. Cytotoxicity testing Viable cells % of the total number of seeded cells a 83 63 90 94 78

Description Initial pH 0 h in bioreactor pH after 3 h of cellular strain experiment pH after cell in bioreactor for 12 h no strain test

pH value 7.81 7.87 7.67

Time in bioreactor h 0 2 4 22 24
a

Standard derivation % 2 9 3 4 6

Based on a seeding density of 80 000 cells/ cm2.

for this particular bioreactor. An approach involving a computational solution complemented by an analytical solution had to be used to comprehensively calibrate the bioreactor. For example, to achieve a cellular WSS of 10 dyn/ cm2 1 Pa, analytically a 56 rpm cone angular velocity would have been sufcient, but it has been shown by the computational study that, in fact, a higher 72 rpm cone angular velocity is needed see dark dotted lines on Fig. 7.
C. Bioreactor cellular toxicity validation study

Fifteen circular silicone substrate samples surface area of 2 cm2 were prepared and sterilized, as previously described in Sec. II G. ECs were then seeded on the samples at a concentration of 80 000 cells/ cm2. The cells were allowed to attach for 12 h and were then placed in the bioreactor lled with fresh EGM-2 for up to 24 h. The bioreactor was placed in an incubator and three samples were removed at different time intervals. An Alamar Blue assay was used to quantify cell viability at each time interval. The assay results showed that there was no signicant cell death after 24 h see Table III. At each time step, a substrate was also stained using an Hematoxylin and eosin H&E staining assay, which highlights the nucleus and the membrane of the cells.

This staining assay was preformed to investigate the cell conuence after each incubation period. An image of a H&E stained cellular substrate, which was incubated in the bioreactor for 24 h, is shown in Fig. 8. A cellular stretch test was carried out to investigate if the bioreactor would affect the EGM-2 pH level. The pH of the cell media was measured before the cells were placed in the bioreactor, after the cells were in the bioreactor for 12 h with no mechanical testing, and nally a pH measurement was taken after 3 h of cell stretching, as shown in Table IV. The pH readings taken were within 2.6% of each other, indicating that the materials used by the bioreactor are not signicantly affecting the cellular medias pH level and that the bioreactor tests are carried out in a pH balanced environment. A bioreactor test was preformed to investigate cell response to a combination of WSS and THS. A low steady WSS 4 dyn/ cm2 and a range of cyclical THS 0%12% strain at frequency of 1 Hz were applied simultaneously to a number of cellular substrates for three hours. After testing, the cells were xed and stained with a H&E stain. The cells were observed to elongate and align perpendicular to the stretch and parallel to the ow illustrated in Fig. 9, which is a characteristic response to this applied mechanical environment.
IV. DISCUSSION

For the past three decades, many studies have investigated the effects of in vitro forces applied to cultured ECs in an attempt to mimic the in vivo forces WSS, THS, and pressure present in the human vasculature and to understand how these biomechanical forces alter the cellular response, i.e., cell morphology or protein synthesis.7,12 Many of these studies have investigated the cellular response to an isolated force,39,53 but in vivo a number of hemodynamic forces co-

FIG. 8. Color online This image shows a monolayer of HUVECs after 24 h in the bioreactor seeded on a bronectin coated silicone substrate. The cells were H&E stained and photographed at 4 magnication.

FIG. 9. Color online Three images of a monolayer of HUVECs, which were grown on bronectin coated silicone substrates. The cells were exposed to a low steady WSS 4 dyn/ cm2 and cyclical THS 012%. The cells were H&E stained and photographed at three different magnications 4, 10, and 20. Note that the white and black arrows indicate the THS and WSS directions, respectively.

Downloaded 18 Oct 2006 to 140.203.2.9. Redistribution subject to AIP license or copyright, see http://rsi.aip.org/rsi/copyright.jsp

104301-8

Breen et al.

Rev. Sci. Instrum. 77, 104301 2006

exist. Such studies provide valuable information regarding the vascular cell response to isolated stimulus, but not an accurate cellular response similar to the in vivo situation. The bioreactor described in this article is capable of simultaneously applying both wall shear stress 30 to + 30 dyn/ cm2 and tensile hoop stress 0%12% to multiple cellular samples, creating a testing environment that is much more physiological. This novel bioreactor design combines a cone and plate rheometer with multiple exible substrates in order to achieve the desired multistimulus in an in vitro environment. The stretching mechanism, which applies the range of cellular substrate strains, was validated using a video extensometer, and the materials used in the bioreactor were shown to be nontoxic and biocompatible using cell viability assays. Both an analytical approach and computational simulations were used to calibrate the WSS generated by the rheometer. The disadvantage of the analytical approach is that it ignores the ow disturbances created by the irregular rheometer geometry, i.e., the shallow indent 0.1 mm in the rheometers plate surface. The computational simulations represent a more realistic ow and include the ow disturbing shallow indents. From these computational simulations, a master curve was deduced that allows calibration of the bioreactor to generate the desired WSS, as a function of the cone angular velocity. The design of the bioreactor presented here focuses on the combination of WSS and THS in a controlled environment. A potential operating disadvantage is that the device does not allow for independent control of pressure; however, on the other hand, this drawback could be considered as an advantage of the bioreactor, as it allows the user to study the cellular response to WSS and THS isolated from a pulsatile pressure which is not permitted in tubular experimental models. Furthermore, certain in vivo observations suggest that pressure is a less signicant in vivo vascular stimulus,54 whereas THS and WSS play a critical role in vascular homeostasis and pathophysiology, e.g., WSS is an acute modulator of vascular tone.38 This bioreactor design has a number of advantages over tubular experimental models, whereby it can easily apply a wide range of combinations of WSS and THS that would be very difcult to reproduce using a tubular system bioreactor. In the in vivo situation, diverse combinations of hemodynamic forces, e.g., hypertension with quasistatic ow, low THS with a large WSS, and WSS out of phase with THS, can exist because of human irregular shaped vasculature, e.g., arterial bifurcation and acute curvature; all these forces combinations can be easily represented in the bioreactor design. In addition, when using tubular constructs, the users most overcome signicant difculties related to seeding, staining, and examination of cells on tubular substrates. In the bioreactor, all ten cellular samples are submerged in a common pool of cell media. The potential for cross migration of biochemical signals could inuence the protein synthesis of neighboring cellular samples. However, Morigi et al.55 have shown that ECs grown in static conditions with either mechanically preconditioned cell media media taken from a prior mechanical EC test that up regulated ICAM-1

and VCAM-1 or fresh cell media expressed equivalent levels of surface protein synthesis, i.e., ICAM-1 and VCAM-1 expressions. This conclusion shows that the secreted biochemical signals in the mechanically precondition cell media do not affect the level of ICAM-1 or VCAM-1 expression. The results of these studies allowed us to make the assumption that any cellular protein synthesis during bioreactor testing would be a direct result of the mechanical stimulus, exclusively, and are uninuenced by any biochemical signals that may be secreted into the common pool of cell media. Using the current system, after testing, the cellular substrate samples can be examined for signs of cell surface protein synthesis, for example, immunostaining and ow cytometry could be used to quantify specic cellular protein synthesis ICAM-1 and VCAM-1, proteins known to be abnormally regulated during the initiation of vascular disease formation.10 In this article, the bioreactor substrate strains were validated for a sawtooth prole and calibrated for a range of laminar steady WSS conditions. However, in order to mimic realistic physiological hemodynamic forces, the bioreactor must be validated and calibrated for more complex in vitro mechanical stimuli, such as pulsatile and oscillating WSS and THS waveforms. These atheroprone and atheroprotective mechanical waveforms have been reported to be found in the sinus region and the distal region of the carotid artery, respectively.5 This is the topic of ongoing work by the authors. In summary, the bioreactor described above is a laboratory based device that can apply mechanical stimuli to numerous cells or tissue samples in a controlled environment. The device was validated and calibrated to ensure that the desired mechanical stimuli are applied to the samples, and after testing, those samples can be removed from the bioreactor to be examined for signs of cellular dysfunction, which are known to lead to atherosclerosis and vascular disease.
ACKNOWLEDGMENTS

The authors acknowledge support from the Program for Research in Third Level Institutions PRTLI, administered by the Higher Education Authority HEA. The project was carried out at the National Centre for Biomedical Engineering Science NCBES, National University of Ireland, Galway, in association with University of Limerick and Institute of Technology, Sligo. Two of the authors L.T.B. and K.B.H. acknowledge funding from the Irish Council for Science, Engineering and Technology IRCSET under the Embark Initiative Postgraduate Research Scholarship Scheme.
D. L. Fry, Circ. Res. 24, 93 1969. A. M. Malek, S. L. Alper, and S. Izumo, JAMA, J. Am. Med. Assoc. 282, 2035 1999. 3 C. F. Dewey, Jr., S. R. Bussolari, M. A. Gimbrone, Jr., and P. F. Davies, J. Biomech. Eng. 103, 177 1981. 4 C. K. Zarins, D. P. Giddens, B. K. Bharadvaj, V. S. Sottiurai, R. F. Mabon, and S. Glagov, Circ. Res. 53, 502 1983. 5 G. Dai, et al., Proc. Natl. Acad. Sci. U.S.A. 41, 14871 2004. 6 A. Barakat and D. Lieu, Cell Biochem. Biophys. 38, 323 2003. 7 J. J. Chiu et al., Arterioscler., Thromb., Vasc. Biol. 24, 73 2004. 8 D. N. Ku, D. P. Giddens, C. K. Zarins, and S. Glagov, Arteriosclerosis Dallas 5, 293 1985.
1 2

Downloaded 18 Oct 2006 to 140.203.2.9. Redistribution subject to AIP license or copyright, see http://rsi.aip.org/rsi/copyright.jsp

104301-9
9

Novel shear and tensile stress bioreactor

Rev. Sci. Instrum. 77, 104301 2006 D. A. Kass, Am. J. Physiol.: Cell Physiol. 279, C797 2000. M. Moretti, A. Prina-Mello, A. J. Reid, V. Barron, and P. J. Pendergast, J. Mater. Sci.: Mater. Med. 15, 1159 2004. 36 P. F. Davies, A. Remuzzi, E. J. Gordon, C. F. Dewey, Jr., and M. A. Gimbrone, Jr., Proc. Natl. Acad. Sci. U.S.A. 83, 2114 1986. 37 Y. Qiu and J. M. Tarbell, J. Vasc. Res. 37, 147 2000. 38 S. Z. Zhao, B. Ariff, Q. Long, A. D. Hughes, S. A. Thorn, A. V. Stanton, and X. Y. Xu, J. Biomech. 35, 1367 2002. 39 B. R. Blackman, G. Garcia-Cardena, and M. A. Gimbrone, Jr., J. Biomech. Eng. 124, 397 2002. 40 S. R. Bussolari, C. F. Dewey, Jr., and M. A. Gimbrone, Jr., Rev. Sci. Instrum. 53, 1851 1982. 41 M. Sato, N. Ohshima, and R. M. Nerem, J. Biomech. 29, 461 1996. 42 B. R. Blackman, K. A. Barbee, and L. E. Thibault, Ann. Biomed. Eng. 28, 363 2000. 43 C. B. Clark, T. J. Burkholder, and J. A. Frangos, Rev. Sci. Instrum. 72, 2415 2001. 44 P. C. Dartsch and E. Betz, Basic Res. Cardiol. 84, 268 1989. 45 J. H. Wang, G. Yang, Z. Li, and W. Shen, J. Biomech. 37, 573 2004. 46 D. L. Wang, C. C. Tang, B. S. Wung, H. H. Chen, M. S. Hung, and J. J. Wang, Biochem. Biophys. Res. Commun. 195, 1050 1993. 47 F. M. White, Fluid Mechanics Mc-Graw Hill, New York, 1994. 48 L. M. OKeefe and T. McGloughlin, The Proceedings of the Eleventh Annual Conference of the Section of Bioengineering of the Royal Academy of Medicine in Ireland, 2005. 49 Description of cone and rheometer design, www.can-am.net/suppliers/ Brookeld.htm 50 S. Tada, and J. M. Tarbell, Ann. Biomed. Eng. 33, 1202 2005. 51 M. H. Buschmann, P. Dieterich, N. A. Adams, and H. J. Schnittler, Biotechnol. Bioeng. 89, 493 2005. 52 S. Z. Zhao, X. Y. Xu, A. D. Hughes, S. A. Thom, A. V. Stanton, B. Ariff, and Q. Long, J. Biomech. 33, 975 2000. 53 A. R. Brooks, P. I. Lelkes, and G. M. Rubanyi, Physiol. Genomics 9, 27 2002. 54 B. R. Kwak, P. Silacci, N. Stergiopulos, D. Hayoz, and P. Meda, Cell Adhes Commun. 12, 261 2005. 55 M. Morigi et al., Blood 85, 1696 1995. 56 Endothelial Growth Media-2, www.combrex.com 57 National Instruments, www.ni.com 58 Ansys-CFX Version 5, www.ansys.com
35

J. N. Topper and M. A. Gimbrone, Jr., Mol. Med. Today 5, 40 1999. P. F. Davies, Physiol. Rev. 75, 519 1995. S. Chien, Prog. Biophys. Mol. Biol. 83, 131 2003. 12 D. C. Chappell, S. E. Varner, R. M. Nerem, R. M. Medford, and R. W. Alexander, Circ. Res. 82, 532 1998. 13 R. M. Nerem, J. Biomech. Eng. 114, 274 1992. 14 C. K. Glass and J. L. Witztum, Cell 104, 503 2001. 15 C. G. Caro, J. M. Fitz-Gerald, and R. C. Schroter, Proc. R. Soc. London, Ser. B 177, 109 1971. 16 T. Asakura and T. Karino, Circ. Res. 66, 1045 1990. 17 M. Motomiya and T. Karino, Stroke 15, 50 1984. 18 B. K. Bharadvaj, R. F. Mabon, and D. P. Giddens, J. Biomech. 15, 349 1982. 19 S. Glagov, C. Zarins, D. P. Giddens, and D. N. Ku, Arch. Pathol. Lab Med. 112, 1018 1988. 20 X. Bao, C. Lu, and J. A. Frangos, Arterioscler., Thromb., Vasc. Biol. 19, 996 1999. 21 S. Tada and J. M. Tarbell, Ann. Biomed. Eng. 33, 1202 2005. 22 K. Kanda and T. Matsuda, Cell Transplant 2, 475 1993. 23 H. Wang, W. Ip, R. Boissy, and E. S. Grood, J. Biomech. 28, 1543 1995. 24 R. C. Buck, Exp. Cell Res. 127, 470 1980. 25 T. Takemasa, T. Yamaguchi, Y. Yamamoto, K. Sugimoto, and K. Yamashita, Eur. J. Cell Biol. 77, 91 1998. 26 C. Neidlinger-Wilke, E. S. Grood, J.-C. Wang, R. A. Brand, and L. Claes, J. Orthop. Res. 19, 286 2001. 27 K. Naruse, T. Yamada, and M. Sokabe, Am. J. Physiol. Heart Circ. Physiol. 43, H1532 1998. 28 V. P. Shirinsky, A. S. Antonov, K. G. Birukov, A. V. Sobolevsky, Y. A. Romanov, N. V. Kabaeva, G. N. Antonova, and V. N. Smirnov, J. Cell Biol. 109, 331 1989. 29 J. H. Wang, P. Goldschmidt-Clermont, and F. C. Yin, Ann. Biomed. Eng. 28, 1165 2000. 30 J. H. Wang, P. Goldschmidt-Clermont, J. Willie, and F. C. Yin, J. Biomech. 34, 1563 2001. 31 P. Sipkema, P. J. van der Linden, N. Westerhof, and F. C. Yin, J. Biomech. 36, 653 2003. 32 J. J. Cheng, B. S. Wung, Y. J. Chao, and D. L. Wang, Hypertension 28, 386 1996. 33 A. Benbrahim, G. J. LItalien, C. J. Kwolek, M. J. Petersen, B. Milinazzo, J. P. Gertler, W. M. Abbott, and R. W. Orkin, J. Surg. Res. 65, 119 1996. 34 X. Peng, F. A. Recchia, B. J. Byrne, I. S. Wittstein, R. C. Ziegelstein, and
10 11

Downloaded 18 Oct 2006 to 140.203.2.9. Redistribution subject to AIP license or copyright, see http://rsi.aip.org/rsi/copyright.jsp

Das könnte Ihnen auch gefallen