Ex vivo isolated human vessel perfusion system for the design and assessment of nanomedicines targeted to the endothelium

Abstract Endothelial cells play a central role in the process of inflammation. Their biologic relevance, as well as their accessibility to IV injected therapeutics, make them a strong candidate for treatment with molecularly‐targeted nanomedicines. Typically, the properties of targeted nanomedicines are first optimized in vitro in cell culture and then in vivo in rodent models. While cultured cells are readily available for study, results obtained from isolated cells can lack relevance to more complex in vivo environments. On the other hand, the quantitative assays needed to determine the impact of nanoparticle design on targeting efficacy are difficult to perform in animal models. Moreover, results from animal models often translate poorly to human systems. To address the need for an improved testing platform, we developed an isolated vessel perfusion system to enable dynamic and quantitative study of vascular‐targeted nanomedicines in readily obtainable human vessels isolated from umbilical cords or placenta. We show that this platform technology enables the evaluation of parameters that are critical to targeting efficacy (including flow rate, selection of targeting molecule, and temperature). Furthermore, biologic replicates can be easily produced by evaluating multiple vessel segments from the same human donor in independent, modular chambers. The chambers can also be adapted to house vessels of a variety of sizes, allowing for the subsequent study of vessel segments in vivo following transplantation into immunodeficient mice. We believe this perfusion system can help to address long‐standing issues in endothelial targeted nanomedicines and thereby enable more effective clinical translation.

explored in recent years, clinical impact has yet to be realized. [4][5][6] Even in cases with promising preclinical results in rodents, subsequent clinical trials have shown only modest efficacy in humans. 7 This lack of clinical relevance has led some to question whether targeted nanomedicines are a viable therapeutic technology. 8 Notably, the majority of this prior research has attempted to target extra-vascular cells (e.g., in solid tumor) via intravascular NP delivery. In contrast to poorly accessible extra-vascular cells, vascular endothelial cells (ECs) are directly accessible to circulating NPs. Indeed, extensive research on EC-targeted NPs has demonstrated that molecular-targeting can dramatically enhance the retention of NPs on ECs both in vitro and in vivo. [9][10][11] Furthermore, ECs play a critical role in homeostasis and inflammation, and therefore represent an attractive cell population for treatment with targeted nanomedicines in a variety of disease indications. 12 The predominant path for preclinical development of EC-targeted NPs typically begins with static cell culture models. [13][14][15] These in vitro models are easy to use, available to most laboratories, and produce clear quantitative readouts to assess targeting efficacy (i.e., mean fluorescence intensity of cells treated with NPs encapsulating fluorescent dye). However, in most cases, cells in culture have significantly different biological properties compared to cells in their native tissue environment. Moreover, the physiologic and environmental characteristics of static cell culture systems are a poor proxy of a complex in vivo system. Consequently, it is not surprising that NP targeting in vitro is typically poorly predictive of efficacy in vivo. 4,16 Microfluidic flow chambers may provide some improvement by enabling study of NP targeting to EC under physiologic shear conditions, which has been shown to effect NP internalization. 17,18 However, such microfluidic systems may not be available to all researchers and still do not replicate the native 3D archictecture and surrounding cell types of native vessel beds.
Animal models, in contrast to cells in culture, provide a more complex physiological setting within which to evaluate NP targeting efficacy. [19][20][21] Notably, effective vascular targeting of NPs in mouse models has been achieved in select tissues (e.g., lung and brain). 9,[22][23][24][25][26][27][28] These studies have revealed the critical importance of molecular properties on NP targeting efficacy. They have further highlighted the need to optimize both target-receptor selection and NP-ligand properties to ensure effective delivery. Engineering such systems depends on specific characteristics of the receptor targets displayed on the luminal surface of the ECs and the binding properties of the targeting ligand conjugated to the NP surface. 29 As a consequence, it is unlikely that the same reagents that bind to ECs of mice will be relevant in humans. Moreover, genetically identical mice lack the natural anatomic and physiologic variability associated with the human population. 19 Finally, experiments conducted in a complex animal model do not allow for isolation or control over key variables (i.e., perfusate composition, temperature, flow rate, and circulation half-life) obscuring the influence of NP design and environmental factors on NP targeting efficacy.
The aforementioned challenges associated with simple cell culture models and complex animal models suggest the need for approaches that can enable evaluation of targeted NP delivery in three-dimensional (3D) human tissues under circulation. Recently, we have investigated targeted NP delivery to human renal vasculature during ex vivo normothermic machine perfusion (an emerging clinical modality for improving the condition of solid organs prior to transplantation 30 ). This study allowed us to measure the targeting of anti-CD31 NPs to ECs within native vascular beds of human kidney. 31 Though ex vivo perfusion of human organs provides a unique opportunity to evaluate targeted NPs in a native setting, access to nontransplanted human organs is limited and the size of the experiments (typically 500 ml perfusion volume in a human kidney) makes these experiments highly costly. Additionally, it is not currently possible to study long term drug effects in these human tissues.
In other recent work, we have circumvented these issues by delivering NPs to small diameter human vessels ex vivo. These small vessels have the potential for long term evaluation following transplant in humanized mice. In one such study, we have treated human coronary artery segments ex vivo with siRNA-loaded NPs that accumulated within ECs during the treatment period. 32 After NP treatment, the human arteries were transplanted as interposition grafts into the infrarenal abdominal aortae of immunodeficient mice. Over the 14 day study, NPs provided a prolonged knock-down of major histocompatibility complex (MHC) Class II molecules, which reduced inflammatorystimulated damage by adoptively transferred human peripheral blood mononuclear cells from a donor allogeneic to the artery graft. 32 These quantitative studies of NP interactions with human tissue could be optimized by a testing platform that allows for controlled perfusion of multiple vessels collected from the same donor.
Here we have developed a testing platform that enables us to use readily obtained human vessel segments in isolation ex vivo. In this platform-which we call the IVPS (isolated vessel perfusion system)vessel segments are maintained as 3D structures with an intact native endothelial lining, allowing quantification of NP accumulation and cellular response. In proof of principle studies, we used this system to compare the effect of temperature, flow rate, and cellular target on NP targeting efficacy to vascular ECs under flow. Furthermore, we used the system to deliver NPs to vessels ex vivo, and then evaluated the persistence of NPs after in vivo implantation of the treated vessels in immunodeficient mice. In sum, we have developed a reliable, simple, and inexpensive testing platform for quantitative assessment of vascular-targeted nanomedicines in readily accessible human tissues. We believe this platform can complement existing models for NP design optimization (cell culture, in vivo animal models, ex vivo nontransplanted human organs) and thereby improve clinical translational of vascular-targeted nanomedicines.

| Fresh collection of umbilical arteries from C-section ensures endothelial integrity
Human umbilical arteries can potentially provide an abundant, accessible, and consistent source of intact endothelium for NP targeting efficacy experiments. We first sought to identify optimal conditions to maintain endothelial integrity in these arteries prior to experimentation.
Confocal microscopy of en face mounted specimens revealed that the endothelial layer of umbilical arteries was continuous and intact when evaluated within 3 hrs after cesarean section (C-section) for tissues stored at 4 C (Figure 1(a)); 3 hrs was the earliest time point we could reliably retrieve arteries from the delivery room and then dissect for experimentation. Prolonged tissue storage of 8 or 24 hrs at 4 C reduced EC layer integrity. Umbilical cords maintained at either 18 C (room temperature) or 37 C after C-section also appeared to have intact EC layers at the 3 hr time point, but similar to 4 C, EC were lost after prolonged storage. Based on this evidence, we concluded that tissues should be obtained fresh from C-section (as opposed to unscheduled, natural birth) so that the time and handling conditions between tissue collection and experiment could be reliably controlled. For all subsequent experiments, we maintained cords at 4 C after recovery to reduce cellular metabolism and any injury associated with warm ischemia.
Finally, we assessed the integrity of EC junctions following recovery and storage at 4 C for 3 hrs prior to staining. Vessels were fixed, permeabilized, and stained for either CD31 or VE-cadherin as an indicator of EC coverage and junctional integrity ( Figure 1b). Confocal images at a variety of magnifications (×10, ×40, and ×100) demonstrated intact EC layers with well-defined EC junctions. Collectively, these data suggest that umbilical arteries recovered from C-section can provide a reliable source of human vessels for subsequent ex vivo experimentation.

| The IVPS enables ex vivo perfusion of human vessel segments
To develop a platform for assessment of targeted NP interactions with human blood vessels, we constructed a closed-loop, ex vivo perfusion system. A schematic of our design is shown in Figure 2a. In this system, a peristaltic pump drives fluid at a controllable volumetric flow rate. By adjusting pump speed and tubing resistance, the IVPS can be adjusted to control flow rate, intravascular pressure, and shear stress at the surface of the vascular lumen ( Table 1)

| Physical perfusion parameters affect NP accumulation on endothelium
Temperature is a critical aspect of optimal organ preservation during ex vivo perfusion for transplant; both hypothermic and normothermic perfusion conditions are under clinical evaluation. 33  of NP appears to be primarily junctional at 4 C and less restricted to the EC junctions at 37 C. Additional~2 hr room temperature incubation after perfusion at 37 C led to even more spatial shift of NP away from the junctions perhaps indicating NP internalization ( Figure S1).
While a detailed study of NP internalization is beyond the scope of this method-focused manuscript, these data suggest that the IVPS could be used to study the effects of temperature on NP internalization in intact vessels.
The flow rate of perfusion was also assessed for effects of NP targeting to ECs. NPs were delivered at flow rates of 1.  Figure S2). While reduced NP retention was observed at 5 ml/min, we also noted reduced EC recovery and potential areas of sheared endothelium at this flow rate ( Figure S2). It is possible that 5 ml/min may reflect the upper limit of experimental parameters for assessing NP binding in this system. Nevertheless, we conclude that perfusion at 37 C between 1.5 and 2.5 mL/min results in indistin-

| NP formulation
NPs were formulated following a well-established nanoprecipitation method. 31  We have found that tissue storage conditions have a pronounced effect on the quality of the endothelial layer prior to experimentation.
As the tissue waits between the collection and the experiment, the endothelial layer becomes less continuous, with gaps in CD31+ cells on the luminal surface of the vessel. This progressive decline in endothelial quality is exacerbated as storage temperature increases from 4 to 18 C and to 37 C. Representative images of umbilical arteries stored under each of these conditions are shown in Figure S1. Based on these results, only blood vessels collected immediately after a Csection were used for studies, to reduce variability in wait time and thus variability in the starting endothelium quality.
The perfusion system and the perfusion chambers were easily adapted to various vessel diameters and graft lengths by selecting the cannulated gavage needle diameter and the length of the perfusion chamber (see Figure 1b,c).

| Characterizing flow profile
Based on the capabilities of the peristaltic pump and the diameter of the perfused vessel, a table of obtainable flow rates and resulting shear stress was calculated. Internal pressure at different flow rates with a constant tubing diameter was measured using a probe. The shear stress was calculated by the Hagen-Poiseuille formula: where τ, wall shear stress; Q, volume flow rate; μ, viscosity of fluid (assumed at 0.01 dyneÁs/cm 2 ); and r, inner radius of cylindrical tube. 35 These results are reported in Table 1. EC layer and junctional integrity were assessed with five randomly taken confocal images at ×10, ×40, and ×100 magnification per each sample.

| Implanted grafts histology and immunostaining
The implanted vascular grafts were explanted 7 days after implantation and then snap-frozen. NP retention and endothelial preservation were evaluated using frozen sections, sliced longitudinally with 15 μm thickness, and stained using fluorescent antihuman CD31 as described above. Sections were mounted using antifade mounting medium with DAPI, and imaged using an EVOS fluorescent imaging system microscope.