Targeting therapeutics to endothelium: are we there yet?

Abstract

Vascular endothelial cells represent an important therapeutic target in many pathologies, including inflammation, oxidative stress, and thrombosis; however, delivery of drugs to this site is often limited by the lack of specific affinity of therapeutics for these cells. Selective delivery of both small molecule drugs and therapeutic proteins to the endothelium has been achieved through the use of targeting ligands, such as monoclonal antibodies, directed against endothelial cell surface markers, particularly cell adhesion molecules (CAMs). Careful selection of target molecules and targeting agents allows for precise delivery to sites of inflammation, thereby maximizing therapeutic drug concentrations at the site of injury. A good understanding of the physiological and pathological determinants of drug and drug carrier pharmacokinetics and biodistribution, may allow for a priori identification of optimal properties of drug carrier and targeting agent. Targeted delivery of therapeutics such as antioxidants and antithrombotic agents to the injured endothelium has shown efficacy in preclinical models, suggesting the potential for translation into clinical practice. As with all therapeutics, demonstration of both efficacy and safety are required for successful clinical implementation, which must be considered not only for the individual components (drug, targeting agent, etc.), but also for the sum of the parts (e.g. the drug delivery system), as unexpected toxicities may arise with complex delivery systems. While the use of endothelial targeting has not been translated into the clinic to date, the preclinical results summarized here suggest that there is hope for successful implementation of these agents in the years to come.

Introduction

The endothelial cells that form a thin monolayer lining the inner surface of blood vessels exert many vitally important regulatory functions. The endothelium controls vascular permeability via dynamic intercellular gaps [1], intracellular fenestrae, and vacuolar pathways initiated in endocytic vesicles [2, 3]. It also regulates: i) blood fluidity, ii) vascular tone, signaling, and angiogenesis; and iii) blood cell trafficking [4–6]. Endothelial damage, pathological activation, and other abnormalities are implicated in the pathogenesis of ischemia, thrombosis, inflammation, tumor growth, hypertension, stroke, atherosclerosis, and other maladies (Table 1). In these conditions, endothelial cells represent important therapeutic targets [7–10].

Table 1.

Examples of pathological conditions involving endothelial dysfunction

Pathological conditions	Role of endothelium
Oxidative stress	Reactive oxygen species are produced by NADPH oxidase in the lumen of endothelial endosomes in response to cytokines, among other endothelial cells (ECs) sources such as mitochondria [26, 27, 5]
Inflammation	Endothelial cells regulate traffic of white blood cells (WBC) to sites of inflammation [28]
Thrombosis	Activated endothelial cells expose numerous procoagulatory molecules [29]
Lung/brain ischemia/reperfusion; lung transplantation	The pulmonary endothelium constitutes an important source and target of reactive oxygen species (ROS) and other damaging agents generated during ischemia and reperfusion [30–33]
Brain ischemia-reperfusion injury	Elevated levels of ROS increase neuronal activity [34]
High blood pressure	Increased oxidative stress and increased amounts of reactive oxygen species in the vascular wall results in quenching of nitric oxide (NO) and impairment of endothelial function and vascular tone [35]

Alas, drugs have no natural affinity to endothelial cells. To overcome this downside, ligands that bind to the endothelium can be used as affinity "vectors" delivering their cargoes to the target cells. Use of optimally devised ligands, configuration of drug delivery systems, and routes of administration allows for concentrating of pharmacological interventions in desirable areas of vasculature, types and phenotypes of endothelia. Furthermore, targeting allows preferential delivery of therapeutic and imaging agents to selected sub-cellular destinations, i.e., to, into, or across endothelium [11–16]. Since the 1980s, several groups have been pursuing this strategy (AKA “vascular immunotargeting") [17–23, 16, 24, 25]. Here we briefly review this actively evolving field, with focus on endothelial targeting of drugs alleviating inflammation, thrombosis, and oxidative stress.

Endothelial Targeting: specific features

The concept of vascular targeting is based on a fairly straightforward idea: drugs and their carriers can be conjugated chemically or via recombinant techniques with the ligands of the endothelial surface markers (Figure 1). These ligands include monoclonal antibodies and their fragments, natural ligands, and peptides [10] [36]. Endothelial proteins tested for targeting include constitutive angiotensin-converting enzyme (ACE) [37, 22, 24], caveolar proteins such as aminopeptidase P2 [38], growth factors, integrins, and transferrin receptors, among others [39]. This approach for the specific delivery of drugs to the endothelial epitope of choice has been extensively studied by several labs [22, 40, 16, 23, 21, 13, 41–44].

Figure 1.

Endothelial determinants for targeted drug delivery. (A) Pan-endothelial molecules clustered in apical plasmalemma, (B) located in the caveolae; (C) Cell adhesion molecules (CAMs); (D) Molecules expressed during angiogenesis or tumor-related; (E) Molecules concentrated in cellular junctions.

In particular, cell adhesion molecules, or CAMs, involved in the migration of leukocytes in sites of inflammation represent good targets for favorable vascular localization of agents alleviating pathologies such as oxidative stress, inflammation, and thrombosis [45]. CAMs include P- and E-selectins, as well as Ig-family members pan-endothelial Platelet-Endothelial Cell Adhesion Molecule-1 (PECAM-1) [46], Intercellular Adhesion Molecule-1 (ICAM-1) [47], and Vascular Cell Adhesion Molecule 1 (VCAM-1) [48]. PECAM-1 is stably and constitutively expressed at high density on the luminal surface of all endothelial cells, whereas ICAM-1 is constitutively exposed on endothelium in the vasculature and is further up-regulated by pro-inflammatory agents, abnormal blood flow, and oxidative stress [49, 50]. Molecules exposed exclusively on activated endothelium (e.g., E- and P-selectins and VCAM-1) represent attractive targets for delivery of drugs and imaging probes to pathological sites in the vasculature [51–54, 14].

Research in the area of therapeutic targeting to CAMs and other molecules on endothelium, has recently intensified with the development of a large number of affinity moieties, including antibodies and antibody fragments (e.g. scFv). In particular, monoclonal antibodies (mAbs) directed to different extracellular epitopes and domains of PECAM-1 have been used as affinity ligands for endothelial targeting of drugs and nanocarriers [55–57] and have shown therapeutic efficacy in both in vitro and in vivo studies. For example, antioxidants conjugated with PECAM-1 antibodies bind to endothelium, accumulate in the pulmonary vasculature, and detoxify reactive oxygen species (ROS) [58].

Pharmacokinetics of Endothelial Targeted Biotherapeutics

Key pharmacokinetic (PK) processes determining biotherapeutic absorption, distribution, metabolism, and elimination (ADME) will control the degree of and time course of delivery of targeted therapeutics to the vascular endothelium. For targeted therapeutics, PK processes can be further sub-divided into non-specific (target-independent) and specific (target-mediated) processes. Delivery of biotherapeutics to the pathologically altered endothelium is controlled by the relative efficiency of target-independent, off-site target-dependent, and on-site target-dependent pathways. While biotherapeutics (namely therapeutic proteins) may be administered extravascularly [59], the focus of this section will be on determinants of biotherapeutic distribution following intravascular administration. Delivery of biotherapeutics may either be via passive mechanisms, taking advantage of favorable physiology in disease sites, or via active targeting, utilizing targeting ligands such as monoclonal antibodies.

Target-Independent Processes

Non-specific distribution and elimination of biotherapeutics is largely controlled by the physicochemical properties of the delivery agent (e.g. size and charge). Following administration, larger biotherapeutics often have a relatively small volume of distribution, which approximates the plasma or blood volume in many cases, due to poor extravasation into tissues [60]. Non-specific distribution of biotherapeutics is largely controlled by convective flow via intercellular pores, thereby key controlling factors in this pathway are the relative size of the biotherapeutic compared to the size of pores present in vasculature [61] and the transvascular fluid flow rate. Intuitively, this pathway is more efficient for peptides and small proteins, as compared to large proteins and particles. It should be noted that the efficiency of this pathway may be altered in disease, due to changes in endothelial physiology. Passive distribution of liposomes can be mediated by liposome-specific parameters, (e.g. surface charge [62]), organ-specific parameters (e.g. leaky vasculature [63] localized temperature gradient [64, 65]), or a synergy between the two (e.g. a drop in local pH effecting a change in the liposome [66]).

Similar to distribution, the elimination pathways for therapeutic proteins are largely controlled by molecular (or particle) size. For peptides and small therapeutic proteins, elimination is often rapid due to efficient renal filtration, hampering delivery to endothelial cells beyond the first pass, due to a short in vivo half-life (t_1/2) [67]. For proteins in the size range of albumin and greater, non-specific elimination is often slower due to poor renal filtration, and is controlled by non-specific pinocytosis and delivery to the lysosome [60]. It should be noted that molecules containing albumin or Fc regions (e.g. antibodies) may have a greater half-life than would be predicted based solely on size, due to interactions with the neonatal Fc receptor (FcRn) [68–70]. Finally, for particulate drug delivery systems (e.g. liposomes), non-specific elimination is often through the cells of the reticuloendothelial system (RES), which typically leads to high levels of uptake in the liver and spleen [71]. Many technologies for modulating RES uptake to reduce non-specific elimination have been proposed, including stealth (e.g. PEGylated) liposomes [72, 73] and liposomes with different charges, sizes, and shapes (e.g. non-spherical) [74]. More efficient elimination of biotherapeutics via any of these non-specific pathways will lead to shorter persistence in vivo and (potentially) reduced targeting to endothelium.

Target-Dependent Processes

Delivery of biotherapeutics to the endothelium is reliant on the specific interaction between targeting ligand and target protein expressed on the endothelium. Specific targeting to endothelium relies on several parameters, including relative target expression in diseased and normal tissues, target trafficking and turnover, and binding affinity and avidity. Interaction between biotherapeutic and target can have a significant impact on pharmacokinetics, leading to saturable tissue distribution and blood clearance, often termed target-mediated drug disposition (TMDD) [75, 76]. Classically, TMDD will manifest as non-linear pharmacokinetics, with more rapid clearance of drug at low doses (sub-target saturation), relative to higher, saturating doses.

Following binding to target, a biotherapeutic can release from the target, remain surface bound, or be internalized. Target-mediated internalization of biotherapeutics can lead to many downstream processes, including recycling, transcytosis, and lysosomal degradation, all of which can alter PK. For those biotherapeutics that are efficiently sorted to the lysosome following binding, target interactions will likely have a significant impact on clearance, when the dose is such that target binding is favored (based on affinity) and below a dose that would saturate the target (based on target expression). Avidity of the targeted therapeutic may also play a significant role in the cellular fate of biotherapeutics, with low avidity agents (e.g. scFvs or mAbs) being processed differently than high avidity agents (e.g. targeted nanoparticles) [77, 78]. While this may lead to rapid elimination of the targeting moiety, any cargo that is not destroyed in the lysosome, and is able to escape the cell, will likely have favorable biodistribution, as it will be preferentially released at the target site.

However, it is important to consider target expression both in the diseased tissue as well as in normal tissues, as this ratio may play a significant role in specific drug delivery. One method to overcome potentially unfavorable disease/healthy tissue target expression ratios would be to administer the drug into blood vessels that flow directly to the diseased organ, taking advantage of first-pass target extraction. Briefly, first-pass extraction refers to the fraction of the dose that is taken up via the first organ seen by the dose, and is most often discussed when considering hepatic extraction of orally administered drugs. A relevant example of this for systemically dosed therapeutics is administering biotherapeutics intravenously for delivery to the pulmonary vasculature, as this will allow the dose to be ‘seen’ by the lung prior to any other tissues that may also express target or clear the biotherapeutic [37]. In situations where rapid and stable target binding is favored, particularly in first-pass extraction, targeting to the pulmonary vasculature will be kinetically favored. This would lead to not only an advantage over specific uptake in normal tissues, but also may be able to counteract poor targeting due to rapid elimination via non-specific processes, such as RES uptake of nanoparticles or renal clearance of small proteins.

Use of the intravenous route to bypass clearance organs such as the liver, directs the entire first pass of drug in blood to the lungs [79]. The pulmonary vasculature represents ~25% of the total endothelial surface in the body and collects the entire cardiac output of venous blood from the right ventricle, whereas all other organs share the arterial output, which is of an equal volume to that seen by the lung alone. A local intra-arterial infusion via a catheter advanced to the conduit vessel favors first-pass carrier interaction with vascular cells in an organ or a vascular area immediately downstream from the vessel [80, 79]. The microvasculature (arterioles, capillaries, and venules) is the preferable target for endothelial nanomedicine. The extended luminal surface area, micron-scale vessel caliber, and low flow rate favor interaction of particles with endothelium in this vascular segment. Hydrodynamic conditions in arteries are less favorable for particle interactions with endothelium than in veins. A significant (if not predominant) fraction of transport from blood to arterial walls occurs from the vasa vasorum, i.e., the microvascular network in the external adventitia layer [81–87].

Impact of Pathology on Targeting

Targeted biotherapeutics are often used to enhance delivery to pathologically altered tissue(s) thus necessitating careful consideration of how disease-mediated changes in physiology may alter pharmacokinetics and biodistribution. Classically, targeting approaches have been used to take advantages of disease-specific upregulation of surface markers, allowing (in theory) for enhanced selectivity of uptake in target sites. Upregulation of target may be through several mechanisms, including de novo protein synthesis, mobilization of intracellular stores of protein, or by unmasking of target proteins.

While these mechanisms are relatively self-explanatory and are readily detected via classical methods (e.g. Western blot), unmasking of proteins may be more challenging to measure, and could occur through changes in molecular conformation, protein localization on the cell membrane, or by removal of barriers preventing access (e.g. glycocalyx). Just as many endothelial determinants increase in expression, exposure, or accessibility in the setting of disease, some constitutively expressed targets (e.g. ACE and thrombomodulin) are lost from the vascular lumen under pathological conditions [88–91].

In addition, it is apparent that pathological changes, both systemic and local, may lead to changes in other determinants of biotherapeutic PK/biodistribution. Local alterations may involve the whole tissue or an organ, although these processes are rarely truly homogeneous, but rather are present as focal heterogeneous patches of abnormal and relatively healthy regions of tissue. A prime example of a spatially heterogeneous condition in which the endothelium is believed to play an important role in disease pathogenesis is that of the acute respiratory distress syndrome (ARDS). A recently reported murine model of ARDS, in which injury was induced within a single lobe of the lung, aimed to reproduce the heterogeneous pathology of this disease in order to study its effect on the biodistribution of ligand targeted therapeutics [92]. The injured lobe of the lung showed changes in protein expression, hypoxic vasoconstriction (decreased blood flow), and enhanced capillary leak (edema), relative to naïve lung. For example, PECAM-targeted liposomes were shunted towards healthy regions of the lung, primarily due to hypoxic vasoconstriction in the injured lobe, while ICAM-targeted liposomes accumulated in the injured lobe, mainly due to injury-mediated upregulation of ICAM-1 expression [68]. Additionally, untargeted particles preferentially accumulated in the injured lobe, due to capillary leak (and increased extravasation). This is but one example of how changes in physiology beyond mere changes in target protein expression could possibly alter PK and targeting.

Mathematical Modeling of Targeting

Development of predictive pharmacokinetic models is an approach that could be useful in understanding mechanisms controlling biotherapeutic pharmacokinetics and distribution. While most work to date has focused on modeling of distribution of targeted proteins, similar principles may be applied for liposomes and other drug carriers. Mechanistic models of targeted drug delivery should incorporate the key physiological and pathological determinants of biotherapeutic PK/biodistribution, which have been discussed above.

A key starting point for development of models for targeted therapeutics would be the general model of TMDD [76], or one of its approximations or extensions [93–95], as these models are structured to incorporate the impact of target binding and turnover on pharmacokinetics. Modeling approaches have been used to generate hypotheses regarding optimal binding properties for antibodies based on target properties [96, 94]. Sensitivity analyses such as those presented by these groups could be useful in the design of new targeting ligands and carriers to increase the probability of maximal targeting and efficacy. Additionally, incorporation of physiological determinants of drug distribution is capable by structuring models in physiologically-relevant manners. Several physiologically-based pharmacokinetic models have been developed for protein therapeutics that are capable of making a priori predictions of PK [97–101]; however, the primary focus of these models has been on prediction of target-independent PK or of delivery to solid tumors. While some models have been developed for untargeted nanoparticles [102, 103], little effort has been put into developing mechanistic models for targeted particles.

To date, there have been very few attempts at developing mechanism-based models to characterize the kinetics of endothelial drug delivery. Recently, we have proposed a semi-physiologic model that was able to predict heterogeneous lung uptake of targeted nanocarriers in a mouse of model of acute respiratory distress syndrome (Figure 2) [92], using similar principles as described for antibody therapeutics. Additionally, sensitivity analyses with this model were in agreement with experimental observations regarding the key mechanisms controlling nanocarrier targeting to the injured lobe of the lung, suggesting the utility of this approach. This type of approach could be useful in identification of novel targets for endothelial delivery and optimal dose levels.

Figure 2.

Mathematical modeling of the impact of local injury on biotherapeutic pharmacokinetics. (A) General structure of a semi-physiologic model for nanocarrier disposition, (B) Lung tissue model for nanocarrier disposition, separating the tissue into injured and healthy regions, and (C) pathological processes involved in altering nanocarrier pharmacokinetics and biodistribution.

Delivery of antioxidants and antioxidant enzymes

Oxidative mechanisms

Cells generate reactive oxygen species (ROS) as a by-product of normal respiration in mitochondria [104, 105] and by several enzyme systems including xanthine oxidase [106] and NADPH oxidases [107] (Figure 3). Moreover, there is growing body of evidence that ROS plays an important roles in many signaling pathways [105] through transcription factors such as Nrf2 [108], NF-κB [109], and others.

Figure 3.

Reactive oxygen species (ROS) in vascular cell signaling pathways and oxidative stress. Superoxide is produced by several cellular enzyme systems including respiratory chain, NADPH-oxidases, xanthine oxidase (XO), cyclooxygenase (COX), etc. It can react with NO decreasing functional NO pool and producing highly reactive peroxynitrate anion ONOO− and. Superoxide spontaneously or by action of superoxide dismutase (SOD) may be reduced into hydrogen peroxide H2O2. Hydrogen peroxide can produce highly reactive hydrogen radical •OH in the presence of transition metals or hypochlorous acid by myeloperoxidase. Catalase and glutathione peroxidases protect cells against hydrogen peroxide. ARDS, acute respiratory distress syndrome; GSHPx, glutathione peroxidases; MPO, myeloperoxidase.

Superoxide radical O₂^·− reacts at an extremely high rate with NO forming a strong oxidant, peroxinitrate (ONOO⁻) that, in turn, causes the inactivation of NO, an important mediator of vasorelaxation and an anti-thrombotic agent [110]. The superoxide radical spontaneously dismutates into hydrogen peroxide H₂O₂, another major ROS. The last reaction is catalyzed by superoxide dismutase (SOD), and enzyme with three isoforms: cytosolic CuZnSOD (SOD1), mitochondrial MnSOD (SOD2), and extracellular SOD (SOD3).

Removing superoxide radical SOD preserves NO^· and blocks ONOO⁻ formation [111]. On one hand, H₂O₂ is able to form the extremely reactive and injurious hydroxyl radical ·OH and hypochlorous acid HOCl in either the presence of free transition metals or by the action of enzymes such as myeloperoxidase (MPO) [112]. On the another hand, H₂O₂ plays an important role as a signaling mediator [113]. The level of H₂O₂ is controlled by catalase that enzymatically converts it into oxygen and water. The source of cellular damage due to oxidative stress in cells may be caused by a deficiency of SOD, catalase, or of other antioxidant systems, and/or by the overproduction of ROS, [114]. Moreover, the abnormal production of signaling ROS may result in cellular dysregulation, causing severe pathologies (Figure 3).

Antioxidant interventions

Reactive oxygen species, particularly O₂^·− or ·OH, have relatively short lifetimes, which determine the spatiotemporal limitations on their signaling and damaging effects. Since these effects are compartmentalized within a short distance of their sites of generation, an effective treatment would require precise delivery of antioxidant drugs on a cellular level [115]. Otherwise, O₂^·− dismutation may be inefficient or may even aggravate the injury through accumulation of H₂O₂ [116]. Many classes of antioxidants have been tested as antioxidant treatments (Table 2).

Table 2.

Examples of antioxidants tested for in vivo use.

Class of antioxidants	Examples	Mechanism of action
Antioxidant enzymes	Superoxide dismutase	Catalytic dismutation of superoxide anion into O2 and H2O2
	Catalase	Disproportionation of H2O2 into O2 and H2O
Flavonoids	Catechins (proanthocyanidine, etc.), flavonols, flavons	Both transition metal chelators and ROS scavengers
Isothiocyanates	Sulforaphane, oltipraz	Phase II enzyme enhancers
Phenolic compounds	BHA, t-BHQ, curcumin, resveratrol	Enhancers of phase II enzymes and conjugating enzymes, eNOS activation
Thiols	N-acetylcysteine, cysteine, dithiols, reduced glutathione	Intracellular redox buffer, metal chelators, radical quenchers
Vitamins	Vitamin C, vitamin E	ROS scavengers
NOX inhibitors	Apocynin, MJ33	Inhibition of ROS generation
SOD mimetics	MnTBAP, EUKs	ROS scavenging
Others	Melatonin, MitoQ	ROS scavengers
	Soluble beta-glucans	Immune response regulation, NO production, MPO inhibition, ROS scavenging
	Selenium (sodium selenite, ebselen, diphenyl diselenide)
	NO and NO donors (S-nitrosothiols, etc.)	GSHPx cofactor
		Free radical scavenging, blocking of ROS generation

^*BHA, butylated hydroxyanisole; t-BHQ, t-butylhydroquinone 5; ROS, reactive oxygen species.

It was demonstrated that increased oxidative stress due to excessive generation of ROS in mitochondria could be significantly attenuated by subcellular delivery of antioxidant drugs to the mitochondria [117]. Membrane-permeable and mitochondria-targeted synthetic SOD mimetics are promising therapeutic agents against mitochondria-derived oxidative stress both in cell culture and in animal models [118, 119]. Furthermore, SOD2/3 chimeric enzymes are able to bind to cell surfaces via the negatively charged glycocalyx and thereby transfer to cells. These chimeric enzymes demonstrated both antioxidant and anti-inflammatory capacities [120]. This class of antioxidant agents may optimally quench cellular oxidative stress through the use of membrane-permeable materials such as peptides [121].

Several antioxidants were tested in cancer therapy [122], including SOD overexpression [123]. However, because non-targeted drugs lack specific affinity to the cell surface and are characterized by non-specific delivery to different tissue and cellular compartments, such approaches often result in incomplete and ineffective delivery of antioxidants [124, 125]. In this review, we focus on targeted endothelial delivery of antioxidant enzymes for treatment of acute oxidative stress or pathological signaling in the vasculature. Although liposomal delivery of oil-soluble antioxidants may improve delivery via improved solubility and tissue retention times, small molecule antioxidants have not shown efficacy in clinical trials. In contrast, as described earlier, antioxidant enzymes are highly potent and specific, detoxifying ROS with high efficacy and rates of reaction.

Targeting antioxidants to endothelium

Delivery of antioxidant enzymes (AOE) to the endothelium may provide suitable protection against oxidative stress, particularly in the pulmonary vasculature, which receives a significant fraction of the injection dose following intravenous (I.V.) administration of drug delivery systems with high affinity to endothelial targets [37]. Local upstream infusion may be optimal for drug delivery to other organs, by precluding depletion of the drug during passage through other organs [80].

A significant obstacle facing the translation of antioxidant therapies into the clinical domain is the inadequate delivery of these agents to their intended site of action. Indeed, the mixed results of several decades of antioxidant research, including large scale clinical trials [126–128], have demonstrated that successful antioxidant therapy must: 1) localize to the cells suffering oxidative stress, 2) effectively detoxify ROS, and 3) do so within an appropriate time-frame for therapy.

One approach to overcome these obstacles is to build SOD and catalase conjugates that are able to bind to cells via hydrophobic or electrostatic interactions. For example, enzymes were coupled to cationic membrane-permeating peptides such as TAT [129]. Fusion constructs of cytosolic CuZnSOD and heparin-binding peptides derived from extracellular SOD showed improved protective effects in animal models of inflammation [120]. Using carriers with affinity to endothelial surface molecules enables more effective and specific targeting of antioxidant enzymes [130].

On the other hand, targeted delivery may alleviate the rapid clearance of antioxidant enzymes by liver and kidney that occurs within minutes after I.V. injection seen with unconjugated catalase and SOD [131, 132]. Several approaches, including conjugation of the enzymes with polyethylene glycol [132], encapsulation in PEG-liposomes [133] or PEG-coated polymeric carriers [134], and conjugation with PEG-based pluronics [135], among other modifications showed increased circulation times of catalase and SOD. Longer blood circulation augmented drug potency of enzymes in some forms of systemic oxidative stress in animals [136, 137]. However, because these carriers lack cellular specificity, they are limited in their efficacy in vivo. Additionally, despite their stealth characteristics, neither PEG-catalase nor PEG-SOD showed any improvement in the setting of in vitro or in vivo delivery to endothelium as compared to non-conjugated enzymes [138].

Antioxidant enzymes antibody conjugates

Antioxidant enzymes SOD and catalase have several advantageous features as agents for protection against oxidative stress, including the lack of requirement for co-factors or external energy input and excellent kinetic properties. For example, one molecule of catalase degrades over 40 million molecules of H₂O₂ per second [139], a level of ROS quenching not attainable even with a large payload of small molecule antioxidants. These enzymes are relatively robust proteins allowing chemical modification without significant enzymatic inactivation. AOE have been conjugated with antibodies to ACE [24], ICAM-1 [47], PECAM-1 [140],, and other epitopes. Enzyme-containing conjugates demonstrated effective, specific binding to the endothelium both in vitro and in vivo [141, 138, 78]. Furthermore, catalase and SOD conjugated to anti-PECAM-1, but not to control IgG protected endothelial cells against the toxic effects of H₂O₂ [78] and O₂⁻ [141], respectively. They were able to inhibit the cellular necrosis and apoptosis induced by oxidative stress [141].

Additionally, antibody-antioxidant enzyme conjugates targeted to ACE, ICAM-1 or PECAM-1, but not control IgG-containing formulations or PEGylated enzymes, were shown to accumulate in the pulmonary vasculature in rats, mice, pigs, and dogs after I.V. injection [24, 47, 138]. Moreover, anti-PECAM/SOD conjugates were specifically shown to accumulate in the intracellular vesicles of pulmonary endothelial cells [138]. The specific endothelial uptake in vivo depends on several parameters of the delivery system including size, shape, and surface charge. Thus, tissue uptake of PECAM-1 directed conjugates was specific for particle sizes from 30 to 500 nm, with the optimal size for anti-PECAM/AOE conjugates being close to 300 nm [142]. While intracellular endothelial delivery is critical for functional activity of the conjugates, the optimal size range shifts towards smaller particles (100–300 nm), since endocytosis of larger particles by endothelial cells is less effective [143].

The functional activity of the injected conjugates was initially validated in animal studies employing an artificial influx of ROS to the pulmonary vasculature. In the first model, lung tissue isolated from rats pre-injected with Ab/catalase conjugate was protected from infusion of H₂O₂, affirming the functionality of delivered antioxidant [144, 78]. In the second model, Ab/catalase, but not PEG-catalase or IgG/catalase, co-injected in mice with anti-TM/GOX (a conjugate that accumulates and generates H₂O₂ in the pulmonary vasculature) decreased oxidative stress in lungs, markedly attenuated edema, and reduced lethality from 100% to <20% [144]. As expected, Ab/SOD did not protect against anti-TM/GOX induced pulmonary injury, since H₂O₂ is the injurious ROS directly produced by GOX [145].

Further functional activity of the delivery of antioxidant enzymes to lungs was validated in several physiologically relevant in vivo models of oxidative stress. PECAM-targeted catalase, but not control formulations, injected in donor rats prior to lung harvest significantly reduced acute oxidative stress, edema, tissue injury, and leukocyte sequestration in lungs transplanted to recipient rats after 18 h of cold ischemia and improved blood oxygenation [144]. The nanocarriers were effective in a model of lung transplantation in pig ameliorating graft functions, including improvement of gas exchange and reduction of intrapulmonary shunt fraction [30]. Similarly, ACE-directed catalase was highly effective in a model of lung transplantation in rats [146]. In an ischemia-reperfusion model, anti-PECAM/catalase formulation attenuated lung injury in situ in ventilated mice [145].

Signaling effects of targeted antioxidant enzymes have also been demonstrated recently in Krev Interaction Trapped Protein 1 (KRIT1)-deficient animals. The intracellular scaffold protein KRIT1/CCM1 regulates microvascular barrier function; its deficiency leads to development of vascular disorder cerebral cavernous malfunction [147, 148]. Loss of KRIT1 causes an increase in ROS generation and exacerbates inflammation-mediated vascular permeability. Tandem delivery of PECAM-targeted antioxidant enzymes SOD and catalase significantly reversed the increase in permeability in KRIT1 deficient mice. Furthermore, localized antioxidant enzymes rescued the redox state and restored responsiveness to TNF-α in KRIT1 deficient arterioles (Goitre, Glading, 2017, Sci. Rep., in press).

The selective role of ROS may be revealed by studying the comparative effects of antioxidant enzymes in different models of injury. In a model of pulmonary ischemia/reperfusion injury, PECAM-targeted catalase, but not SOD demonstrated protection, implicating hydrogen peroxide H₂O₂ as the chief damaging ROS in this injury model [145]. In contrast, PECAM-targeted SOD, but not catalase or free untargeted SOD formulations, inhibited vasoconstriction induced by Angiotensin II in mice, confirming the key role of superoxide O₂^·− produced by endothelial NADPH oxidase in quenching NO [145]. Furthermore, Ab/SOD, but not the Ab/catalase or untargeted free SOD, inhibited pathological endothelial activation induced by cytokines in cells and by LPS in vivo, manifested by changes in expression of VCAM-1 in the pulmonary endothelium in mice [138]. Moreover, studies in cell culture demonstrated that anti-PECAM/SOD particles accumulating in the endosomes quenched superoxide anion produced in the vesicular lumen by NADPH oxidase, thereby demonstrating that internalization of the drug is critical, and as such requires specific targeting that cannot be achieved with untargeted formulations [138, 27].

Liposomal antioxidant enzyme delivery

Liposomes have long been used as a means to deliver antioxidants, protecting them from clearance and deactivation in vivo, and facilitating access to injured or inflamed cells and tissues [149]. The amphiphilic nature of the particle allows for drugs to be entrapped in the inner volume, conjugated to the surface, or intercalated in the bilayer. This enables the delivery of both hydrophilic and hydrophobic species, including lipid-soluble antioxidants incorporated within the hydrophobic bilayer (e.g. vitamin E, carotenoids, flavonoids, soy isoflavones, among many others) [150]. Alternatively, water-soluble antioxidants such as ascorbate, urate, and glutathione have been studied as liposomal cargoes, enclosed in the hydrophilic core of the liposome. Inclusion of tocopherol and other oil-soluble antioxidant vitamins into the lipophilic inner membrane of liposomes has been shown to lend greater antioxidant efficacy, while also serving to protect the light and oxidation-sensitive molecules from damage by environmental exposure [151].

Studies in a rat arthritis model showed that SOD liposomes inhibited edema more effectively than naked SOD [152]. The incorporation of an acetylated hydrophobic derivative of SOD, Ac-SOD, improved loading efficiency of the enzyme compared to unmodified SOD. Since the Ac-SOD localized to the liposomal bilayer and 50% of the enzyme was exposed to the exterior, the site of activity was focused at the surface of the liposome, or ‘enzymosome’, instead of within the aqueous interior. The change in conformation reduced the effect of release rate on the activity of the liposome and increased bioavailability of the enzyme [153]. This effect of SOD localization within the liposome structure was tested in a rat adjuvant arthritis model comparing PEG-coated liposomes with either Ac-SOD or plain SOD. The circulation time of the PEGylated liposomes increased regardless of the SOD type included, and a faster anti-inflammatory effect was observed with the Ac-SOD PEG liposomes versus the plain SOD PEG liposomes [154]. Varying methodology and content of liposomal SOD formulations allows for modulation of the loading efficacy, localization within the liposomal compartments, and enzymatic activity of resultant SOD/liposomes [155].

NADPH-oxidase inhibitor MJ33 loaded into liposomes targeted to the endothelium by PECAM antibody accumulated in the endothelial cells, inhibited ROS production, and provided more potent protective effects than non-targeted counterparts against oxidative stress and inflammation in vitro, ex vivo in perfused mouse lungs, and in mice in vivo [156]. Similarly, targeted liposomes loaded with EUK-134, a superoxide dismutase/catalase mimetic, bound to endothelial cells in culture and in pulmonary vasculature in mice and provided anti-inflammatory effects in a mouse model of endotoxin lung injury, whereas untargeted IgG/EUK liposomes provided neither delivery of cargo to endothelium nor protection [157].

ICAM-1 and VCAM-1 targeted echogenic liposomes filled with NO have been used to detect inflammatory changes in atherosclerotic lesions using contrast-enhanced ultrasound imaging. Acoustic enhancement was enabled by the pretreatment of the endothelium with targeted NO-loaded echogenic liposomes combined with localized ultrasound activation, providing improved detection of site-specific inflammatory changes [158, 159].

Potential benefits of peripherally delivering an opioid using ICAM-targeted loperamide HCl loaded liposomes loaded were studied in rats. The anti-inflammatory and analgesic effects, as well as particle biodistribution localized focally to the inflamed tissues, avoiding brain accumulation, producing “highly significant analgesic and anti-inflammatory effects over the 48 hours”. This approach directs pain control to specific injured tissues, highlighting a potentially clinically relevant role for peripheral anti-inflammatory opioid targeting [160].

Other carriers and perspectives

An advanced approach for antioxidant enzyme delivery is to encapsulate AOE into porous nanocarriers permeable to small molecule enzyme substrates that prevent access to the enzymes by proteases, thereby protecting the protein drugs from degradation. Freeze-thawing double emulsion technique allows encapsulation of catalase into spherical polymer nanocarriers (200–400 nm diameter) based on PEG-PLGA and similar di-block copolymers, permeable to H₂O₂ (but not to superoxide), and protecting catalase from proteases [161]. Enzyme PEGylation may increase the protection against proteases, while modulating the molar ratio and size of PEG and PLGA chains in the copolymer controls the shape of nanocarriers, allowing geometries ranging from spherical to filamentous [162]. Another example of a protective nanocarrier is porous polymersomes that allow loading of SOD inside the particle providing access to superoxide, while preventing SOD degradation [163].

PECAM-targeted PEG-PLGA nanocarriers loaded with catalase provide effective endothelial delivery in vitro and in vivo, and prolonged antioxidant protection of the endothelium [164]. Encapsulation of either catalase or SOD into micelles formed by controlled precipitation of magnetic nanoparticles using calcium and oleate provides composite nanocarriers (200–300 nm diameter) containing active catalase or SOD accessible to either H₂O₂ or superoxide, and protected from proteases [165].

Traumatic brain injury is another pathological condition that may benefit from antioxidant enzyme delivery. Oxidative stress occurs within minutes of mechanical impact, playing a critical role in secondary injury after trauma, including inflammation, excitotoxicity, and cell death [166]. Accordingly, antioxidant therapy was proposed as a potential therapeutic intervention in brain trauma [167]. Indeed, targeting catalase to ICAM demonstrated significant protective effects in a model of traumatic brain injury [168]. ICAM-targeted catalase reduced both total ROS generation (including specific production of H₂O₂) and a marker of oxidative stress, 3-nitrotyrosine. The delivered enzyme also preserved the integrity of the blood-brain barrier, which was severed by traumatic tissue damage, and attenuated neuropathological indices. Moreover, targeted catalase significantly decreased injury-induced pro-inflammatory activation of microglia [168].

The last two decades have demonstrated significant progress in the formulation of delivery systems carrying antioxidant enzymes and their applications to pathologies with a critical role of oxidative stress such as organ transplantation, ischemia-reperfusion, or inflammation. In the future, tests of new target molecules as well as new delivery system,s or optimization of current platforms should be important steps in its imminent translation into the clinical domain, thereby improving management of disease conditions involving this pathological mechanism.

Endothelial targeting of antithrombotic agents

Most pharmacologic efforts in this area have centered on systemic administration of endothelial-derived agents, rather than attempt to directly alter the function of the monolayer. Examples include small molecules, such as prostacyclin (epoprostenol) and nitric oxide (the active metabolite of nitroglycerine, sodium nitroprusside, and isosorbide mononitrate), and a number of biotherapeutics, including tissue plasminogen activator (tPA), activated protein C (APC), tissue factor pathway inhibitor, and antithrombin [169–174]. While some of these drugs have found clinical application in select settings, this overall strategy has been plagued by narrow therapeutic indices and life-threatening toxicity. These disappointing clinical results, in spite of enormous investment of effort and resources, suggest that systemic infusion may simply be poorly capable of replicating the tightly regulated and highly localized production of endogenous endothelial-derived molecules.

Over the past twenty years, our laboratory and others have worked to develop an alternate strategy, in which antibodies or other immunologic ligands are used to direct therapeutic proteins to ECs themselves, in the hope of reconstituting or boosting their endogenous protective pathways [175, 176]. Initial efforts focused on delivery of tissue plasminogen activators (tPA), potent fibrinolytic enzymes approved for the treatment of myocardial infarction and acute ischemic stroke [177]. Motivated by the well-documented limitations of the currently available tPAs [178], the enzyme was cross-linked to antibodies specific for endothelial cell adhesion molecules, with the intention of localizing its fibrinolytic activity to specific vascular beds [179]. While theoretically attractive, this approach ultimately encountered a number of practical limitations, including a relative lack of vascular-bed specific determinants and the rapid internalization of endothelial-targeted multivalent antibody-drug complexes. These considerations led to the development of a new series of protein therapeutics, which incorporate single chain variable fragments (scFv) [80, 55]. These monovalent protein affinity ligands can be genetically fused to protein cargo and are generally believed to have a lower rate of internalization following binding to surface targets, making them attractive options for delivery of protein cargo to the luminal surface of ECs.

Advantages of using scFv and small affinity peptides include: i) lack of side effects mediated by Fc-fragment of IgG, including activation of complement and Fc-receptor bearing cells; ii) lack of endothelial activation and internalization induced by CAM cross-linking; iii) established techniques for humanization and affinity maturation; and iv) modular recombinant format for insertion of mutations endowing products with novel, favorable pharmacokinetics and/or functional features.

To date, the majority of efforts have focused on scFv fusion proteins as a means to augment the endothelial protein C (PC) pathway [55]. Unlike plasminogen activators, which are secreted by ECs, the key components of this anti-thrombotic, anti-inflammatory, and vascular barrier-stabilizing system are membrane-bound glycoproteins, which function at the endothelial surface. Thrombomodulin (TM) binds thrombin, changes its enzymatic specificity, and inhibits its pro-thrombotic and edematogenic activities [180]. TM then partners enzymatically with a second protein, the endothelial protein C receptor (EPCR), which optimally positions PC for thrombin-induced cleavage and activation [181]. Loss of endothelial TM and EPCR, which has been demonstrated in multiple disease states, results in a relative imbalance between thrombin and APC, a vasculoprotective protease previously developed as a recombinant biotherapeutic and approved for the treatment of severe sepsis in humans [182, 183]. As an alternative to systemic infusion of APC, a murine PECAM-1 specific scFv/TM fusion protein was used to reverse the underlying loss of endothelial TM and demonstrated protective effects in multiple mouse models of pulmonary vascular injury, while imparting a lower risk of spontaneous bleeding [90].

While PECAM-1 is an attractive target from the standpoint of high-level expression and pan-endothelial distribution, further investigation has shown that a scFv/TM fusion protein bound to this cell adhesion molecule does not interact with EPCR in the same way as endogenous TM [24]. Subsequent work has produced two distinct strategies of fusion protein-mediated intervention: one in which TM is anchored to endothelial ICAM-1, which allows superior interaction with endogenous EPCR, and the other in which TM and EPCR are co-delivered to PECAM-1, enabling enzymatic partnering between the recombinant proteins (Figure 4) [184, 185]. The latter strategy utilizes paired PECAM-1 specific scFv, which demonstrate the collaborative enhancement effect described in detail later in this review.

Figure 4.

Endothelial targeting to boost endogenous antithrombotic and anti-inflammatory pathways. The endothelial protein C (PC) pathway, which centers on the membrane glycoproteins, thrombomodulin (TM) and endothelial protein C receptor (EPCR), is a critical regulator of inflammation, thrombosis, and vascular permeability. Infusion of soluble biotherapeutics, like soluble human TM (shTM) and recombinant human APC (rhAPC), suffers from poor PK and lack of interaction with key components of the endogenous system. Anchoring recombinant TM and/or EPCR to endothelial cell adhesion molecules, like PECAM-1 and ICAM-1, localizes therapeutic action to the surface membrane and may exert superior effects through restoration of natural anticoagulant effects and signaling through protease-activated receptors (PARs). For these purposes, scFv fusion protein biotherapeutics may be preferable to large monoclonal antibody (mAb)-protein complexes, which can crosslink CAMs and induce endosomal uptake and endothelial disruption.

The most recent development has been the creation of scFv/TM fusion proteins incorporating human TM and human specific or species cross-reactive scFv [186] [187]. These agents allow confirmation of therapeutic effects in humanized systems, an important step in motivating further pre-clinical development. In particular, a human ICAM-1 targeted scFv-thrombomodulin fusion protein was tested and found to reduce intravascular fibrin deposition in a human whole blood microfluidic model of tissue factor-driven, inflammatory thrombosis. In addition to proving more effective than other candidate therapeutics (soluble TM and tissue factor inhibitory antibody) in this model, the fusion protein demonstrated a synergistic anti-thrombotic effect when paired with supplemental PC, suggesting this combination as a novel translational therapeutic strategy [187].

Dual targeting to vascular endothelium

In addition to prevalent delivery tactics that utilize a “one type-ligand” targeting, a recently discovered phenomenon known as Collaborative Enhancement of Paired Affinity Ligands (CEPAL) offers the possibility of dual therapeutic targeting to vascular lumen. CEPAL is described as an increased binding of paired antibodies directed to adjacent, yet distinct epitopes of PECAM-1 [188], and is observed for both mouse and human proteins. The effect is shown to have a comparable amplitude in various systems, including cells expressing mutant form of PECAM-1 unable to form homodimers and soluble recombinant PECAM-1, which indicates that CEPAL does not interfere with cellular functions or require homophilic PECAM-1 interactions [189]. Interestingly, PECAM-1 antibodies enhance binding of multivalent nanocarrier spheres coated with paired antibodies, and the effect is modulated by nanoparticle avidity, epitope selection, and flow (Figure 5A–C) [190].

Figure 5.

Proposed model of CEPAL to PECAM-1. Proposed hypothetical model of CEPAL to PECAM-1. (A and B) Targeting of drug conjugates with antibody mAb1 (a) or mAb2 (B). Collaborative enhancement of antibody targeting observed in CEPAL. (D) Examples of in vivo endothelial targeting: Lung to blood ratios in biodistribution of anti-muPECAM-1 [125I]-mAbs 390 and MEC13.3 [1] (left), coadministration of Mec13 scFv/EPCR with 390 scFv/TM increases pulmonary targeting of 390 scFv/TM [12] (right).

One possible explanation of CEPAL effect is conformational changes in the PECAM-1 molecule are induced after binding of antibodies, which leads to the exposure of a partially hidden, cryptic epitope. Allosteric conformational changes in the antibody molecule after binding to proteins was studied by using multiple experimental methods, including crystallography, surface plasmon resonance, and isothermal titration calorimetry. On the other hand, these antibody-antigen interactions lead to the formation of new protein-protein complexes that also may affect an antigen itself, as it is experiencing conformational changes as well (27). As a result, the antigen molecule may now expose previously hidden regions that are now more accessible for binding of other antibodies (Figure 5D).

Collaborative enhancement effect has been confirmed in vivo, as demonstrated by improved pulmonary accumulation (2-fold) of intravenously administered radiolabeled PECAM-1 mAb when co-injected with an unlabeled paired mAb in mice [188]. Surprisingly, the effect is even more pronounced in the case of delivery of relatively large multivalent NCs coated with PECAM antibodies co-injected with free antibodies (8.5-fold) [190]. Dual targeting by means of CEPAL is an attractive strategy for “partner” drugs. Agents for which biologic effect is enhanced in enzymatic reaction benefit the most when delivered in close proximity on the surface of the vascular lumen. An example of such drugs includes two aforementioned endothelial membrane proteins: TM and EPCR. Biotherapeutic fusion proteins consisting of either recombinant TM or EPCR linked to scFvs of paired anti-PECAM-1 antibodies have shown therapeutic efficacy in endothelial cell culture and in mice [184]. Co-treatment of cells with scFv/TM and scFv/EPCR caused a significant increase in APC generation via two mechanisms: increased binding and increased enzymatic partnering of the drugs. In mice, co-delivery of fusion proteins scFv/EPCR and scFv/TM had a synergistic effect, as shown via enhanced pulmonary targeting and increased generation of APC (Figure 5E).

Modulation of endothelial targeting by using paired antibodies to adjacent epitopes of one molecule, or CEPAL, has a critical advantage for tandem agents. As it was shown to be a feasible strategy in vascular targeting of fusion proteins as well as nanocarriers, there is a need for further investigation of CEPAL’s full potential. Furthermore, investigation of other molecules where CEPAL exists and the mechanisms behind it represents a scientific challenge. Induced binding of monoclonal antibodies was previously reported in the works of Lubeck et al [191] for hemagglutinin molecule of influenza A/PR/8/34 and Towbin and colleagues [192] for Interleukin-1β. All described molecules have a diverse nature and functions, which may be a hint of a more generalizable law of immune response.

Cautionary notes: first do no harm

The goal of vascular targeting is to improve management of maladies in human patients. As this tentative translational strategy approaches (and in few studies enters) the clinical domain, considerations of the benefit/risk ratio come to the forefront of the agenda. In particular, the risk of inflicting adverse effects remains, arguably, to be addressed with the rigor and depth commensurate for the clinical translation of any current strategy for targeting drugs to vascular endothelium.

In most clinical scenarios, drug delivery to endothelium should be free of adverse effects on the target cell and other cell types taking up the drug (e.g., phagocytes, renal and hepatic cells), as well as free of systemic side effects such as activation of complement and other host defense systems in the bloodstream. Noteworthy, the biocompatibility of the drug delivery system is not necessarily equal to that of its components [193]. Loading a relatively safe agent into a relatively safe carrier decorated by innocuous ligands may yield a toxic product.

Ligands and especially ligand-driven carriers may activate endothelial cells, induce shedding and/or internalization of target determinants, or otherwise disturb the endothelium. For example, targeting to thrombomodulin, a very useful model in animal studies [194, 195], is unlikely to find clinical use because of the high risk of thrombosis and inflammation [196] caused by inhibition of thrombomodulin’s protective functions [180]. Inhibition of endothelial enzymes ACE and APP results in elevation of the level of one of their common peptide substrates, bradykinin, which may lead to side effects associated with enhanced vascular permeability, a known and generally tolerable side effect of ACE inhibitors.

The criteria of safety are different in targeting tumors or tumor endothelium as opposed to targeting drugs for management of cardiovascular, pulmonary, neurological and metabolic maladies [197]. Toxic effect to the tumor cells is often viewed a bonus, whereas the specificity of targeting must be maximal to avoid collateral damage. In contrast, endothelial disturbance must be minimized in non-tumor applications to avoid aggravation of oxidative stress, inflammation, and thrombosis. However, the criteria of specificity are less stringent in this case, because drugs alleviating these conditions (often associated with systemic pathologies) are expected to be less likely to cause systemic harmful effects; therefore, pan-endothelial delivery of antioxidant, anti-inflammatory, or anti-thrombotic agents throughout the vasculature is a suitable option.

Conclusions and perspectives: we are not there yet, but will be there

Almost two decades have passed since the initial prototype studies of drug targeting to endothelial adhesion molecules. Types of cargoes delivered to endothelial CAMs using diverse drug delivery systems have included small chemical drugs [198], biotherapeutics, and imaging agents [199]. Drugs conjugated with anti-CAM exert therapeutic effects superior to untargeted drugs in cell cultures [200–202, 78], isolated perfused organs [47, 144, 78], and animal models of human pathology [144] [144]. A short summary of preclinical outcomes for select endothelial-targeted drugs is given in Table 3.

Table 3.

Preclinical outcomes for select endothelial-targeted drugs

Therapeutic	Delivery Mechanism	Outcome
Superoxide dismutase	Anti-PECAM mAb conjugate	Inhibited LPS-induced VCAM expression in mouse lung [138]
	Anti-PECAM mAb conjugate + NO donor	Suppressed LPS-induced VCAM, TNF-α, and MIP-2 in mice [27]
	PECAM-targeted nanocarriers	Mitigated endotoxin-induced lung inflammation [203]
Catalase	Anti-PECAM mAb conjugate	Reduced in alveolar edema and attenuated arterial oxygen decline in lung ischemia/reperfusion [145]
		Protection from glucose oxidase induced lung injury [145, 144]
		Reduced oxidative stress, ischemia/reperfusion injury, and improves function in rat lung transplant [144]
		Improved outcomes in pig lung transplant [30]
	PECAM-targeted nanocarriers	Alleviated pulmonary edema and leukocyte infiltration in LPS-induced lung injury [203]
	Anti-ICAM mAb conjugate	Reduced markers of oxidative stress, preserved blood-brain barrier integrity, and improved neurological outcomes in traumatic brain injury [168]
	Anti-ACE mAb conjugate	Reduction in H2O2-induced ACE shedding, lung edema, tracheal and pulmonary arterial pressure in perfused rat lungs [47]
		Reduced inschemia-reperfusion induced increases in endothelin-1 and iNOS and enhanced arterial oxygenation [146]
MJ33	PECAM-targeted liposomes	Reduced endothelial permeability and suppressed ROS production in LPS-challenged mice [156]
EUK-134	PECAM-targeted liposomes	Provided >60% protection against pulmonary edema in LPS-challenged mice [157]
Loperamide HCl	ICAM-targeted liposomes	Provided analgesic and anti-inflammatory effects in a rat paw inflammation model [160]
Tissue Plasminogen Activator	Anti-ICAM mAb conjugate	Disolved fibrin microemboli in rat lungs [204]
Urokinase	RE8F5 mAb Conjugate	Enhanced thrombolytic potency in a rat model of pulmonary embolism [205]
Thrombomodulin	Anti-PECAM scFv fusion protein	Reduced fibrin deposition and leukocyte infiltration in ischemia-reperfusion [90]
		Suppressed lung inflammation and edema following LPS-challenge [90]
	Anti-ICAM scFv fusion protein	Reduced MIP-2, VCAM, and E-selectin expression and improved endothelial barrier function in LPS-challenged mice [185]
Thrombomodulin/EPCR	Paired anti-PECAM scFv fusion proteins	Decreased VCAM expression and improved endothelial barrier function in LPS-challenged mice [184]

Abbreviations used in table: PECAM (Platelet-Endothelial Cell Adhesion Molecule), ICAM (Intercellular adhesion molecule), VCAM (Vascular cell adhesion molecule), mAb (Monoclonal Antibody), scFv (Single Chain Variable Fragment), MIP-2 (Macrophage Inflammatory Protein 2), EPCR (Endothelial Protein C Receptor), LPS (Lipopolysaccharide), TNF-α (Tumor Necrosis Factor alpha), NO (Nitric Oxide), ACE (Angiotensin-Converting Enzyme), iNOS (Inducible Nitric Oxide Synthase)

Drug targeting includes phases of molecular recognition and anchoring, followed by either residence on the plasmalemma or internalization, concluded by elimination. These complex processes are controlled by features of the target (including surface density and accessibility of the anchoring determinant molecules and their epitopes, parameters of flow, and functional and phenotypic characteristics of the cell), as well as features of the ligand (affinity, number and accessibility of binding sites), its configuration in the drug delivery system (valence, surface density, interactive freedom), and features of the drug delivery system (size, shape, PK). The immense multitudes of these entanglements dictate efficacy and the benefit/risk ratio of given approaches for vascular drug delivery.

Endothelial targeting of drugs has been achieved in animals and in perfused human organs, delivering agents to diverse endothelial compartments. Careful selection of targets, features of the carriers, such as valence, surface density of ligands, and carrier geometry, pro-drug activation features, and encapsulation into protective polymeric carriers provide powerful tools for the control of sub-cellular addressing, activation, and duration of the effects of the cargoes.

Scaling-up synthesis and quality control of targeted drug delivery systems with a standard, FDA-acceptable level of homogeneity is a challenge for translation of these delivery systems into the clinical domain. Recombinant mutant pro-drugs fused with scFv may be more amenable this development. Established expression systems enable large-scale, GMP-production of homogeneous monovalent scFv/PA fusions [206, 207].

Prophylactic application of endothelial-targeted drug delivery is well suited for treatment of ischemia in organ transplantation or in cardiopulmonary bypass. Anti-PECAM scFv achieves peak pulmonary targeting in a few minutes of IV delivery [55], permitting injection into a donor prior to organ removal; hence, a recipient will receive only the drug bound and eventually metabolized within the graft, which further boosts safety and selectivity. It is tempting to hope that in the next decade targeted interventions into endothelial cells will be translated from successes reported in animal studies to medical practice.