Targeting functionalized nanoparticles to activated endothelial cells under high wall shear stress

Abstract Local inflammation of the endothelium is associated with a plethora of cardiovascular diseases. Vascular‐targeted carriers (VTCs) have been advocated to provide focal effective therapeutics to these disease sites. Here, we examine the design of functionalized nanoparticles (NPs) as VTCs that can specifically localize at an inflamed vessel wall under pathological levels of high shear stress, associated for example with clinical (or in vivo) conditions of vascular narrowing and arteriogenesis. To test this, carboxylated fluorescent 200 nm polystyrene particles were functionalized with ligands to activated endothelium, that is, an E‐selectin binding peptide (Esbp), an anti ICAM‐1 antibody, or using a combination of both. The functionalized NPs were investigated in vitro using microfluidic models lined with inflamed (TNF‐α stimulated) and control endothelial cells (EC). Specifically, their adhesion was monitored under different relevant wall shear stresses (i.e., 40–300 dyne/cm2) via real‐time confocal microscopy. Experiments reveal a significantly higher specific adhesion of the examined functionalized NPs to activated EC for the window of examined wall shear stresses. Moreover, particle adhesion correlated with the surface coating density whereby under high surface coating (i.e., ~10,000 molecule/particle), shear‐dependent particle adhesion increased significantly. Altogether, our results show that functionalized NPs can be designed to target inflamed endothelial cells under high shear stress. Such VTCs underscore the potential for attractive avenues in targeting drugs to vasoconstriction and arteriogenesis sites.

immune system cells (e.g., lymphocytes, monocytes, macrophages) and the endothelial cell layer veering the vascular walls. 1 The basic interaction between these two cells types at the initial steps of inflammation relies on inflammatory receptor-ligand interactions, 2 and therefore, can be employed for drug delivery using vascular-targeted carriers (VTCs).
Endothelial cells (EC) response to inflammatory mediators is termed endothelial activation and it varies according to the specific mediator. For example, histamine is known to promote short-term response of ECs, while TNF-α stimulates a long-term response. 3 While short-term EC response occurs within a few minutes and involves rapid expression of membrane receptors, long-term response induces new adhesion molecules synthesis. 3 The inflammatory receptors interact with their ligands on leukocytes surface and can similarly be utilized for targeted drug delivery to activated ECs. The first receptors to be expressed on ECs are receptors of the selectin superfamily. One of them, E-selectin, is crucial for capturing leukocytes circulating within blood flow and enables their rolling over the inflamed EC under physiological flow 4-6 (see schematic of Figure 1a). E-selectin is stored in α granules inside resting EC, but following an inflammatory stimuli these vesicles fuse rapidly (within seconds) with the cell membrane. 6 However, E-selectin expression declines generally within the first 24 hr post stimulation. 6 This rapid post-stimulation expression of E-selectin may be utilized for efficient drug-targeting. 7,8 The next step in ECleukocyte cascade, following rolling, consists of a firm adhesion 3,9,10 as shown in Figure 1b. This step is mitigated by overexpression of immunoglobulins superfamily members, such as Intra-cellular adhesion molecule-1 (ICAM-1) and vascular cell adhesion molecule-1 (VCAM-1). 3,11,12 In contrast to E-selectin, ICAM-1 has abroad distribution in the body due to its function in cell-cell and cell-matrix adhesion. While under resting conditions ICAM-1 expression is low, upon chemokines simulation or shear stress alterations 6,12 it is upregulated.
ICAM-1 becomes overexpressed mostly after 4 hr, and even 48 hr later there is still altered expression of the ICAM-1 receptor. 13 Yet, the stable pairing of the ICAM-1 receptor to its ligands provides firm adhesion, which is highly desirable in drug delivery approach. 7,8 Based on the natural molecular interaction occurring during inflammation between leukocyte and EC, vascular-targeted carriers (VTCs) that target activated ECs have been sought and developed ( Figure 1b) as a potent drug delivery strategy. 14 Briefly, the active targeting of VTCs is based on functionalizing these carriers with ligands and antibodies that enable specific ligand-receptor pairing with their complimentary overexpressed inflammatory receptors on the EC. 13,15 Achieving effective targeting remains a challenging task where VTCs are designed to adhere to specific receptor overexpressed on a limited number of cells at the disease site while avoiding adhesion to vascular cells throughout the body, where cells exhibit lower receptor expression. 16 Therefore, various studies have focused on targeting functionalized nanoparticles to inflamed endothelium based on over-expression of adhesion receptors such as: F I G U R E 1 Schematic of adhesion process of (a) leukocyte and (b) leukocyte mimetics VTC to activated EC under flow, as mediated via EC inflammation receptors (including ICAM 1 and E-selectin) that are locally over-expressed upon inflammation. (c) The VTCs are perfused by a syringe pump with controlled flow rate inducing shear stress on the seeded EC in a microfluidic channel. The illustration was created with BioRender.com ICAM-1, VCAM, and E-selectin. 17,18 However, in the cardiovascular system, this task needs to be performed under dynamic conditions including blood flow and its associated hemodynamic forces. 13,19 22 Additionally, to improve site specificity, the use of multivalent targeting, using two or more ligands which can provide synergetic adhesion, have also been explored, under these low WSS conditions. 13,23 Yet, so far these have been limited to healthy or low WSS conditions and have not addressed VTC adhesion under pathological levels of high WSS where WSS can be 100 dyne/cm 2 and higher. 24,25 Studies addressing NPs targeting under high WSS region have so far focused on targeting thrombotic arteries 26 and not the endothelium. However, high WSS and endothelial activation are critical aspects known to occur in several disease conditions including vaso-constricted arteries or at early stages of arteriogenesis following occlusion of a neighboring artery. 20,24,27,28 Thus, although not studied before, targeting NPs to the activated EC under these high WSS conditions can be valuable.
In the present work, we explore in vitro the adhesion of VTCs to

| Microfluidic device fabrication
Microfluidic devices were produced from Polydimethylsiloxane (PDMS) using conventional soft-lithography, as reported previously. 29 Briefly, a simple channel (22 mm in length and 2 mm in width) was cut in an 80-μm aluminum adhesive film using a cutter plotter (CE5000, Graphtec, CA). A 2.2 cm long channel was used to assure a uniform, stable WSS at the examined region (see also Computational Fluid Dynamic [CFD] simulation results in Supporting Information and Figure S1). The obtained aluminum strips were placed at the bottom of a petri dish. Next, a PDMS Sylgard 184 (Dow Corning, Midland, MI) mixture containing a 1:10 ratio of crosslinker and resin, respectively was poured into the dish, degassed and kept at room temperature overnight. Afterwards the PDMS was peeled, and inlet and outlet holes were punched. Finally, the PDMS channels were sealed to a glass slide 76 × 50 × 1 mm (Marienfeld, Germany) utilizing oxygen plasma.

| Cell culture
Prior to cell seeding, the surface of the microfluidic device was coated using 100 μg/ml human fibronectin (Sigma Aldrich, Israel) in PBS. Human

| Staining HUVEC cells against E-selectin and ICAM-1
Cells were stained against E-selectin by incubating the fixed TNF-α stimulated cells with 50 μg/ml FITC labeled E-selectin binding peptide, Esbp 30 (purchased from GL Biochem, China, >95% purity validated by HPLC and MS), 30 for 2 hr at room temperature. Also, ICAM-1 receptors were stained using incubation with 2.5 μg/ml anti ICAM-1 antibody (ThermoFisher, MA) for 2 hr, and labeled using a secondary antibody Alexa fluor 647 anti-mouse (ThermoFisher, MA). Cell nuclei were stained using DAPI. The microfluidic channels were washed twice with PBS between staining steps. Confocal images were obtained using a Nikon based confocal microscopy system (Andor, Belfast GB) and recorded by an Andor DS-Ri2 camera. Images were acquired by an Andor iQ3 software and fluorescent intensity analysis was performed via an ImageJ software. Data was averaged from 3 to 5 images per experiment and three repeat experiments were conducted for each condition. The amount of conjugated aICAM-1 on particle surface was estimated using the supernatant protein content by a Smith assay with bicinchoninic acid kit for protein determination (Sigma Aldrich, Israel).

| Particle functionalization
The absorption reads were taken by a CARRY Bio100 spectrophotometer. For particle Esbp quantification, the FITC fluorescence levels were read using a Microplate reader (Varioskan LUX, ThermoFisher, MA). In addition, particle size and zeta potential of coated particles were measured by a DLS Zetasizer Nano instrument (Malvern, GB).

| Perfusion system and adhesion experiments
The perfusion system comprises a HUVEC seeded PDMS microfluidic device, a syringe pump (KDS Scientific, MA) connected to a syringe filled with particle solution at a concentration of 3.4 ± 0.5 μg/ml suspended in PBS which was supplemented with 1%(w/v) BSA (Millipore, MA) buffer. Particle concentration was measured via fluorimetry utilizing a calibration curve of the stock solution. The particle solution was perfused through the channel at a controlled constant flow rate ranging between 512 and 3,840 μl/min, to monitor the influence of various levels of WSS (see below). The particle solution was perfused between 15 to 30 min in each experiment. Particle adhesion was assessed via florescent confocal microscopy.

| Shear stress and adhesion calculations
Flow-induced WSS corresponds a tangential force applied on a surface due to viscous fluid flow. In our microfluidic model, we set the WSS using the analytical formula established under laminar flow conditions for a wide channel (w h) such that where τ w is the WSS, μ the fluid viscosity, Q is the flow rate, h and w are the channel height and width, respectively. Here, we assume a fully-developed laminar flow for a Newtonian incompressible fluid obeying no-slip conditions at the wall.

| Data analysis and statistics
Confocal time-lapse images were taken for each flow experiments.
Using a custom analysis software (Matlab ® ), we extract the number of present particles in each frame and the slope representing the average adhesion rate over time (i.e., number of particles per mm 2 per min).
Additionally, the particle adhesion probabilities were also calculated as described All statistical analyses were determined using GraphPad Prism 8 ® software.

| E-selectin and ICAM-1 ligand adhesion to EC
In this work we have focused on VTCs functionalized with two common inflammatory ligands, namely an Esbp and an anti ICAM-1 antibody. The Esbp is an artificial peptide, first synthesized by Shamay et al. 30 The Esbp CDITWDQLWDLMK-CONH2 sequence labeled with FITC-Lys was used in our study to allow its fluorescence detection. The peptide binds E-selectin with high affinity but not P-selectin and L-selectin, members in selectins superfamily. 30 For I-CAM1 targeting we used an anti ICAM-1 monoclonal antibody from mouse origin, which reacts with human ICAM-1, and has been widely studied for VTCs. 7

| Effect of ligand surface density on NPs adhesion under high WSS
To test the effect of particle surface ligand density on adhesion under high WSS, 200 nm polystyrene NPs were first functionalized at two surface densities: 1,000 (low density) and 10,000 (high density) copies per μm 2 . The ligand density, hydrodynamic diameter and zeta potential of the various studied formulations are summarized in Figure 3.
ECs were then cultured in the microfluidic devices and inflamed with TNF-α over 4 or 6 hr for Esbp and aICAM-1, respectively. These  Figure 4a. Moreover, we observed that the adhesion rate was shear-dependent, that is, increasing with higher shear.
To test the effect of ligand surface density on adhesion under pathologically high WSS, we performed experiments with the low-density Esbp NPs and compared them to the high-density Esbp NPs. The results show a strong trend whereby high density Esbp NPs adhere more than lowdensity Esbp NPs, as shown in Figure 4b for representative images at WSS = 300 dyne/cm 2 . Quantitative analysis of the results (Figure 4c) show that at 100 and 300 dyne/cm 2 , the adhesion rate of high -density Esbp NPs to activated EC is more than two folds higher compared to low-density Esbp NPs. Thus, Esbp NPs appear to provide a selective attachment to activate the inflamed ECs and increasing the ligand density has a positive impact on this adhesion.
To test aICAM-1 NPs, the same perfusion experiments were repeated for conditions of low-density and high-density aICAM-1 NPs; their adhesion under flow to inflamed, (6 hr) TNF-α stimulated ECs as well as normal ECs was monitored and quantified. For highdensity aICAM-1 NPs (Figure 5), the selective adhesion to activated ECs compared with unstimulated ECs was consistently observed for all the WSS values examined (i.e., 40, 100, and 300 dyne/cm 2 ). In contrast, for low-density aICAM-1 NPs, significantly less adhesion was observed under the same WSS conditions, as observed in Figure 5b showing representative confocal microscopy images at WSS of 300 dyne/cm 2 . Quantitative analysis of the adhesion rate of both particles ( Figure 5c) underlines a >3-fold increase in particle binding of high-density aICAM-1 NPs, when spanning 100-300 dyne/cm 2 . On the other hand, low-density aICAM-1 NPs show a decrease in binding when WSS increases, as well as significantly less overall adhesion.
Regarding the high aICAM-1 results ( Figure 5c) the observed behavior of the NPs can be attribute to a more complex interaction between the anti-body and the ligand, which can also be influenced by flow.
F I G U R E 3 NPs ligand quantification, number mean value and Z-average of the particles' size, particles' PDI, and the ζ-potential evaluation  Figure 3). Specifically, 5,000 copies of each ligand were immobilized on the same particle to produce the same total molecules as compared under high-density Esbp and aICAM-1 NPs. We then perfused the NPs on 0.5-hr TNF-α stimulated EC or 4 hr stimulated EC, while monitoring particle adhesion under high WSS. Then we compared them with high-density Esbp and aICAM-1 NPs.
As shown in Figure 6a NPs are superior to Esbp NPs across all the examined WSS including at a physiological level of 40 dyne/cm 2 . The difference between the two particles is more pronounced than at 0.5 post TNF-α simulation.
Thus ICAM-1 may offer a more firm and stable adhesion compared to E-selectin and similarly aICAM-1 NPs may offer stronger adhesion compared to Esbp NPs, 33 suitable for high WSS conditions.
When dual targeting particles were tested, these particles exhibited significant higher adhesion compared to both Esbp and aICAM-1 particles after half an hour of inflammation (Figure 4), at all the examined WSSs, and showed a higher adhesion probability under high WSS (100 and 300 dyne/cm 2 , see Figure S2). Thus, suggesting synergetic effects of E-selectin and ICAM-1 binding to their ligands. At 4 hr of TNFα stimulation, this synergetic effect diminishes probably due to the fact that the inflammatory receptor overexpression profile has changed.
Therefore, the ratio of 1:1 does not fit optimally the receptors' expression as ICAM-1 is significantly more dominant than the E-selectin. 13 Additionally, for each ligand a lower surface density, that is, 5,000 copies per μm 2 , on the dual targeting particles exists and as ICAM-1 interaction becomes dominated the adhesion compared to the higher 10,000 copies per μm 2 aICAM-1 is reduced. This phenomena was also well investigated by Eniola-Adefeso et al 13  the ligand-density leads to a more firm adhesion. 13 Indeed, the high density coated NPs adhere better to the inflamed ECs under high WSS, however, as shown in Figure 6, this comes at the cost of a significant increase in the adhesion of the high density coated NPs to non-inflamed ECs. When looking at the specificity of the different NPs formulation, defined as the ratio between the adhesion of NPs to activated ECs (on target) divided by the adhesion to normal ECs (offtarget), at different levels of WSS ( Figure S4) a complex behavior is observed where the coating type, coating density and activation time all play a role. For both Esbp ( Figure S4a) and aICAM-1 NPs ( Figure S4b) the higher density NPs correspond to higher specificity at all the examined high WSS. Additionally, the high-density aICAM-1 NPs show higher specificity than high-density Esbp NPs. As shown in