In Vitro Leukocyte Adhesion in Endothelial Tissue Culture Models Under Flow

Atherosclerosis, an inflammatory disease which causes thickening and stiffening of arteries, is a major cause of death in the United States (Lloyd-Jones et al., 2010). These deaths occur because of vessel occlusion created by atherosclerotic plaques and thrombus shedding, leading to heart attack, ischemia, or stroke. Atherosclerosis is expected to be the leading cause of death worldwide within 10 years (Lloyd-Jones et al., 2010). Inflammation plays a significant role in the initiation and progression of atherosclerosis. The cells that line the arteries, endothelial cells (ECs), mediate the inflammatory process. The forces created by blood flow affect the inflammatory response of ECs and the interaction with blood components. This chapter summarizes the background in our recent studies on the response of ECs to blood forces and the interaction of inflammatory cells.


Introduction
Atherosclerosis, an inflammatory disease which causes thickening and stiffening of arteries, is a major cause of death in the United States (Lloyd-Jones et al., 2010). These deaths occur because of vessel occlusion created by atherosclerotic plaques and thrombus shedding, leading to heart attack, ischemia, or stroke. Atherosclerosis is expected to be the leading cause of death worldwide within 10 years (Lloyd-Jones et al., 2010). Inflammation plays a significant role in the initiation and progression of atherosclerosis. The cells that line the arteries, endothelial cells (ECs), mediate the inflammatory process. The forces created by blood flow affect the inflammatory response of ECs and the interaction with blood components. This chapter summarizes the background in our recent studies on the response of ECs to blood forces and the interaction of inflammatory cells.

Vascular anatomy
Arteries have three tissue layers: the intima, media, and adventitia. The intima is lined with a monolayer of ECs in direct contact with blood. The ECs act as a protective membrane, allowing diffusion from the blood stream into the artery. ECs are capable of expressing specific genes in response to physical stresses which cause the vessel to remodel leading to the development of atherosclerosis. In addition, the intima can contain other cells (smooth muscle cells, fibroblasts and inflammatory cells), an extracellular matrix (ECM), and is only a few cell layers thick in healthy tissue. The internal elastic membrane, consisting of a layer of elastic connective tissues, separates the intima and media. The media layer is mainly comprised of smooth muscle cells (SMCs) and ECM. Although the media is involved in atherosclerosis, remodelling is less evident. The adventitia is relatively unaffected by atherosclerosis. It is separated from the media by the external elastic membrane comprised mainly of collagen, providing structural support yet allowing for artery expansion when required (Waller et al., 1992). Over time, an atherosclerotic plaque grows by the accumulation of lipids, inflammatory cells, vascular cells and matrix material in the intima. It often produces a fibrous cap, over a necrotic lipid core, which can weather and rupture over time. The artery is able to compensate for some intimal thickening by expanding outwards, instead of allowing for the plaque to impede blood flow. Eventually the vessel can no longer expand outwards, and negative remodelling can occur. Blood flow is therefore disturbed through the formation of a stenosis (Shah, 2006). This may lead to ischemia and angina pectoris (Libby, 2002). Atherosclerosis can occur in any size of artery. However, clinical manifestations frequently occur in medium and large arteries when the EC layer is breached by erosion or disruption of the fibrous cap. Disruption may occur from a thinning of the fibrous cap as there is increased lipid accumulation, inflammatory cell recruitment and matrix metalloproteinase (MMP) expression, as well as the expression of cytokines inhibiting collagen synthesis. When the plaque is opened up to the blood stream, platelets cause blood coagulation and thrombus formation. There are two possible outcomes after thrombus formation. First, the thrombus may be broken down and reabsorbed. A second outcome is that the thrombus is disrupted by the blood flow and detached from the site of injury. This embolism may then travel through the vasculature to small arteries, where it causes ischemia and potential heart attack or stroke (Libby, 2002).
In advanced lesions, macrophages are unable to take up any additional Ox-LDL, and these lipid molecules begin to accumulate within the intima instead (Schwartz et al., 1991). Ox-LDLs are toxic, and begin to injure and kill ECs, SMCs, and macrophages. When the macrophage is injured, its lipid contents are released into the intimal layer, forming a lipid core, also comprised of enzymes, cytokines, and growth factors that have accumulated within the intima (Schwartz et al., 1991;Shah, 2006). In order to protect the body from the growing necrotic lipid core within the intima, a fibrous plaque is formed over the lesion. This prevents direct contact between the accumulation of cells within the intima and blood flow. The fibrous plaque is composed primarily of collagen, elastin, and proteoglycans that were found in the original fatty streak (Ross, 1999). T lymphocytes and macrophages release MMPs, which break down the extracellular matrix within the intima, allowing for these components to be used within the fibrous plaque (Libby, 2002).

Mechanisms in leukocyte-endothelium adhesion 2.2.1 Intercellular adhesion at the arterial surface
The initiation of atherosclerosis is hypothesized to start with endothelial injury, which triggers inflammatory pathways integral to the progression of the disease (Ross et al., 1977). Leukocytes, including neutrophils and monocytes, preferentially adhere to sites of inflammation. The events that take place during leukocyte recruitment are shown in the representative drawing, Figure 1. Fig. 1. Leukocyte recruitment to a site of endothelial injury. When injured, the endothelium expresses an increase in cell adhesion molecules (CAMs). (A) Leukocytes circulate within the blood stream. (B) Ligands on the leukocytes attach to selectins on the endothelium, effectively tethering the leukocyte. (C) The leukocyte begins to roll across the endothelium, reducing its velocity by forming and breaking selectin-ligand bonds. (D) Leukocytes begin to make bonds between CAMs and integrins. This firmly attaches the leukocyte to the endothelium. (E) Leukocytes migrate towards a cellular junction through CAM-integrin interactions. (F) CAM-CAM interactions allow the leukocyte to migrate through the endothelium by diapedesis.

Leukocyte tethering and rolling
Leukocytes circulating within the bloodstream must make their way to the site of injury, located on the endothelium. Although leukocytes flow in close contact with the endothelial layer, they do not stick until the inflamed endothelium starts to express adhesive molecules (Kelly et al., 2007). Tethering and rolling of leukocytes along the endothelial wall is due to a class of cell adhesion molecules (CAMs) known as selectins. There are three selectins involved: E-selectin, P-selectin, and L-selectin. Both E-and P-selectins are expressed on the endothelium; E-selectin is synthesized and expressed after endothelial stimulation, whereas P-selectin is expressed constitutively and stored, then quickly released upon stimulation (Kelly et al., 2007). L-selectin differs in that it is not expressed on the endothelium, but instead is constitutively expressed on the leukocyte surface. Both E-and P-selectin recognize carbohydrate ligands on the surface of leukocytes, while L-selectin recognizes a series of ligands expressed on the endothelium. When the selectins come in contact with their ligands they will bind, tethering the leukocyte to the endothelium (Miyasaka et al., 1997). Tethering facilitates leukocyte rolling along the EC surface. The velocity of the travelling leukocyte will be reduced as more selectin-ligand bonds form, allowing leukocyte adhesion (Kelly et al., 2007;Kubes & Kerfoot, 2001).

Leukocyte adhesion
Firm adhesion begins when integrins, another class of CAMs, are activated on the surface of leukocytes by chemoattractant cytokines, termed chemokines. Chemokines are secreted by circulating leukocytes and ECs (Kelly et al., 2007). One class of integrin responsible for neutrophil adhesion is the β 2 -integrin. When activated by cytokines, certain β 2 -integrins bind to intracellular adhesion molecules 1 and 2 (ICAM-1 and ICAM-2, respectively) on the EC surface, effectively adhering the leukocyte to the EC. Also activated by cytokines, α 4 β 1integrin will bind with vascular cell adhesion molecule 1 (VCAM-1) (Kelly et al., 2007;Rao et al., 2007). The bound integrin-CAM complex results in firm adhesion of the leukocyte to the endothelial surface (Miyasaka et al., 1997). Although under transcriptional regulation, both adhesion molecules ICAM-1 and VCAM-1 have been shown to be upregulated at sites in the vasculature prone to developing atherosclerotic lesions (Iiyama et al., 1999).

Leukocyte migration
Once the leukocytes have firmly adhered to the endothelium, they may migrate from the lumen of the blood vessel into the subintimal space. Initially, leukocytes must make their way to the closest EC junction through a process termed locomotion. The movement of the leukocyte is made possible through interactions with leukocyte integrins and both ICAM-1 and -2 located on the endothelial surface (Schenkel et al., 2004). At the junction, the leukocytes will encounter another cell adhesion molecule, called platelet endothelial CAM 1 (PECAM-1). PECAM-1 is expressed both on the leukocyte and the endothelium. An interaction between the complementary PECAM-1 molecules allows leukocytes to migrate through the gap junction by diapedesis (Rao et al., 2007;Schenkel et al., 2004), a process also called transmigration.

Idealized arterial hemodynamics
Hemodynamics, the mechanics of blood flow, influence many of the physiological processes of the vascular system (Glagov et al., 1988). From an engineering perspective, blood flow through medium and small arteries (such as the right and left coronary arteries) is often simplified by assuming steady laminar flow in a straight, rigid vessel (Ku, 1997;Nichols & O'Rourke, 1990). Additionally, blood is assumed to be a Newtonian fluid to simplify the flow dynamics to Hagen-Poiseuille flow (Nichols & O'Rourke, 1990). Such assumptions allow us to reduce the governing equations describing pressure-driven flow into a onedimensional velocity profile in the form of Hagen-Poiseuille flow (Nichols & O'Rourke, 1990). For a cylindrical vessel model of arterial perfusion, the velocity profile as a function of a radial dimension is described by: (1) where Q is the volumetric flow rate of blood and R is the hydraulic radius of the vessel. Wall shear stress (WSS) is a tangential force per unit area of a fluid-wall interface that results from flow parallel to the vessel wall. For fluids with constant dynamic viscosity μ (Newtonian fluid), the WSS is the product of the viscosity and the shear rate γ, evaluated at the vessel wall: The wall shear rate of a fluid is the velocity gradient evaluated at the fluid-wall interface. For Hagen-Poiseuille flow through a cylindrical vessel (Equation 1), the wall shear rate is expressed as: Combining Equations (2) and (3) produces an expression for WSS that is dependent on both the vessel geometry and the volumetric flow rate: This equation is accepted as a reasonable model of the average WSS for arteries that are absent from serious geometric disturbances (Ku, 1997). Arterial WSS values range from 5 to 70 dyne/cm 2 with average WSS values of approximately 15 dyne/cm 2 being observed in coronary arteries (Glagov et al., 1988;Malek et al., 1999). Moderate levels of steady, laminar shear stress (> 10-15 dyne/cm 2 ) are believed to induce an atheroprotective EC phenotype while low shear stresses (< 4 dyne/cm 2 ) are believed to induce an atheroprone EC phenotype (Malek et al., 1999). An atheroprone EC phenotype describes one which facilitates the disease pathway marked by an increase in adhesion molecules and a decrease in vasodilators (as described in Sections 2.1 & 2.2) (Libby, 2002;Malek et al., 1999). The fluid flow regime is determined by the dimensionless Reynolds number (Re), which represents the ratio of inertial to viscous forces. For flow in a cylindrical channel, the Reynolds Number is described by: where ρ is the fluid density, D is the hydraulic diameter, and U is the average fluid velocity. A Reynolds Number below 2300 indicates laminar flow that will behave predictably while a value above this threshold suggests the presence of flow disturbances. The average arterial conditions are within a laminar flow regime (Nichols & O'Rourke, 1990) and vary depending on the artery and metabolic demand (Myers et al., 2001;Nichols & O'Rourke, 1990). www.intechopen.com

Localized hemodynamics of leukocyte adhesion
The progression of leukocyte adhesion is strongly influenced by local hemodynamic forces. As the cell is passing along the wall, the torque imparted on the cell by the blood stream causes the cell to spin. As a result, the state of loose attachment with selectin-ligand bonds constantly forming and breaking has become known as cell rolling. The blood stream imposes not only torque but also shear stress on the slow moving cell. In turn, the membrane of the cell will try to distribute this stress by elongating in the direction of flow, allowing for increased binding with the vessel wall. Firrell and Lipowsky found that leukocytes rolling along rat arteriolar walls would elongate by around 140%, allowing their contact area to jump from approximately 14 µm 2 to 50 µm 2 (Firrell & Lipowsky, 1989).
Modelling of bond forces and leukocyte attachment is well documented in the literature (Cozens-Roberts et al., 1990;Evans et al., 2004;Lawrence et al., 1997;Tees & Goetz, 2003). For successful adhesion, a fine balance between the adhesive force and the hemodynamic force must be met. This adhesive force is dependent on several factors, including: the receptor density of both cells, the rate of reaction with respect to both bond formation and dissociation and the strength of the bonds and their response to strain. For instance, the bonds between E-, P-, and L-selectins and their respective ligands behave as catch-slip bonds (Lawrence et al., 1997;Marshall et al., 2003;Sarangapani et al., 2004). Though receptor-ligand bonds spontaneously dissociate (Tees & Goetz, 2003), slip bond behaviour describes an increasing probability of dissociation with increasing tensile force. Catch-slip bond behaviour, on the other hand, strengthens with increasing tensile force until some optimal force has been met. Furthermore, the rate at which this force is increased, known as the force gradient, also affects the strength of the selectin-ligand bonds (Evans et al., 2004). This is of particular interest to the study of stenotic arteries as plaque formation leads to distinct regions of varying force gradients.

Leukocyte adhesion in Parallel-Plate Flow Chambers (PPFCs)
The flow between two parallel plates has often been used to investigate the effects of blood flow on ECs and their interactions with blood components. Traditionally, parallel-plate flow c h a m b e r s ( P P F C s ) h a v e b e e n u s e d t o p r o v ide an environment suited for tissue and suspension culture experiments under laminar flow. In classical PPFCs, fluid is driven through a channel formed by two narrowly separated plates in parallel. ECs are cultured on the bottom surface of the upper plate (often, a glass coverslip) while suspension cultures of leukocytes (neutrophils or monocytes) are prepared in the perfusion medium and their movement visualized within the chamber (Lawrence et al., 1987). This allows for regional or complete surface quantification of cells that are either adherent or, if observed in real-time, leukocytes that are undergoing rolling adhesion. A schematic of a PPFC is presented in Figure 2. In a well defined PPFC, the WSS can be accurately characterised for steady, laminar flow of a Newtonian fluid as a function of a constant measurable volumetric flow rate Q: where w is the width of the plate perpendicular to flow and h is the height of the interstitial gap between the two plates. For a constant volumetric flow rate, the velocity profile is parabolic and the WSS is uniform across the upper and lower plates save for the boundaries defined by the gasket (where the flow field approaches zero) and in the region of developing flow at the inlet of the chamber (Bacabac et al., 2005;Lawrence et al., 1987). In practice, PPFCs are designed with a large w/h ratio allowing most of the flow field to be homogenous over the surface of the cells (Bacabac et al., 2005). Studies of leukocyte adhesion using PPFCs have become the benchmark for revealing the role of shear in the adhesion pathway. Early results revealed a discreet shear dependence on both non-specific (Forrester & Lackie, 1984) and adhesion molecule-mediated adhesion (Alon et al., 1995;Finger et al., 1996;Lawrence et al., 1997). Further studies highlighted the role of endothelial dysfunction and inflammation as a precursor to adhesion when conditioned with flow (Alcaide et al., 2009;Sheikh et al., 2003;Sheikh et al., 2005). These findings complement the paradigm of leukocyte adhesion at sites of vascular inflammation; however, they do not address focal adhesion in non-uniform shear fields. This was considered by performing flow experiments using a step disturbance across the plate of the flow chamber (Burns & DePaola, 2005;Chen et al., 2006). The flow fields created by the step introduce regions of flow reversal and spatial WSS gradients to represent physiological hemodynamics (Burns & DePaola, 2005;Chen et al., 2006). Despite disturbed flow, leukocyte adhesion is increased in areas of high WSS gradients, with the highest incidence in reattachment zones (Chen et al., 2006).

Experimental methods 3.1.1 Asymmetric stenosis model design
Parallel-plate flow chambers are not ubiquitous when characterizing the role of WSS in endothelial dysfunction and leukocyte adhesion. Cone-plate viscometers (Shankaran & Neelamegham, 2001), animal models (Walpola et al., 1993(Walpola et al., , 1995 and three-dimensional (3D) tissue culture models (Hinds et al., 2001) have an increasing presence in the field. A 3D model of an idealized coronary artery with an eccentric stenosis has been developed by our research group to reveal the effect of spatial WSS gradients on both endothelial inflammation and leukocyte adhesion (Rouleau et al., 2010a(Rouleau et al., , 2010b. The eccentric stenosis geometry with a 50% occlusion (i.e. 50% area reduction, orthogonal to flow) has been chosen to represent a clinically relevant atherosclerotic lesion (Brunette et al., 2008;Wexler et al., 1996). The model measures 10 cm in length with an internal diameter of 3.175 mm, Figure 3. Sylgard™ 184 silicone elastomer is cast and cured in PVC moulds to create semicompliant structures that maintain their geometric integrity through the stages of sterilization and cell culture preparation. Fig. 3. Three-dimensional asymmetric stenosis model schematic with regional classifications defined using computational fluid dynamics and photochromic molecular flow visualization (Section 3.1.3).

Perfusion design
The perfusion flow loop consists of a media reservoir with tubing, flow dampeners, and an 8-roller peristaltic pump head with a programmable drive to produce steady or pulsatile laminar flow at the entrance of the models, Figure 4. ECs are cultured in the internal lumen of the model until they form a continuous, confluent monolayer. Inlet WSS values of 4.5, 9 and 18 dynes/cm 2 were chosen to represent moderate physiological shear in coronary arteries whereas inlet WSS values of 1.25 and 6.25 dynes/cm 2 were chosen to represent moderate to high shear for in vitro neutrophil adhesion. The WSS field was experimentally and numerically determined (Rouleau et al., 2010b).

Computational fluid dynamics and photochromic molecular flow visualization
Computational fluid dynamics (CFD) is a theoretical branch of research that relies on the power of modern computers to estimate fluid behaviour. The popularity of CFD lies in its ability to simulate physical experiments, thereby providing direction for further work or allowing for quick testing of key variables. With respect to cardiovascular flow studies, CFD is a numerical solution to a continuum of the Navier-Stokes (NS) equation. Although CFD is a powerful tool, it is only as accurate as the input data (e.g. geometry and mechanical properties). Defining the geometry and mechanics of healthy and diseased tissue is a constant endeavour in biomedical engineering (Choudhury et al., 2009;Tremblay et al., 2010). If the proper information is available, then CFD simulations are feasible, however experimental validation is still crucial. CFD has been performed for our stenosis model, yielding flow profiles at 6.25 dynes/cm 2 and 1.25 dynes/cm 2 , respectively ( Figures 5 & 6). We have used the photochromic molecular flow visualization (PMFV) technique to validate our CFD flow in the stenosis model (Ethier C.R. et al., 2000;Mahinpey et al., 2004). Using the flow profile at a given position, we are able to estimate the shear stress acting on the wall of a channel. Photochromic species reversibly change conformation when excited by a light source, such as a laser, resulting in an observable colour change. A PMFV setup includes a solution of photochromic dye, a laser, and a high-resolution camera. In practice, the photochromic dye solution is pumped through a micro-channel, the laser is triggered and the resulting pulse passes through the solution orthogonal to flow, activating any dye it contacts (Couch et al., 1996;Park et al., 1999). At this moment, a narrow column of visible dye will appear within the solution. The solution is in motion, however, so a fraction of a second later, the excited dye will have displaced with flow. This displacement is recorded by a camera, providing a snapshot of the flow profile and subsequently, the shear stress on the opposing walls. Photochromic visualization results are presented in Figure 6 (c) & (d) to validate our CFD simulation.

Cellular analysis
EC morphology is often a good predictor of EC phenotype. Healthy ECs elongate in the direction of flow, whereas dysfunctional ECs may become randomly oriented and cobblestone in appearance (Dartsch & Betz, 1989). The shape index (SI) is a metric of EC morphology defined as (Nerem et al., 1981): Generally, elongated cells have a lower SI than rounded cells. In concert with the SI, the angle of orientation evaluates the proportion of EC elongation relative to the direction of flow. ECs in regions of observed elongation will be narrowly distributed near 0° (i.e. in the axis of flow) while regions that appear cobblestoned will have a much wider distribution. Protein and mRNA regulation of inflammatory markers and transcription factors defines the endothelial phenotype and relates to leukocyte adhesion and atherogenesis. Gene regulation is quantified for large cultures using Q-PCR, while the resultant protein expression is observed using Western Blotting. Regional inflammation around the stenosis can be observed using immunostaining and confocal microscopy for adhesion molecules and the translocation of inflammatory transcription factors to EC nuclei. Our analysis includes, but is not limited to, specific inflammatory markers and adhesion molecules, including: ICAM-1, VCAM-1, E-selectin and NF-κB.

Results and discussion 3.2.1 Endothelial cell morphology
The stenosis model was first used to investigate the morphological effects of shear gradients caused by the stenosis. A morphological response is one of the last measurable changes which occurs in the cascade of events following introduction of flow. As ECs experience a steady WSS, they tend to become elongated and aligned in the direction of flow, representing a healthy endothelium. When exposed to low shear magnitude or WSS spatial gradients, ECs tend to take on a more cobblestone and random morphology which is indicative of an unhealthy endothelium (Helmlinger et al., 1991;Levesque et al., 1986;Levesque & Nerem, 1985;Nerem et al., 1981;Nerem, 1993). Perfusion experiments to evaluate EC response were run at wall shear stress values of 4.5, 9 and 18 dyne/cm 2 which corresponded to Reynold's numbers of 50, 100 and 200, respectively. It was found that at all times and inlet WSS values the shape indices in the inlet and outlet of the stenosis model were statistically similar to that of a straight model. These results demonstrate that these values can be a good reference point for morphological changes in the regions surrounding the stenosis. Furthermore, the longer the perfusion time, the more elongation was observed in the direction of flow.

Effect of wall shear stress gradients on endothelial cell morphology
The stenosis model allows the observation of the morphological response of the ECs to WSS gradients. Figure 7 shows the WSS patterns within the model as a function of position. A positive shear gradient is found in the proximal region of the stenosis, reaching a peak WSS just upstream of the apex. A negative WSS gradient is observed in the recirculation zone of the stenosis, however, downstream of the flow reattachment point, laminar Hagen-Poiseuille flow resumes. It was found that after sufficient time had passed, morphological trends formed throughout the model. The inlet and outlet of the model showed similar shape indices as the straight tube controls, making them acceptable internal controls. These uniform internal controls can be compared to the other regions of the models. The deceleration, or recirculation zone (depending on the flow patterns, governed by the Re), showed the highest shape index, perhaps indicative of the most inflammatory response. The acceleration, or proximal zone, showed a slightly elevated shape index compared to the control regions, though this value was still statistically lower than that found in the deceleration zone. It was expected that the increased shear in this zone would result in more elongated ECs, however the results suggest that WSS gradients can have a more drastic effect on endothelium health than WSS magnitude alone. It can be concluded that the deceleration zone could potentially present an inflamed endothelium and therefore one would predict to see an increase in regional expression of proteins linked to inflammation in that region. In turn, this should also lead to the largest amount of neutrophil adhesion in the recirculation zone.

Regional inflammation and adhesion molecule expression
WSS magnitude and duration was investigated to determine the effect these factors have on inflammatory response. Straight, cylindrical models were perfused under 4.5, 9 and 18 dyne/cm 2 inlet WSS conditions for various time periods (up to 24 hours). VCAM-1 and ICAM-1 mRNA expression decreased with increasing WSS magnitude and time, indicative of an atheroprotective phenotype. An increase in WSS magnitude resulted in a decrease in E-selectin mRNA expression; however, E-selectin expression increased from 0-12 hours, and sharply fell by 24 hours. Inflammation was considered in the stenosis model by observing regional endothelial CAM expression using immunostaining and confocal microscopy. Perfusions with neutrophils were run at 1.25 dyne/cm 2 , revealing an increase in CAM expression at the peak of the stenosis, Figure 8 ( Rouleau et al., 2010a). Furthermore, perfusions were run at a higher inlet WSS of 6.25 dyne/cm 2 , with no noticeable difference in regional CAM expression. These WSS values are consistent with the conditions used for neutrophil adhesion experiments. Using inlet WSS values of 4.5, 9 and 18 dyne/cm 2 , more physiologically relevant hemodynamics were present. During these perfusions, ICAM-1, VCAM-1 and E-selectin levels were quantified and NF-κB translocation was observed to provide a robust picture of the regional inflammation around the stenosis. Similar to the experiments at 1.25 and 6.25 dyne/cm 2 , there was an upregulation of ICAM-1 and VCAM-1 at the stenosis peak and in the proximal and recirculation zones. Similarly, E-selectin and NF-κB were also upregulated in these areas. It can then be concluded that there is a higher likelihood of increased neutrophil adhesion around the stenosis of the model.

Regional neutrophil adhesion
In perfusion experiments, it was shown that both WSS magnitude and perfusion time significantly affected the adhesion of a leukocyte cell line (NB4 cells). The trends showed that flow conditioned cells resulted in reduced adhesion of the NB4 cells for both the TNF-α stimulated and non-stimulated ECs. The two WSS magnitudes investigated, 1.25 dynes/cm 2 and 6.25 dynes/cm 2 , resulted in a 3 fold and 15 fold decrease in adhesion, when compared to cells that were kept static prior to the adhesion experiments, respectively. Furthermore, experiments were run under even higher WSS conditions (12.5 dynes/cm 2 ) and it was found that very few cells were able to adhere to the ECs. This data demonstrates the influence of hemodynamic and attractive (ligand-receptor) forces acting on the neutrophils. The higher hemodynamic forces push the neutrophils off of the binding sites, overcoming the attractive forces which form during the adhesion of the neutrophils. For both the low (1.25 dynes/cm 2 ) and high (6.25 dynes/cm 2 ) shear stress conditions it was found that an increase in perfusion time from 1 to 6 hours resulted in an increase in adhesion, with a more noticeable increase occurring in non-stimulated ECs, potentially showing that at the shorter time point in TNF-α stimulated cells, a maximum in vitro adhesion is reached.

Regional neutrophil adhesion
The stenosis model presents a unique 3D environment which allowed for the investigation of the spatial differences in adhesion on and around a stenosis. Videos of the adhesion assays showed that there was a region of flow recirculation downstream of the stenosis. Immediately downstream of the separation point there was a distinct line of NB4 cell adhesion. It is postulated that this focal neutrophil adhesion was facilitated by low WSS and minimal fluid momentum caused by backflow in the recirculation zone. For the rest of the analysis, the average adhesion was evaluated for each region of the model, Figure 9. TNF-α stimulation increased adhesion in all regions save for the stenosis peak. It was found that both the WSS magnitude and perfusion duration affected the incidence of adhesion. For example, it was found that at low inlet WSS (1.25 dyne/cm 2 ), ECs in the recirculation and distal regions showed a significant increase in adhesion from 1 to 6 hours. It was found that in general, the recirculation zone tended to have the highest cell adhesion. It is hypothesized that the recirculation of NB4 cells results in a higher concentration of cells flowing along the endothelium (Rouleau et al., 2010a). Furthermore, the leukocytes have a lower momentum in the recirculation zone due to the decreased shear. This would allow for an increase in adhesion in this region. The endothelium in this location is also exposed to reduced WSS, leading to an increased inflammatory response. The proximal and distal regions have lower incidence of adhesion than the recirculation region but more than at the stenosis peak. Interestingly, there seemed to be greater adhesion in the proximal region than the inlet of the model which may be due to a positive wall shear gradient. As the fluid reaches the stenosis, the projected surface area decreases resulting in an increase in WSS. Hinds et al. found a similar result in their studies using monocytes (Hinds et al., 2001). Comparing these two results, it can be seen that there is increased adhesion of leukocytes to ECs in the presence of complex wall shear stress gradients. In all instances, the stenosis peak had relatively low adhesion, which is a result of the high shear forces. By the peak of the stenosis, WSS values were appreciably larger than those found in the inlet (Figure 7). These hemodynamic forces exceed the adhesive force needed for the neutrophils to adhere. Extending static adhesion past 1 hour resulted in little or no additional adhesion. This was likely due to a lack of adhesion sites for the NB4s to bind to. Ultimately, under any conditions, there will be a point where the endothelium becomes saturated with bound NB4s and simply cannot facilitate further adhesion, in vitro. Although this leaves a certain limitation on the results of the aforementioned experiments, it also demonstrates the high levels of focal adhesion which can occur as a result of endothelial inflammation.

Conclusions
Atherosclerosis is an inflammatory disease. In vitro studies of the interaction of inflammatory cells with the endothelium have advanced our understanding of the role of inflammation in atherosclerosis development and progression. Our novel three dimensional dynamic cell culture model of a coronary stenosis has shown the importance of spatial gradients in wall shear stress in EC response and leukocyte attachment. Leukocyte attachment is increased in the proximal and distal regions of the stenosis. The increased attachment occurs in regions where the ECs have an inflamed phenotype. The results suggest that the hemodynamics created by the stenosis geometry create an inflammatory response of the endothelial cells that promotes leukocyte attachment. These results help to explain disease stability in established coronary stenoses.