I. Introduction

Top Previous Next References

Macrophages are widely distributed and strategically placed in many tissues of the body to recognize and clear altered and senescent cells, invading particulates, as well as macromolecular ligands via a multitude of specialized plasma membrane receptors (Gordon, 1995). This propensity of macrophages for endocytosis/phagocytosis of foreign particles in the past has provided an opportunity for the efficient delivery of therapeutic agents to these cells with the aid of colloidal drug delivery systems (usually in the form of liposomes, polymeric nanospheres, micelles, and oil-in-water emulsions), following parenteral administration (Poznansky and Juliano, 1984). For example, macrophages that are in contact with blood or lymph serve as sites of proliferation for certain microbes during some or all of the infection process. Numerous logical strategies for using colloidal drug carriers in the treatment of certain infectious diseases involving macrophages have been developed, based on analysis of the pathogenesis of microbial diseases (Alving, 1988; Agrawal and Gupta, 2000). For instance, the proposed lysosomotropic-parasitotropic process for delivery of liposome-encapsulated drugs to Leishmania within macrophages involves uptake of the phagocytosed liposome and delivery to a lysosome, fusion of lysosome with the parasitophorous vacuole, and delivery of the liposome to a lysosome within the parasite (Alving, 1988). Ambisome (Gilead, Boulder, CO), a liposomal formulation of amphotericin B, is an example of a world-wide marketed drug delivery product available for treatment of visceral leishmaniasis or confirmed infections caused by specific fungal species.

Some pathogens have also evolved ways of evading intracellular killing to survive within the cell; one example is by escaping the phagosome, which is followed by direct entry into cytoplasm (Provoda and Lee, 2000). To allow further cytoplasmic delivery, new strategies have been developed. A classical example is the use of a pH sensitive liposome (Horwitz et al., 1980). Such vesicles maintain stable phospholipid bilayers at neutral pH or above but destabilize and become fusion-competent at the acidic pH of endosomes and subsequently release their entrapped contents first into endosomes and then into cytoplasm. However, the efficiency of cytoplasmic delivery has further been enhanced by employing natural mechanisms that breach the endosomal membrane once the liposomal contents are delivered into the endosomal lumen. For example, Lee et al. (1996) have adopted a strategy used by a facultative intracellular pathogen (Listeria monocytogenes) that mediates bacterial passage from the phagosome into cytosol. After being internalized into phagosomes, L. monocytogenes permeabilizes phagosomal membranes with the aid of listeriolysin O to enter the cytosolic space of host cells. Indeed, coencapsulation of listeriolysin O along with protein-based antigens in liposomes resulted in efficient delivery of protein antigens to the cytosolic pathway of antigen processing and presentation in macrophages. This approach seems highly desirable in vaccine production to polysaccharide antigens that are poorly immunogenic. Similarly, breaching of the endosomal membrane is also critical for survival of DNA after delivery to macrophages.

Metabolic processes within the mature macrophage have also served as targets for drug carriers. These included degradation of accumulated macromolecules within lysosomes (storage diseases) and iron overloading (Poznansky and Juliano, 1984). Apart from therapeutic goals, colloidal carriers have proved to be useful for diagnostic purposes, for example, to assess macrophage phagocytic and clearance functions. Similarly, particulate colloids tagged with a suitable radiopharmaceutical or contrast agent were shown to be helpful for imaging certain pathologies (e.g., deep-seated tumors) not via targeting but through macrophage loading (Seltzer, 1989; Tilcock, 1995; Kostarelos and Emfietzoglou, 1999). By this approach, it is the surrounding parenchyma, and not the pathology, that will change in intensity.

The rapid sequestration of intravenously injected colloidal particles from the blood by hepatic midzonal and periportal Kupffer cells is problematic for efficient targeting of drug carriers or diagnostic agents to a desired macrophage population (e.g., splenic red pulp macrophages) as well as to nonmacrophage sites. As a result, there has been a growing interest in the engineering of colloidal carrier systems that upon intravenous injection avoid rapid recognition by Kupffer cells and adequately remain in the blood. It is the aim of this article to critically discuss and evaluate the developments in the field of long-circulating and target-specific colloidal carrier systems for intravenous administration (and in selected cases with regard to subcutaneous injection). We have viewed the application of long-circulating particles in sustained-release of drugs within vasculature, transfusion medicine (artificial erythrocytes), and targeting. The latter is defined either as passive or active. With regard to passive targeting, this article identifies and discusses the opportunities afforded by certain physiological and pathological conditions that are amenable for treatment or diagnosis by using long-circulating colloidal carriers. In relation to active targeting, attention is focused toward identification of potential and accessible tissue-specific antigens and their ligands. For particle engineering, we have taken a lead from nature. That is, exploring the surface mechanisms that afford red blood cells long-circulatory lives and the ability of certain pathogenic microorganisms to evade macrophage recognition. Our analysis is centered where such strategies have been translated and fabricated to design a wide range of particulate carriers (e.g., nanospheres, liposomes, micelles, oil-in-water emulsions) with prolonged circulation and target specificity. Throughout our approach, we critically evaluate these engineered systems on the basis of biodistribution, target specificity, therapeutic end-points, biological activities, and toxicity.

	II. Theoretical Applications of Long-Circulating Particulate Carriers in Experimental and Clinical Medicine

Top Previous Next References

Similarly, one can also foresee long-circulating nanoparticles as carriers of radiopharmaceuticals or contrast agents for use in the imaging of vasculature. These may include blood-pool imaging, detection of vascular malformations, and gastrointestinal bleeding. Technetium-99m-labeled red blood cells are used routinely in nuclear medicine for blood-pool imaging, but technetium-99m is not strongly bound to the red cell and dissociates, resulting in the clearance of free technetium through the kidney and bladder (Phillips, 1998). This dissociation greatly interferes, for example, with the detection of sites of bleeding in the lower abdomen. This, together with the potential hazards associated with blood handling, justifies efforts for the development of synthetic long-circulatory particles in nuclear medicine.

The unique structural changes associated with a given vascular pathophysiology could also provide opportunities and insights for the use as well as engineering of long-circulating particulate carrier systems. For example, particle escape from the circulation is normally restricted to sites where the capillaries have open fenestrations as in the sinus endothelium of the liver (Roerdink et al., 1984) or when the integrity of the endothelial barrier is perturbed by inflammatory processes (e.g., rheumatoid arthritis, infarction, infections) (Turner and Wright, 1992) or by tumors, although such defects are not a consistent feature of all tumors (Jain, 1989; Hobbs et al., 1998). In the liver, the fenestrae-associated cytoskeleton controls the hepatic function of endothelial filtration, where the size of fenestrae can be as large as 150 nm (Braet et al., 1995). The size of fenestrae in certain inflammatory vessels as well as tumor capillaries can be up to 700 nm. Currently, there is evidence in support of liposome extravasation to hepatic parenchyma (Roerdink et al., 1984) as well as increased capillary permeability to liposomes and polymeric nanospheres (in the size range of 50-200 nm) during inflammation (Boerman et al., 1997a,b; Dams et al., 1998, 1999, 2000b) and in specific cancers (Poznansky and Juliano, 1984; Yuan et al., 1994; Hobbs et al., 1998; Drummond et al., 1999) based on observations in experimental animals and in humans. However, the efficiency and kinetics of particle delivery varies from one model to another. Particles of less than 10 nm can also leave the systemic circulation through the permeable vascular endothelium in lymph nodes (Wiessleder et al., 1994; Moghimi and Bonnemain, 1999). The sinus endothelium of bone marrow is also capable of removing small-sized particles from the systemic circulation. Here, the sinus endothelium can remove particles from the circulation by both transcellular and intracellular routes (Moghimi, 1995a). The transcellular route is through the diaphragmed fenestrae of endothelial walls, whereas the intercellular route is associated with the formation of bristle-coated pits containing matter on the luminal surface of the endothelium. However, regional dilation increased branching of sinuses, and even complete loss of bone marrow sinus endothelium can occur in certain pathological conditions thus resulting in particulate accumulation in affected areas. For instance, marked fragmentation of endothelial cytoplasm, as well as complete loss of the endothelium of bone marrow sinusoids, have been reported for dogs infected with a canine hepatitis virus (Linblad and Bjorkman, 1964).

Certainly one would expect this because of the prolonged residence in blood, long-circulating carriers with the appropriate sizes have a better chance of reaching the above-mentioned targets, resulting in improved treatment or diagnosis. Another topic that deserves investigation is the use of a tissue-specific pharmacological mediator that, by opening intercellular connections in selected vessels, may facilitate long-circulating colloidal entities to escape from the circulation and extravasate into the tissues of organs whose postcapillary endothelia contain receptors for the mediator (Rosenecker et al., 1996). In spite of the preceding evidence, unless the means are found to actively control local permeability and access, the issue of tissue selectivity is not properly addressed by the use of engineered nanoparticulate complexes, and significant therapeutic gain is unlikely.

Given the potency (and toxicity) of modern pharmacological agents, tissue selectivity is a major issue. In the delivery of chemotherapeutic agents to solid cancers, this is particularly critical, since the therapeutic window for these agents is often small and the dose-response curve is steep. Therefore, the idea of exploiting the well documented vascular abnormalities of tumors, restricting penetration into normal tissue interstitium while allowing freer access to that of the tumor, becomes particularly attractive. This has been the subject of considerable research; however, the data suggest that the solution to this problem will depend upon more than simply reducing mechanical barriers to trans-endothelial passage. Let us examine the idea of passive targeting more closely with regard to the process of extravasation of long-circulating particles from the blood to a solid tumor, a target that has received the most attention in the drug delivery field. A solid tumor comprises two major cellular components. These are the tumor parenchyma and the stroma, which incorporates the vasculature and other supporting cells. To meet the metabolic requirements of the expanding population of tumor cells, the pre-existing blood vessels become subject to intense angiogenic pressure (Folkman, 1995). Several factors produced by tumor cells are believed to signal the development of new capillaries from these vessels (angiogenesis). Scanning electron microscopy of microvascular corrosion casts has allowed visualization of the geometry of tumor architecture. From these studies, it has become apparent that tumor vessels are highly irregular and show gross architectural changes (e.g., the presence of interrupted endothelium and an incomplete basement membrane) with increased blood vessel tortuosity as well as abnormal and heterogeneous vessel density (Jain, 1988). As a consequence of these abnormalities, there may be profound physiological changes in blood flow within tumors and in transport properties of the tumor vessel. Due to the vascular heterogeneity of solid tumors, the blood flow, and consequently tissue oxygenation, tends to be nonuniform. Therefore, solid tumors usually contain well perfused, rapidly growing regions, and poorly perfused, often necrotic areas (Jain, 1988). As in normal tissues, diffusive and convective forces govern the movement of molecules into the interstitium of tumors. However, diffusion is believed to play a minor role in the movement of solutes across the endothelial barrier compared with bulk fluid flow. Examination of pressure gradients in experimental tumors has suggested that the movement of drugs and particulates out of tumor blood vessels into the extra-vascular compartment is remarkably limited (Jain, 1989). This has been attributed to a higher-than-expected interstitial pressure, in part due to a lack of functional lymphatic drainage, coupled with lower intravascular pressure. In addition, interstitial pressure tends to be higher at the center of solid tumors, diminishing toward the periphery, creating a mass flow movement of fluid away from the central region of the tumor.

These pathophysiological characteristics have serious implications for the systemic delivery of not only low-molecular weight drugs and macromolecules (e.g., antibodies, polymer conjugate therapeutics), but also particulate delivery vehicles (e.g., liposomes), and simply enhancing the plasma half-life of these agents will not necessarily lead to an increase in therapeutic effect. However, it is likely that long-circulating particles may have a better safety profile in terms of normal tissue toxicity than free anticancer agents, as well as anticancer drugs encapsulated in phagocyte-prone carriers. The utility of conventional colloidal carriers as vehicles for drug delivery in cancer treatment is inappropriate since the class of drug being used is able to induce apoptosis in macrophages of the RES (Daemen et al., 1995). In the liver, restoration of Kupffer cells may take up to periods of 2 weeks (Daemen et al., 1995). A potentially harmful effect is that bacteriemia may occur during the period of Kupffer cell deficiency (Daemen et al., 1995).

The ability to target drugs and gene therapeutics to nonmacrophage cells within the vasculature has been one of the most sought after goals in clinical therapeutics. Therefore, attachment of specific (but not macrophage-recognizable) ligands onto the surface of macrophage-evading carriers will open the possibility of targeting specific cell types or subsets of cells within the vasculature and even elements of vascular emboli and thrombi. One example is the abnormal lymphocyte differentiation antigens on leukemia cells (Raso et al., 1982), which could serve as target molecules for ligand-directed targeting of macrophage-evading carriers containing anticancer agents. Another important target is vascular endothelial cells. The vascular endothelium is remarkably heterogeneous; endothelial cells from different tissues of the body differ in expression of surface antigens and receptors (Kumar et al., 1987; Belloni and Nicolson, 1988; Rajotte et al., 1998; Thorin and Shreeve, 1998). Endothelial cells also display phenotypic changes in response to various cytokines and growth factors (Pober and Cotran, 1990; Bevilacqua, 1993; Kraling et al., 1996; Murray, 1997). For example, ELAM-1 (E-selectin) is not detectable on the surface of quiescent endothelial cells of normal vessels, but is strongly up-regulated on the luminal surface by pro-inflammatory cytokines such as IL-1 and TNF- (Lin et al., 1992). E-selectin induction on endothelial cells is clearly associated with inflammatory and immune processes (Pober and Cotran, 1990) as well as some noninflammatory angiogenic states (Kraling et al., 1996). E-selectin is known to be involved in the arrest of monocytes, neutrophils, and subsets of T-lymphocytes within the circulation. Therefore, it is not surprising that this molecule has become a potential target for several strategies designed to enhance delivery of therapeutic agents to, or to improve imaging of, particular vascular beds. For example, Wickham et al. (1997) have demonstrated that the enhanced expression of E-selectin may be used as a means to target entry of adenoviral vectors into endothelial cells. A bi-specific antibody, which reacts with both E-selectin and with a "FLAG" epitope genetically incorporated into one of the adenoviral coat proteins, was able to direct binding and entry of the adenovirus into endothelial cells pretreated with the TNF- but not into quiescent nonstimulated cells. Another antibody against E-selectin, 1.2B6, has been used for radionuclide imaging of rheumatoid arthritis in humans (Jamar et al., 1997). In this study, the sensitivity of anti-E-selectin coupled to indium-111 was compared with a nonspecific antibody conjugated with technetium-99m, in terms of their ability to detect synovitis. This study showed that targeting with the E-selectin antibody was more intense and specific than with a nonspecific antibody (Jamar et al., 1997). Furthermore, the anti-E-selectin antibody was able to detect joint abnormalities in areas that were clinically silent.

Biochemical differences between the vasculature of the tumor and the normal tissues are also noteworthy. Indeed, various approaches have been taken to identify differences in luminal protein expression between the endothelial cells of tumor vessels and those of normal tissues. These include two-dimensional gel electrophoresis of cell surface proteins isolated from cultured endothelial cells (P. Hewett and J. C. Murray, manuscript submitted), screening of monoclonal antibodies on tissue sections (Hagemeier et al., 1986; Rettig et al., 1992), in vivo lectin labeling (Belloni and Nicolson, 1988), and the application of bacteriophage display technology in vivo (Rajotte et al., 1998) to study endothelial heterogeneity. These markers, which are usually up-regulated on the surface of tumor-associated endothelial cells, include proteins involved in cell-cell or cell-matrix interactions (e.g., E-selectin, vascular cell adhesion molecule-1, and the _v3 integrin complex) and growth factors receptors (Cheresh, 1987; Matthews et al., 1991; Kim et al., 1993; Brooks et al., 1994; Dumont et al., 1994; Sato et al., 1995; Max et al., 1997; Pasqualini et al., 1997; Charpin et al., 1998; Lauren et al., 1998; Lin et al., 1998; Maurer et al., 1998; Ellerby et al., 1999). The growth factor receptors of potential interest are the vascular endothelial growth factors-2 (KDR-2), which are up-regulated in a wide range of solid tumors (Matthews et al., 1991), and the Tie family of endothelial receptor tyrosine kinases (Tie-1 and Tie-2) (Lauren et al., 1998). Targeted disruption of gene function in murine embryos indicates that Tie-2 plays a pivotal role in both angiogenesis and vascular remodeling/stability, whereas Tie-1 is required for the maintenance of vascular integrity (Dumont et al., 1994; Lin et al., 1998). The angiopoietin growth factors have been identified as the ligands for Tie-2, whereas Tie-1 remains an orphan receptor. Since the development of new blood vessels is essential for tumor growth (Griffioen and Molema, 2000), an attractive antitumor strategy is to exploit these biochemical differences and aim at the tumor vasculature for therapeutic intervention with ligand-bearing long-circulating particles. This strategy might overcome the physiological differences within tumors that have been described as heterogeneous and unpredictable. One must also be cautious with regard to heterogeneity in angiogenic stages in human tumor vasculature. This is because in the animal model, the majority of tumor vasculature is usually in a proangiogenic state, whereas in human tumors, the percentage of proangiogenic vessels is variable and usually low (Griffioen and Molema, 2000). Hence, such antiangiogenic approaches may only affect a minority of the tumor vasculature.

Even if engineered colloidal carriers can be targeted successfully to specific subsets of circulating blood cells or endothelia, subsets may differ significantly in their capacity to internalize particles bound to the cell surface. Therefore, in selected cases, it may be necessary to use particle-immobilized ligands that upon binding and signal transduction initiate local effects (e.g., apoptosis or cytokine production).

	III. Rational Approaches in the Design of Long-Circulating Particles

Top Previous Next References

If monocytes and macrophages in contact with blood are not the desired target, then how can we avoid their phagocytic "barrier"? To date, a classical approach has been to administer large doses of placebo carriers in an effort to impair the phagocytic capacity of macrophages in contact with the blood, thereby allowing subsequently administered particles containing active material to remain in systemic circulation for prolonged periods or to reach designated targets (Abra et al., 1980; Souhami et al., 1981; Moghimi and Davis, 1994). Another strategy is transient apoptotic destruction of liver and spleen macrophages by prior administration of gadolinium chloride (Hardonk et al., 1992) or liposomes with entrapped clodronate (Naito et al., 1996; Schmidt-Weber et al., 1996; van Rooijen et al., 1997). The latter approach has received some attention for the in vivo gene transfer protocols involving adenoviruses (Wolff et al., 1997). These strategies, although successful in experimental models, have little justification in clinical practice as they suppress the essential defense system of the body.

"If one way be better than another, you may be sure it is nature's way" [Aristotle]. It seems appropriate to design a long-circulating carrier based on nature's principles. For example, healthy erythrocytes evade the macrophages of the immune system and fulfill their function of transporting oxygen with a life span of 110 to 120 days. A multitude of physicochemical and physiological factors are believed to control the life span of red blood cells. These include surface characteristics (e.g., surface charge, membrane phospholipid composition, surface antigens) as well as bulk properties (e.g., shape and their extent of deformability) (Weiss and Tavassoli, 1970; Schnitzer et al., 1972; Chen and Weiss, 1973; Moghimi, 1995b; Oldenborg et al., 2000). For instance, red blood cells may avoid macrophage surveillance with the protection of a barrier of oligosaccharide groups. Furthermore, their deformable nature allows them to bypass the human splenic filtration process at the IES in the walls of venous sinuses. In the same context, two decades ago Densen and Mandell (1980) pointed out, "it is naive to assume that microbes passively accept their fate at the hands of phagocytes". In fact, for the successful growth, certain pathogens have deployed a clever array of surface strategies to avoid phagocytosis by macrophages (Ram et al., 1998; Alcami and Koszinowski, 2000). Perhaps we should also learn to translate feasible microbial surface strategies in the engineering of long-circulating or macrophage-evading nanoparticles.

A. The First Few Steps

1. Physicochemical Characteristics of Nanoparticles and Their Effect on Protein Adsorption and Opsonization. It has been repeatedly emphasized that the clearance behavior and tissue distribution of intravenously injected particulate drug carriers are greatly influenced by their size and surface characteristics (Poznansky and Juliano, 1984; Patel, 1992; Moghimi and Davis, 1994). These physicochemical parameters can control the degree of particle self-association (Ahl et al., 1997) in the blood as well as particle opsonization in biological fluids. The size of a particle may change substantially upon introduction into a protein-containing medium (e.g., plasma). Therefore, in the blood, particles and their aggregates should be small enough so that they are not removed from the circulation by simple filtration in the first capillary bed encountered (e.g., rat or mouse lung following tail vein injection). The opsonization process is the adsorption of protein entities capable of interacting with specific plasma membrane receptors on monocytes and various subsets of tissue macrophages, thus promoting particle recognition by these cells (Chonn et al., 1992; Moghimi and Davis, 1994; Gref et al., 1995; Moghimi and Patel, 1998; Moghimi and Hunter, 2000a). Classical examples of opsonic molecules include various subclasses of immunoglobulins, complement proteins like C1q and generated C3 fragments (C3b, iC3b), apolipoproteins, von Willebrand factor, thrombospondin, fibronectin, and mannose-binding protein (Absolom, 1986; Patel, 1992; Serra et al., 1992; Chonn et al., 1995; Moghimi and Patel, 1996; Szebeni, 1998). On exposure to blood, particles of differing surface characteristics, size, and morphology attract different arrays of opsonins as well as other plasma proteins, the content and conformation of which may account for the different pattern in the rate and site of particle clearance from the vasculature (Moghimi and Patel, 1998). Since opsonization plays a major role in particle clearance from the blood, then interindividual variations in blood opsonic activity and concentration must also be considered. Undoubtedly, a clear understanding of such events is the first rational step for the design of colloidal carriers that target not only a relevant macrophage population (see Section III.A.2.) but also for the engineering of long-circulating or macrophage-evading particles. It should also be emphasized that the interaction of particles with blood protein may have effects beyond opsonization. These may include interference with the blood-clotting cascade, a process that may lead to fibrin formation, and anaphylaxis because of complement activation.

2. Macrophage Heterogeneity, Physiological Status, and Species Differences. It is important to realize that not all macrophages and monocytes are identical; considerable heterogeneity with respect to phenotype and physiological properties (e.g., phagocytosis) exist between different types of phagocytes and even among phagocytic cells of the same tissue (Gordon et al., 1992; Naito, 1993; Rutherford et al., 1993). For example, rat Kupffer cells are heterogeneous with regard to phagocytosis and accessory functions such as cytokine production, tumoricidal capability, and antigen presentation to T cells (Sleyster and Knook, 1982; Hoedemakers et al., 1994). Kupffer cells in the periportal area are larger and have a higher lysosomal enzyme activity (on a per cell basis as compared with cells in other regions of the liver), which reflects their higher level of phagocytic and scavenging activities than those of central macrophages. On the other hand, macrophages of the pericentral region are more active in cytokine production and have a higher tumoricidal capability than the larger periportal Kupffer cells (Hoedemakers et al., 1994). With regard to phagocytic receptors, human Kupffer cells and splenic macrophages serve as good examples, Table 1 (Buckley et al., 1987; Tomita et al., 1994). For instance, in human liver the expression of the IgG-coated particle receptor (FcRIIIA) is restricted only to Kupffer cells located in the central area of the liver lobule (Tomita et al., 1994). In murine, macrophages found within the marginal zone, which surrounds the lymphoid regions of the spleen, differ in morphology, endocytosis, and receptor expression from those in the red pulp. For example, unlike other splenic macrophages, the marginal zone macrophages express a specific scavenger receptor (known as MARCO) that is distinct from type 1 and type 2 scavenger receptors (Ito et al., 1999). It is therefore conceivable that the particular populations of macrophages may respond differently to particulate delivery systems and, hence, employ one particular recognition mechanism. Therefore, understanding of macrophage properties, heterogeneity, and recognition mechanisms could provide new insights and opportunities for the design of long-circulating particles as well as carriers that can selectively deliver drugs and therapeutic materials to particular macrophage subpopulations (e.g., for optimizing delivery to recruited monocytes versus local resident macrophages, and selective and apoptotic ablation of macrophages serving as reservoirs of infectious agents, such as HIV and tuberculosis). Here, consideration must also be given toward the physiological status of macrophages of the RES. A responsive phagocyte can become primed and later selectively activated for one or more functions (Adams and Hamilton, 1992). Examples of such functions include enhanced phagocytosis, chemotaxis, antigen processing and presentation, and destruction of tumor cells. The capacity to accomplish a complex function is dependent, in turn, upon acquisition of the requisite capacities such as enhanced mobility of a given population of plasma membrane receptors for a given chemotactic agent or an immobilized opsonin, increase in the number or selective expression of a particular group of scavenging receptors (e.g., the expression of plasma membrane FcRIA which is restricted to human Kupffer cells at the sites of inflammation and is apparently unrelated to the type of liver disease), and the amount of mediators secreted etc. There are strong indications that activated macrophages can recognize engineered long-circulating particles from the blood (Moghimi et al., 1993a; Moghimi and Murray, 1996; Schmidt-Weber et al., 1996; Moghimi and Gray, 1997) (discussed under Section VII.). Hence, such long-circulating particles may act as vehicles for delivery of therapeutic or diagnostic agents to such macrophage populations.

3. Splenic Filtration. The size and the deformability of particles plays a critical role in their clearance by the sinusoidal spleens of humans and rats. Particles must be either small or deformable enough to avoid the splenic filtration process at the IES in the walls of venous sinuses (Chen and Weiss, 1973; Moghimi et al., 1991). The IES in sinusoidal spleens provides resistance to flow through the reticular meshwork. The endothelial cells of the sinus wall have two sets of cytoplasmic filaments: a set of loosely associated tonofilaments and a set of filaments tightly organized into dense bands in the basal cytoplasm containing actin and myosin, which can probably vary the tension in the endothelial cells and, hence, the size of IES (Drenckhahn and Wagner, 1986). However, the slit size rarely exceeds 200 to 500 nm in width, even with an erythrocyte in transit (Chen and Weiss, 1973). Hence, retention of blood cells and blood-borne particles at the IES depends on their bulk properties, such as size, sphericity, and deformability. These cell slits are the sites where erythrocytes containing rigid inclusions (e.g., Heinz bodies, malarial plasmodia) are believed to be "pitted" of their inclusions, which are eventually cleared by the red pulp macrophages (Weiss and Tavassoli, 1970; Schnitzer et al., 1972; Groom, 1987). Idealy, the size of an engineered long-circulatory particle should not exceed 200 nm. If larger, then the particle must be deformable enough to bypass IES filtration. Alternatively, long-circulating rigid particles of greater than 200 nm may act as splenotropic agents (Moghimi et al., 1991).

4. Confinement to Vasculature. Earlier we discussed the possibility of particle extravasation from the blood to selected sites in the body (Section II.D.). However, if we intend to keep long-circulating particles within the vasculature, as in certain cases (e.g., a circulating drug reservoir), then a lower size window in the design of long-circulating particles must also be considered. There is a clear-cut relationship between particle size (e.g., as in the case of liposomes) and the extent to which they reach the hepatic parenchyma. For instance, it has been demonstrated that the smaller the liposomes (usually those of below 100 nm in diameter), the larger the contribution of hepatocytes in total hepatic uptake (Roerdink et al., 1984). This is probably a reflection of the size of the fenestrations in the hepatic sinusoidal endothelium, which are ~100 to 150 nm in diameter (Braet et al., 1995). Even highly deformable vesicles of up to 400 nm in diameter can also reach hepatocytes after intravenous injection by a process of "extrusion" through endothelial fenestrations (Romero et al., 1999). Therefore, as a rough approximation, the size of long-circulating particles, providing that they are rigid structures, should be in the range of 120 to 200 nm in diameter to substantially avoid particle trapping in space of Disse and hepatic parenchyma.

	IV. Translation of Microbial and Related Mammalian Technologies to Nanoparticle Engineering

Top Previous Next References

Considerable information has become available from pathogenic microorganisms that employ various surface strategies to avoid recognition by macrophages. For example, mucoid strains of Pseudomonas aeruginosa from patients with cystic fibrosis display a polyuronic acid polysaccharide that interferes with phagocytosis by virtue of its hydrophilicity (Cabral et al., 1987). The external envelope glycoproteins of HIV type-1 and its simian counterpart are heavily glycosylated. Recent studies have demonstrated that these carbohydrates form a barrier to help shield the virus from immune recognition and limit effective antibody responses to the virus (Reitter et al., 1998). A major constituent of various microbial envelope glycoproteins is sialic acid (Ram et al., 1998). Such sialyated proteins favourably bind factor H (Meri and Pangburn, 1994). The binding of factor H contributes to pathogenicity by inhibiting complement-mediated destruction; factor H acts as a cofactor for factor I-mediated cleavage of C3b and also inactivates alternative pathway convertase by dissociating Bb from the C3b, Bb complex (Fearon and Austen, 1977; Sim et al., 1981). Examples of factor H binding proteins include the streptococcal M6 protein (Fischetti et al., 1995), the Yad A protein of Yersinia enterocolitica (China et al., 1993), and also the envelope glycoproteins 120 and 41 of HIV-1 (Stoiber et al., 1996; Reitter et al., 1998). Microorganisms such as certain species of Staphylococcus aureus fix complement in a cryptic subcapsular location inaccessible for ligation by complement receptors. Eukaryotic infectious agents, such as Candida albicans and related species, subvert C3-mediated opsonization processes by expressing a surface protein that is antigenically, structurally, and functionally related to the mammalian integrin CR3 and p150,95 (Heidenreich and Dierich, 1985). Several viruses express complement regulatory proteins to reduce complement attack (Alcami and Koszinowski, 2000).

Can coating of synthetic nanoparticles with a hydrophilic microbial polysaccharide, factor H, or a factor H ligand fend off the body's defense system? Can such constructs release their entrapped materials in a controlled manner within the systemic circulation or at the desired pathological sites? And, finally, what are the toxicological variables inherent with these types of systems? One of the earliest attempts undertaken was the coating of liposomes with linear dextrans, which are polymers of -D-glucose units (Pain et al., 1984). Linear dextrans are produced by the fermentation of sucrose with a strain of the bacterium Leuconostoc mesenteroides followed by hydrolysis and fractionation to give dextrans with different average molecular weights. Linear dextrans have frequently been used as plasma expanders in medicine; they remain in the systemic circulation for extended periods of time which are proportional to their molecular weights. The circulatory persistence of drugs (e.g., doxorubicin) and proteins such as carboxypeptidase G, superoxide dismutase, and arginase have all been increased following conjugation to dextran (Melton et al., 1987a,b; Wileman, 1991). Liposomes coated with dextran (molecular weight of 70,000) were also more efficient in retaining the entrapped radioactive markers within the circulation than uncoated liposomes (Pain et al., 1984). The rate of clearance of dextran-coated liposomes was dependent on the density of dextran molecules on the liposome surface. The steric brushes of the dextran macromolecules are believed to reduce protein adsorption, resulting in enhanced stability of liposomes in the blood (Pain et al., 1984). Perhaps, one of the most successful applications of dextran in nanoparticulate engineering is the development of ultrasmall superparamagnetic iron oxide particles, which are used in magnetic resonance imaging (Bengele et al., 1994; Guimaraes et al., 1994; Wiessleder et al., 1994; Moghimi and Bonnemain, 1999). These particles consist of hexagonally shaped iron cores of 4 to 5 nm in diameter surrounded by 20 to 30 hydrated "brush-like" structures composed of dextran molecules. By laser light scattering, such crystals exhibit a unimodal hydrodynamic radius of 20 to 25 nm. Because of their physical properties (small size and hydrophilic nature), iron oxide particles are poorly recognized by Kuppfer cells and splenic macrophages and, concomitantly, exhibit prolonged circulation in the blood with a half-life of 3 to 4 h (Moghimi and Bonnemain, 1999). The eventual macrophage recognition of such particles may be attributed to gradual opsonization with antidextran antibodies and complement activation. These partricles have found clinical applications in lymph node imaging (see Section VIII.C.).

Another example is the bacterial exopolysaccharide pullulan (an -D-glucan). Cholesterol-pullulan conjugates with various degrees of substitution as well as monoalkyl(hexadecyl)-pullulan have all been shown to confer some degree of protection to the liposome surface and moderately reduce vesicle uptake by both the liver and spleen macrophages in vivo (Kang et al., 1997). The capsular polysaccharide of N-acetyl neuraminic acid (polysialic acids) from Escherichia coli K1 and K92, Neisseria meningitidis B and C as well as their shorter chain derivatives should also be considered for nanoparticle surface engineering. Indeed, preliminary experiments have confirmed that long-chain polysialic acids exhibit prolonged circulation half-lives (Gregoriadis et al., 1993; Fernandes and Gregoriadis, 1997); due to their high hydrophilicity, little or no binding occurs between factor H and polysialic acids (Pangburn et al., 1991; Meri and Pangburn, 1994). Recently, successful attempts were made to increase the half-life of asparaginase following coupling to such polysialic acids (Fernandes and Gregoriadis, 1997). Theoretically, conjugation of liposomes or nanospheres to polysialic acids could be carried out by a variety of methods, depending on the reactive groups available on the interacting entities. However, caution is required for utilizing polysialic acids for human use. This is because coupling reactions could potentially alter or damage the tertiary structure of the long-chain polysialic acids and, hence, affect the clearance patterns of the engineered colloids. The bacterial polysialic acids C and K92 are immunogenic in humans whereas polysaccharide B and K1 are T-independent antigens and do not induce immunological memory (Moreno et al., 1985). However, it is possible that polysialic acids B and K1 could become T cell dependent antigens and induce memory when grafted to nanocarriers. Another concern is the possible antigenicity of polysialic acids; low levels of antibodies against polysialic acids exist in the circulation (Mandrell and Zollinger, 1982). Macrophages as well as other myeloid cells express several sialic acid binding receptors such as sialoadhesin, CD33, and siglec-5 (Crocker et al., 1991; Munday et al., 1999). The sialic acid-binding proteins are not phagocytic receptors (Munday et al., 1999). However, they could cooperate with phagocytic receptors to increase the efficiency of recognition and uptake, but they function primarily as ligands in cellular recognition events and signaling. Therefore, it will be of interest to determine whether interactions between sialic acid binding proteins and polysialic acids influence host defense functions.

Other successful approaches might be to tag drug carriers with complement regulatory proteins or specific complement inhibitors. Several novel complement inhibitors and chimeric or modified human complement regulatory proteins are currently available for pursuing such studies. Perhaps one of the easiest approaches is to enrich the surface of particulates with factor H ligands. For example, factor H has the ability to bind to polyanions such as heparin, and most sulfated glycosaminoglycans, to include dextran sulfate and chondroitin sulfate A (but not chondroitin sulfate C, keratan sulfate, hyaluronic acid, or polyaspartic acid) (Pangburn et al., 1991; Meri and Pangburn, 1994). It should be emphasized that the interaction between factor H and polyanions is specific and may depend upon the number, orientation, and polymeric arrangement of the anionic groups on nanoparticle surface. Recently, Passirani et al. (1998) covalently attached heparin to the surface of monodispersed poly(methyl methacrylate)-based nanoparticles with a mean diameter of 160 nm. Following intravenous injection into mice, heparin-coated nanoparticles exhibited an initial phase of elimination from the blood with a half-life of 5 h, the remaining heparin nanoparticles circulated for ~48 h. The prolonged circulation time of heparin nanoparticles was suggested to arise from the inhibition of complement activation (Passirani et al., 1998). Although, complement inhibition was only demonstrated in human serum, it is unlikely that the inhibition of complement activation in mice can explain the prolonged circulation time of nanoparticles and their poor hepatic localization. In the event of complement activation, resident murine Kupffer cells lack complement receptors with scavenging functions (Lepay et al., 1985; Lee et al., 1986; Gordon et al., 1992). The prolonged circulation time of nanoparticles in mice is most likely due to their high degree of surface hydrophilicity. It is also likely that the vascular endothelial cells may eventually play a critical role in the clearance of heparin-coated particles because these cells possess heparin receptors, although species variation must be considered (Patton et al., 1995).

Heparin-like polysaccharides can also be obtained from microorganisms. For example, E. coli serotype K5 synthesizes a form of a relatively low molecular mass (~50 kDa) desulfatoheparin (Vann et al., 1981). This molecule has repeating units of 4--D-glucuronosyl-1 and 4--N-acetyl-D-glucosamine, which is similar to N-acetylheparosan, a biosynthetic precursor of heparin. The bacterial product resembles type II glycosaminoglycuronan chains, which are synthesized in the Golgi complex of eukaryotes and then joined to core proteins. Unlike the eukaryotic products, the bacterial polymer lacks L-iduronosyl and sulfate residues. Because of its structural similarity to heparin and the abundance of the negatively charged glucronic acid, these heparin-like molecules may be worthy of particle-surface modification. A related and successful strategy has been the grafting of polyglucuronide to a liposome (100 nm) surface that conferred some degree of vesicle invisibility to macrophages of the RES of both mice and rats (Namba et al., 1992; Oku et al., 1992; Oku and Namba, 1994). For example, at 22 h after intravenous administration into rats, the ratios of the total amount of liposomal associated radioactive marker in the liver and the spleen to that in the blood were 2.4 and 6.3 for liposomes containing 20 mol% palmitylglucuronide and control formulation, respectively.

Sialic acid is an essential component of eukaryotic cell surfaces that plays an important role in preventing destruction of host tissue by constant low-grade activation of the alternative pathway. This is evident from desialyation of erthyrocyte membranes, which results in their conversion from nonactivators to activators of the alternative complement pathway due to reduction in factor H binding (Pangburn and Muller-Eberhard, 1978; Kazatchkine et al., 1979). Surolia and Bachhawat (1977) conducted an elegant approach to rendering liposomes invisible to macrophages by mimicking the mammalian cell surface. Here, the liposome surface was enriched with sialic acid following the incorporation of cell derived glycolipids into the liposomal bilayer. This work initiated a major breakthrough in the liposome field; since then, various modifications of vesicles with gangliosides (Allen and Chonn, 1987; Gabizon and Papahadjopoulos, 1988; Allen et al., 1989; Chonn et al., 1992; Liu et al., 1995), ganglioside derivatives (Park and Huang, 1993; Yamauchi et al., 1993, 1995) and glycophorin (Yamauchi et al., 1993) have been made and investigated. Indeed, sialic acid containing gangliosides (e.g., ganglioside GM1), Fig. 1, when incorporated into liposomes, inhibit the alternative complement pathway as it promotes the binding of factor H to liposome-bound C3b (Michalek et al., 1988). In mice, DSPC or DPPC vesicles of 70 to 200 nm in size, containing high concentrations of cholesterol (>30 mol%) and GM1 at 5 to 7 mol%, behave as long-circulatory (Allen and Chonn, 1987; Gabizon and Papahadjopoulos, 1988; Allen, 1994a). Interestingly, in rats, GM1-containing vesicles are cleared rapidly from the circulation by hepatic Kupffer cells (Liu et al., 1995). The presence of naturally occurring anti-GM1 IgM in rat (but not in mouse) blood, through the activation of the classical pathway of the complement system, is thought to be responsible for the rapid detection of GM1-incorporated vesicles by Kupffer cells (Wassef et al., 1991; Alving and Wassef, 1992; Liu et al., 1995). What mechanisms are therefore responsible for prolonged circulation time of GM1 liposomes in mice? GM1 at 5 to 7 mol% is miscible with both DPPC and DSPC bilayers and forms the most stable vesicles in the presence of cholesterol (Bedu-Addo and Huang, 1996). The stability of the bilayer reduces vesicle susceptibility to lysis by mouse plasma components (e.g., lipoproteins) and to perturbation or penetration by plasma or cell surface proteins. Furthermore, due to the inhibition of complement activation, the assembly of the membrane attack complex at the vesicle bilayer is prevented. Gabizon and Papahadjopoulos (1988) postulated that the negative charge in GM1 is "shielded" by a bulky, neutral, hydrophilic sugar moiety that contributes to macrophage avoidance by decreasing or preventing protein adsorption or opsonization processes. This hypothesis is further supported by the fact that liposomal incorporation of other negatively charged glycolipids, where the negative charge is shielded by neutral sugars (e.g., phosphatidylinositol and sulfatides), also prolongs the blood retention time of the vesicle. Replacing phosphatidylinositol with phosphatidylinositol phosphate, in which an "exposed" charged phosphate is added to the inositol molecule, or replacing GM1 with G_T1, in which two silaic acid molecules with exposed carboxyl groups are present, enhances the hepatic clearance of liposomes (Gabizon and Papahadjopoulos, 1988, 1992). In spite of these experiments, the validity of this hypothesis is still questionable. Several studies have demonstrated that negatively charged glycolipids or phospholipids with exposed and unshielded carboxylic groups such as ganglioside GM3 (Fig. 1) (Yamauchi et al., 1993), [2-(2-palmitoylamido-1-ethyl)-5-acetoamide-3,5-dideoxy-D-glycero--D-galacto-2- nonulipyranoside]onate (a synthetic derivative of sialic acid) (Yamauchi et al., 1995), N-glutaryldioleoylphosphatidylethanolamine, and N-adipyl dioleoylphosphatidyl-ethanolamine (Park et al., 1992)] show a considerable ability to prolong the circulation time of sub 200 nm liposomes in mice. Furthermore, derivatives of GM1 have also been synthesized such that their carboxyl group is either methylated or reduced (Park and Huang, 1993). These derivatives showed considerable ability in preventing hepatic sequestration of liposomes in mice. Reductive amination of a GM1 derivative containing a C7 analog of sialic acid in the presence of -alanine adds to GM1 oligosaccharide an additional carboxyl group, which is not shielded by the neutral sugar residues, and yet the -alaninyl GM1 had an activity very similar to native GM1 (Park and Huang, 1993). In contrast to GM1, ganglioside GM3 containing liposomes are long-circulatory in rats (Yamauchi et al., 1993). Therefore, it appears that particular stereo organization of the ganglioside sugar residues can control the extent of opsonization and the macrophage recognition of glycolipid-incorporated liposomes.

View larger version (17K): [in this window] [in a new window]   Fig. 1.   Structure of gangliosides GM1 and GM3.

Studies by Moghimi and Patel (1989a,b, 1998) demonstrated that rat serum displays a dual role in Kupffer cell recognition of liposomes. In addition to stimulating liposome uptake by Kupffer cells via a noncomplement-mediated process, serum was shown to contain at least two heat-stable proteins that suppressed liposome recognition by Kupffer cells. Subsequently, it was suggested that a balance between the opsonic molecule and these suppressive proteins (dysopsonins) could regulate the quantity and the rate of clearance of liposomes from the blood by the hepatic macrophages (Moghimi and Patel, 1998). Dysopsonins could modulate the rate of liposome uptake by reducing the amount of liposome-bound opsonin and, hence, protect the phagocytic cells from being destroyed by excessive binding and ingestion of liposomes, particularly those vesicles which are more resistant toward lysosomal esterases. On the basis of this hypothesis, Park and Huang (1993) suggested that the prolonged circulation time of GM1-incorporated vesicles in mice may also be due to the binding of a putative dysopsonin onto the liposome surface leading to a reduced level of both opsonization and/or inactivation of bound opsonins. The dysopsonin hypothesis therefore suggests another possibility for enhancing the circulation time of liposomes (and perhaps other particulates). If dysopsonins play a significant role in vivo, then vesicles enriched with these proteins should be expected to display long circulation times. This approach warrants the identification of these blood dysopsonic molecules and their mode of action with regard to different types of colloidal carriers.

Attempts have also been made to introduce sialic acid groups onto the surface of polymeric nanospheres. One example was the noncovalent adsorption of glycoproteins rich in sialic acid such as orsomucoid onto the surface of poly(isobutylcyanoacrylate) nanoparticles (Olivier et al., 1996). This attempt was disappointing, since the half-life of these particulates was similar to those of uncoated systems. This was presumably due to insufficient density or altered conformation of the clustered glycans on the nanoparticle surface. Alternatively this could have been due to the desorption and exchange of sialic acid rich glycoproteins as soon as the particles were introduced into the systemic circulation.

Another potentially interesting mammalian system for disguising foreign particles as "self" is the integrin-associated protein CD47. This molecule has been suggested to function as a marker of self on murine red blood cells and lymphocytes since in the absence of CD47 the cells are cleared from the bloodstream by the splenic red pulp macrophages (Oldenborg et al., 2000). This marker protects the cells against elimination by binding to the SIRP. CD47 analogues are also encoded by smallpox and vaccina viruses (Lindberg et al., 1994). It seems that these pathogens are taking advantage of SIRP signaling to disable normal defenses. Therefore, CD47-SIRP may represent a viable approach for the design of macrophage-evading colloidal carriers.

	V. Synthetic Polymers in Colloid Engineering

Top Previous Next References

From the preceding discussion it is now apparent that a relatively successful approach for prolonging the circulation times of colloidal particles in the blood is to create a steric surface barrier of sufficient density. Because of the possible immunological consequences associated with some bacterial polysaccharides and the high cost of recombinant complement regulators, tremendous efforts have been directed to design synthetic polymers that can fulfill these criteria. The design of long-circulating particles can be traced back to the late sixties. It has long been known that stabilization of emulsion systems may be achieved at the interface by the addition of an emulsifying agent or a surfactant. In a pioneering experiment Geyer (1967) demonstrated that intravenously injected lipid emulsions prepared with high molecular weight members of POE/POP copolymer nonionic surfactants (poloxamers and poloxamines) as emulsifiers remained in the blood for relatively long periods. This behavior was initially thought to arise from the interference of surfactants with the lipoprotein-lipase activity (Hart and Payne, 1971). Later, it was suggested that high molecular weight POP/POE surfactants in some way prevent lipid particles from sticking to the blood vessel endothelium as well as inhibiting recognition by macrophages (Jeppsson and Rossner, 1975).

Poloxamers consist of a central POP block that is flanked on both sides by two hydrophilic chains of POE (Fig. 2). A slightly different structure is exhibited by the poloxamines that are tetrafunctional block copolymers with four POE/POP blocks joined together by a central ethylene diamine bridge (Fig. 2). Numerous investigators have now demonstrated that such copolymers adsorb onto the surface of oil-in-water emulsions or any hydrophobic nanoparticulate systems [e.g., polystyrene, gold, PLGA, poly(phosphazene), poly(methyl methacrylate) and poly(butyl 2-cyanoacrylate)] via their hydrophobic POP center-block (Troster et al., 1990; Moghimi et al., 1993c; Stolnik et al., 1994; Storm et al., 1995; Vandorpe et al., 1997; Monfardini and Veronese, 1998). This mode of adsorption leaves the hydrophilic POE side-arms in a mobile state as they extend outward from the particle surface and provide stability to the particle suspension by a repulsion effect through a steric mechanism of stabilization involving both enthalpic and entropic contributions (Moghimi et al., 1993c). The strength of polymer adsorption and the resultant polymer conformation is dependent on the proportion and the size of both POP and POE segments as well as the physicochemical properties of the nanoparticle surface (Moghimi and Hunter, 2000b). With the help of field-flow fractionation, electron spin resonance and conventional labeling techniques, detailed analytical characterization of the adsorption complexes formed between poloxamers and polystyrene beads of different sizes have been accomplished. These studies have also pointed toward the importance of nanoparticle surface curvature on polymer chain mobility and conformation (Li et al., 1994). For a given triblock polymer, it was found that both surface concentrations and adlayer thicknesses are strongly related to the particle size, such that smaller particles (sizes below 100 nm) take up fewer polymer molecules per unit area than the larger ones. The reduced crowding around each POE chain results in thinner adlayers and higher chain mobilities. Therefore, the surface density decreases with decreasing particle size. For a particle of a given size, it is the size of the surfactant's hydrophobic center block (POP), rather than its flanking tails, that determines the surface concentration or density. Thus, triblocks of similar POP size showed comparable surface concentration, while the longer POE chains were associated with thicker adlayers as well as greater chain dynamics (Li et al., 1994). Therefore, by keeping the particle size constant, one can gain insight into the effect of POE chain lengths on plasma protein adsorption and phagocytosis. Indeed, among the various copolymer members, poloxamine-908, poloxamine-1508, poloxamer-238, and poloxamer-407 (Fig. 2) have proved to be among the most effective copolymer nonionic surfactants for prolonging the circulation time of hydrophobic nanoparticles of 15 to 150 nm in mice and rats. For example, reported half-lives of poloxamine-908-coated nanospheres in mice and rats vary from a few hours to 1 to 2 days, depending on both the particle size and its initial surface hydrophobicity (Storm et al., 1995; Monfardini and Veronese, 1998; Moghimi and Hunter, 2000b).

View larger version (19K): [in this window] [in a new window]   Fig. 2.   The structure of selected poloxamines (a) and poloxamers (b).

It has also been shown that surface modifications with poloxamers and poloxamines before intravenous injection is not really necessary for making nanoparticles long-circulatory. Intravenously injected uncoated 60 nm polystyrene nanoparticles (which are susceptible to phagocytosis by Kupffer cells) were converted to long-circulating entities in those rats that received a bolus intravenous dose of either poloxamer-407 or poloxamine-908, 1 to 3 h earlier (Moghimi, 1997, 1999). It can be argued that the altered biodistribution profile of nanoparticles is the result of cell-surface modification by the administered copolymers. For instance, block copolymers could adhere to cell membrane hydrophobic domains via their hydrophobic center block or act as an effective membrane-spanning entity (Watrous-Peltier et al., 1992). The extracellular steric constraints resulting from hydrophilic POE tails of copolymers will then prevent the interaction between an approaching particle and the cell. Interestingly, this was not the primary mechanism. Instead, nanoparticles were shown to acquire a coating of copolymer and/or copolymer-protein complexes in the blood (Moghimi, 1997); this event explains their phagocytic resistance.

Following intravenous injection to mice and rats, poloxamer- or poloxamine-coated sub-200 nm nanoparticles of albumin (Lin et al., 1994), poly(phosphazene) (Vandorpe et al., 1997), and PLGA (Stolnik et al., 1994) as well as liposomes (Woodle et al., 1992; Kostarelos et al., 1999) do not exhibit prolonged circulation times (reported half-lives are usually less than 2-3 h). Therefore, alternative coating of grafting materials have been tailored. The majority of these materials are based on PEG and its derivatives (Dunn et al., 1994; Lin et al., 1994, 1997, 1999; Gref et al., 1994; Stolnik et al., 1994; Vandorpe et al., 1996, 1997; Lacasse et al., 1998; Peracchia et al., 1997a,c, 1998; Monfardini and Veronese 1998; De Jaeghere et al., 2000). PEG is a linear polyether diol that exhibits a low degree of immunogenicity and antigenicity (Abuchowski et al., 1977). The polymer bac