I. Introduction

Top Previous Next References

The rapid development in current pharmaceutical drug discovery has resulted in the emergence of increasing numbers of novel therapeutic drugs for the treatment of a variety of diseases. However, at present the main problem associated with systemic drug administration is likely to include even biodistribution of pharmaceuticals throughout the body, the lack of drug-specific affinity toward a pathological site, nonspecific toxicity, and other side effects resulting from high doses. An attractive strategy to enhance the therapeutic index of drugs is to specifically deliver these agents to the defined target cells thus keeping them away from healthy cells, which are sensitive to the toxic effects of the drugs. Polymer- and liposome-based delivery systems show potentials as specific and target-oriented delivery systems (Langer, 1998; Maruyama et al., 1999; Vyas and Sihorkar, 2000; Vyas et al., 2001). Naturally existing proteins (such as transferrin) have also received major attention in the area of drug targeting since these proteins are biodegradable, nontoxic, and nonimmunogenic. Moreover, they can achieve site-specific targeting due to the high amounts of their receptors present on the cell surface. The efficient cellular uptake of transferrin (Tf¹) pathway has shown potential in the delivery of anticancer drugs, proteins, and therapeutic genes into primarily proliferating malignant cells that overexpress transferrin receptors (TfRs) (Kratz and Beyer, 1998b; Singh, 1999; Vyas and Sihorkar, 2000; Kircheis et al., 2002).

In this review, we summarize the biochemistry and molecular biology of Tf and the TfR, including the structure, function, and regulation of TfR expression as well as the latest progress in understanding the mechanism of TfR-mediated iron uptake by cells and transport across the blood-brain barrier. In particular, we address the role of Tf and TfR in the targeted delivery of therapeutic drugs, including small molecule drugs, therapeutic peptides, proteins, and even genes, into the malignant tissues or cells. The role of TfR in brain drug targeting and delivery is also included.

	II. Transferrins

Top Previous Next References

During the past decades, intensive studies have been made toward understanding these unique iron-binding proteins. Several reviews have given a broad range of coverage of Tf and TfR functional properties, structures, metal binding properties, and their potential in biomedical processes (Baker, 1994; Morgan, 1996; Richardson and Ponka, 1997; Aisen, 1998; Sun et al., 1999; Andrews, 2000; Lieu et al., 2001).

The transferrins comprise a family of large (molecular mass ca. 80 kDa) nonheme iron-binding glycoproteins, believed to originate with the evolutionary emergence of vertebrates or prevertebrates. Three major types of transferrins have been characterized. Serum Tf occurs in blood and other mammalian fluids including bile, amniotic fluid, cerebrospinal fluid, lymph, colostrom, and milk. Ovotransferrin (oTf) is found in avian and reptilian oviduct secretions and in avian egg white (Jeltsch and Chambon, 1982; Williams et al., 1982), and lactoferrin (Lf) is found in milk, tear, saliva, and other secretions (Baggiolini et al., 1970; Metzboutigue et al., 1984). Transferrin is mainly synthesized by hepatocytes, with a concentration of 2.5 mg/ml and 30% occupied with iron in blood plasma (Leibman and Aisen, 1979). Recently, a new member of the transferrin family, melanotransferrin (also called p97), has been identified as an integral membrane protein in human malignant melanoma cells and in some fetal tissues (Brown et al., 1982; Rose et al., 1986). It can also be expressed by a wide range of cultured cell types, including liver and intestinal cells (Qian and Wang, 1998).

The principal biological function of transferrins, except melanotransferrin, is thought to be related to iron binding properties. Serum Tf has the role of carrying iron from the sites of intake into the systemic circulation to the cells and tissues (Morgan, 1964; Baker and Morgan, 1969). It is also likely to be involved in the transportation of wide range of other metal ions other than iron, including therapeutic metal ions, radio diagnostic metal ions, and some toxic metal ions (Savigni and Morgan, 1998; Sun et al., 1999). Lactoferrin, on the other hand, is believed to act principally as a bacteriostat, by chelating the iron, which is essential for the growth of microorganism antimicrobial activity. Lactoferrin may also be involved in modulation of the immune and inflammatory responses and act as a growth factor, which is unrelated to iron binding properties (Iyer and Lönnerdal, 1993; Lönnerdal and Iyer, 1995). Ovotransferrin also exhibits antimicrobial activity. The function of melanotransferrin is not yet clear. It seems true that it plays a small role in iron uptake. Instead, it may help in the rapid proliferation of tumor cells and also act as an iron scavenger at the cell surface to prevent lipid peroxidation (Kwok and Richardson, 2002).

The primary structures of over 10 transferrins have been determined and can be found in various protein databases. Transferrins are single-chain glycoproteins containing ca. 700 amino acids with molecular mass ca. 80 kDa. The sequence identity between different species and different members of the family is extremely high, e.g., 78% identity between rabbit and human serum transferrin, 60% between serum transferrin and lacoferrin, and ca. 40% identity between melanotransferrin and other transferrins. The high levels of conservation in their primary structures were also reflected in their three-dimensional structures. Many crystal structures of transferrins (different species and some fragments) are available and have been reviewed previously (Sun et al., 1999). Briefly, the polypeptide chain is folded into two structurally similar but functionally different lobes, referred to as N- and C-lobe, respectively. Two lobes were connected by a short peptide. Each lobe can be further divided into two domains enclosing a deep hydrophilic cleft bearing an iron binding site. At the metal binding site, Fe³⁺ coordinates with distorted octahedral geometry to two oxygens from Tyr, one nitrogen from His, one oxygen from Asp, and two oxygens from a bidentate carbonate (synergistic anion). The ligands are from two domains and two polypeptide strands, which cross over between the two domains at the back of the iron site. This kind of arrangement is crucial to ensure that the domains are able to move apart to form an open conformation, hinged by the backbone strands, which leads to iron release. Bicarbonate is essential for strong binding of iron to the specific site of Tf and may also have a role in iron release. In addition to iron, many metal ions other than iron have been found to bind to the specific iron sites (Sun et al., 1999; Zhong et al., 2002), thus Tf has been implicated in the transportation of other metal ions.

Another distinctive feature of Tf is that it undergoes conformational changes during Fe³⁺ uptake and release that are thought to be crucial for the selective recognition by the receptor of the transporter protein. The mechanism for opening and closing the lobes has been studied intensively but still remains elusive. It has been postulated that the dilysine (Lys209-Lys301 for ovotransferrin) pair may serve a special function in the release of iron. The charge repulsion resulting from the protonation of the dilysine, located in opposite domains, at lower pH may be the trigger to open the cleft and facilitate iron release (Dewan et al., 1993). In addition, this dilysine pair may also serve as an anion binding site for iron release (Baker, 1994; He et al., 1999a). Mutation of either or both lysines to glutamate or glutamine would abolish this trigger and result in an extremely slow release of iron compared with that from the intact N-lobe Tf (He et al., 1999a). For human serum Tf N-lobe, the presence of a trigger mechanism for the domain closure has been claimed on the basis of the absence of full closure in the Fe³⁺-loaded Asp63Ser mutant, as analyzed by X-ray solution scattering (Grossmann et al., 1993a). However, the X-ray crystal structure of Asp60Ser lactoferrin showed a greater domain closure than the parent half-molecule (N-lobe) lactoferrin, which has led to the proposal of an equilibrium between the open and closed forms in solution with a low energy barrier (Faber et al., 1996). Recently, a different trigger mechanism for domain closure was proposed, such that the small triggered motion in Tyr92, an Fe³⁺-coordinating ligand, would induce the extensive rearrangements in the hydrogen bonding networks in the -strand where Tyr92 is located. These networks would work as a driving force for the domain closure (Mizutni et al., 2001).

	III. The Transferrin Receptors

Top Previous Next References

The primary structure of the human transferrin receptor 1 (TfR1) has been deduced from the nucleotide sequence of its cDNA (McClelland et al., 1984; Schneider et al., 1984). It is a transmembrane homodimer that consists of two identical monomers with a molecular mass of approximately 90 kDa; each monomer is joined by two disulfide bonds at Cys89 and Cys98 (Jing and Trowbridge, 1987). It has a short, NH₂-terminal cytoplasmic region (residues 1 to 67), a single transmembrane pass (residues 68 to 88), and a large extracellular portion (ectodomain, residues 89 to 760), which is soluble and bears a trypsin-sensitive site and contains a binding site for transferrin. TfR1 is synthesized in the endoplasmic reticulum and is post-translationally modified with both phosphate and fatty acyl groups (Omary and Trowbridge, 1981; Schneider et al., 1982). The extracellular domain contains three N-linked glycosylation sites at Asn251, Asn317, and Asn727 and one O-linked glycosylation site at Thr104 (Omary and Trowbridge, 1981). These sites are thought to be crucial for the function of TfR1. Mutations at the N-linked glycosylation sites impair transferrin binding activity. Similarly, elimination of the O-linked glycosylation at Thr104 enhances the cleavage of TfR1 and promotes the release of its ectodomain (Williams and Enns, 1991; Rutledge and Enns, 1996).

Crystallographic studies of the ectodomain of human TfR1 (residues 122 to 760) revealed that homodimer of TfR1 is organized as a butterfly-like shape (Fig. 1). Each TfR1 monomer consists of three distinct globular domains (Lawrence et al., 1999). Three domains, identified as the protease-like, apical, and helical domains, form a lateral cleft, which is likely to be in contact with the docked transferrin molecules (Lawrence et al., 1999). The ectodomain of TfR1 is separated from the membrane by a stalk, which probably includes residues involved in disulfide bond formation and the O-linked glycosylation (Fuchs et al., 1998). The amino acid sequence of the globular ectodomain of TfR1 is 28% identical to that of membrane glutamate carboxypeptidase II, which hydrolyzes the most prevalent mammalian neuropeptide, N-acetyl--L-aspartyl-L-glutamate (Lawrence et al., 1999). Therefore, it has been suggested that TfR1 is evolved from a peptidase related to membrane glutamate carboxypeptidase II (Bzdega et al., 1997) although it lacks peptidase activity.

View larger version (43K): [in this window] [in a new window]   Fig. 1.   X-ray crystal structure of the ectodomain of the transferrin receptor 1. A ribbon diagram of the dimeric transferrin receptor 1 is organized as a butterfly-like shape. The transferrin receptor 1 monomer contains three distinct domains. The protease-like, apical and helical domains in one monomer are shown in red, green, and yellow, respectively, and the other monomer is in blue. The stalk region is shown in gray and connected to the putative membrane-spanning helices (Lawrence et al., 1999).

A homodimer of TfR1 can bind up to two molecules of Tf. The binding affinity of Tf for various receptors is very high, from 10⁵ to 10¹⁰ M¹ (Sun et al., 1999). Small differences between transferrin receptors can have large effects on the binding affinity for Tf from different mammals and different cells. However, the diferric protein always has a higher affinity for the TfR1 than its mono-ferric and apo-forms. It is not fully understood how and where TfR binds Tf. The primary receptor recognition site of human transferrin is thought to be mainly on the C-lobe of Tf (Zak et al., 1994). However, recent studies show that both C- and N-lobe of human serum Tf are necessary for receptor recognition (Mason et al., 1997; Zak et al., 2002). Other studies using a human/chicken chimeric TfR1 suggest that Tf binds to a region corresponding to the helical domain (Buchegger et al., 1996). Site-directed mutagenesis has shown that TfR1 residues 646-648, which are present in helix 3 of the helical domain, are critical for Tf binding (Dubljevic et al., 1999). However, docking of Tf molecules to the crystal structure of TfR1 shows that the most contact between Tf and TfR1 involves the apical domain of TfR1 and the N₁ and C₁ domains of Tf, although certain parts of the protease-like and helical domains may also participate in binding of Tf (Lawrence et al., 1999). Therefore, further studies are warranted to understand how TfR1 binds Tf.

The TfR1 appears to be expressed in all nucleated cells in the body, such as red blood cells, erythroid cells, hepatocytes, intestinal cells, monocytes (macrophages), brain, the blood-brain barrier (also blood-testis and blood-placenta barriers), and also in some insects and certain bacteria (Levay and Viljoen, 1995; Lönnerdal and Iyer, 1995; Schryvers et al., 1998; Qian et al., 1998, 1999a, 2000; Moos and Morgan, 2000), but differs in levels of expression (Davies et al., 1981; Enns et al., 1982). It is expressed on rapidly dividing cells, with 10,000 to 100,000 molecules per cell commonly found on tumor cells or cell lines in culture (Inoue et al., 1993). In contrast, in nonproliferating cells, expression of TfR1 is low or frequently undetectable.

Expression of TfR1 in nonerythroid cells is regulated at the post-transcriptional level by interactions of iron-regulatory proteins (IRPs) and iron-responsive elements (IREs) in the 3'-untranslated region of TfR1 mRNA. The 3'-untranslated region of receptor mRNA contains a series of five hairpin stem-loop structures required for iron-dependent regulation (Casey et al., 1988). The stem-loop structures called IREs are recognized by trans-acting proteins, known as IRPs (Leibold and Munro, 1988), which control the rate of mRNA translation or stability (Rao et al., 1986; Müllner and Kühn, 1988). Two closely related IRPs (IRP1 and IRP2) have been identified to date (Leibold and Munro, 1988; Henderson et al., 1993). Both display IRE binding properties under conditions of iron deprivation. IRP1 shares 30% sequence identity with m-aconitase, a [4Fe-4S]-containing enzyme that catalyzes the isomerization of citrate to isocitrate. IRP1 has been regarded as a bifunctional "sensor" of iron, switching between RNA binding and enzymatic activities as aconitase depending on cellular iron status (Constable et al., 1992; Haile et al., 1992; Basilion et al., 1994). Under conditions of iron deficiency, IRP1 binds to the IREs in the 5'-untranslated region of ferritin mRNA (Hentze et al., 1987; Bhasker et al., 1993). This binding sterically prevents the recruitment of 43 S translation preinitiation complex and inhibits translation of ferritin. On the other hand, binding of IRP1 to IREs in the 3'-untranslated region of TfR1 mRNA increases mRNA stability and synthesis of TfR1 (Müllner and Kühn, 1988; Koeller et al., 1989; Müllner et al., 1989). In contrast, under high concentration of intracellular iron, IRP1 is enzymatically active and lacks RNA binding activity that leads to the degradation of TfR mRNA, whereas ferritin mRNA is translated efficiently (Koeller et al., 1989; Müllner et al., 1989).

IRP2 shares about 62% identity with IRP1 and, despite sequence similarities, it does not form a [4Fe-4S] cluster and consequently lacks aconitase activity (Guo et al., 1994). IRP2 binds specifically to all known mRNA IREs with an affinity equally as high as that of IRP1; however, their binding specificities to distinct sequences of iron-responsive elements differ significantly (Henderson et al., 1993; Guo et al., 1994). IRP2 differs from IRP1 in the mechanism by which iron levels are sensed (Guo et al., 1995; Iwai et al., 1995). IRP2 undergoes ubiquitination and proteasomal degradation in iron-replete cells, which is mediated by an iron-dependent oxidation mechanism requiring a unique 73-amino acid domain containing three cysteine residues (Iwai et al., 1995, 1998). IRP2 is induced following iron starvation through renewed synthesis of stable IRP2 protein and its inactivation by iron reflects degradation of IRP2 by a translation-dependent mechanism (Guo et al., 1994; Henderson and Kühn, 1995). It is likely that IRP1 and IRP2 perform distinct functions, probably by acting on different target genes. However, IRP1-deficient mice show no abnormalities in iron metabolism (Rouault and Klausner, 1997). Therefore, IRP2 might compensate for the function of IRP1.

The signals other than iron levels, such as nitric oxide and oxidative stress, can also regulate IRPs and modulate cellular iron metabolism (Drapier et al., 1993). It has been recently reported that the increased IRP activity induced by nitric oxide is one of the causes for the exercise-induced low iron status and high TfR expression (Qian et al., 1999b, 2001; Xiao and Qian, 2000; Qian, 2002). Several review papers have given detailed information (Hentze and Kühn, 1996; Aisen et al., 1999; Lieu et al., 2001). Briefly, nitric oxide and H₂O₂ produced from oxidative stress activate IRP1 by a cycloheximide-insensitive post-translational mechanism (Pantopoulos and Hentze, 1995), whereas IRP2 activation by nitric oxide requires de novo protein synthesis (Pantopoulos et al., 1996). Nitric oxide regulates the binding of IRP1 by disassembling the iron-sulfur cluster or by acting as a cytoplasmic iron chelator, which results in a loss of aconitase activity (Drapier et al., 1993; Gardner et al., 1995, Ho et al., 2001; Bouton et al., 2002). The activation of IRP1 by H₂O₂ is dependent on extracellular signaling events (Pantopoulos et al., 1997; Pantopoulos and Hentze, 1998), suggesting that cluster disassembly alone cannot fully account for H₂O₂-induced IRP1 activation and that signaling pathways are involved.

The TfR binds two proteins critical for iron metabolism: Tf and HFE, the protein mutated in hereditary hemochromatosis (Pietrangelo, 2002; Waheed et al., 2002). Both the primary and crystal structure of HFE showed homology to class I major histocompatibility complex protein (Feder et al., 1996; Lebrón et al., 1998), that is composed of a heavy chain associated with ₂-microglobulin (₂M). The mechanism by which HFE regulates iron uptake into the body is unknown. However, HFE was found to coprecipitate with TfR1 in tissue culture cells (Feder et al., 1998). The HFE·TfR1 complex can also be identified in human tissues such as the placenta and intestine (Parkkila et al., 1997; Waheed et al., 1999; Trinder et al., 2002). Upon forming an association complex in the endoplasmic reticulum, HFE cotrafficks with TfR through the Golgi reticulum network to reach the cell surface (Gross et al., 1998; Roy et al., 1999; Salter-Cid et al., 1999; Ramalingam et al., 2000). The wild-type HFE has been shown to negatively regulate Tf-mediated iron uptake in transfected cells (Parkkila et al., 1997; Gross et al., 1998; Roy et al., 1999; Schwake et al., 2002). This inhibitory effect is however attenuated in the Cys282Tyr mutant HFE (Feder et al., 1998; Okamoto, 2002; Schwake et al., 2002), suggesting that patients with HFE mutations might have pathological iron regulation, possibly via an altered TfR1-dependent iron metabolic pathway.

Similar to Fe-Tf, HFE binds tightly to soluble TfR1 at the cell surface pH 7.4 with K_d ca. 0.6 nM, with little or no binding at the acidic intracellular vesicles pH, suggesting that HFE dissociates from TfR1 in acidified endosomes (Lebrón et al., 1998). Both 2:1 and 2:2 TfR1/HFE stoichiometries have been observed (Lebrón et al., 1998; Bennett et al., 2000; West et al., 2001); however, the stoichiometry of TfR1/HFE on the cell membrane is not known. It has been demonstrated that membrane bound or soluble HFE binding to cell surface TfR1 results in a reduction in the affinity of TfR1 for Fe-Tf (Feder et al., 1998; Gross et al., 1998). This has been attributed to the formation of the ternary complex consisting of one Fe-Tf and one HFE bound to a TfR1 homodimer Tf (Lebrón et al., 1998, 1999). The HFE-TfR1 cocrystal structure (Fig. 2) reveals that HFE binds to the helical domain of TfR1, and a large surface area of interaction exists between HFE and TfR1 (Bennett et al., 2000). Binding of HFE induces conformational changes of TfR1. The backbone structure of the protease and apical domains remains unchanged in contrast to the helical domain in uncomplexed TfR1, which is displaced with respect to the other domains when compared with the structure of complexed TfR1. The movement alters the shape of the cleft between the helical and protease-like domains, and thus possibly alters the ability of Tf to bind to the TfR1 (Bennett et al., 2000). Mutation of five TfR1 residues at the HFE binding sites results in significant reductions in Tf binding affinity, which also support the idea that HFE and Tf compete for overlapping binding sites on TfR1 (West et al., 2001). However, the lower affinity of Tf for TfR1 by HFE is not a satisfactory explanation for the lower iron uptake since the concentration of Fe₂-Tf in serum is very high (ca. 5 µM), and the binding of Fe₂-Tf to its receptor is saturated even in the presence of HFE. A recent study showed that overexpression of HFE without overexpression of ₂M results in a decrease in TfR1-dependent iron uptake in transfected Chinese hamster ovary cell lines, whereas overexpression of both HFE and ₂M results in an increase in TfR1-dependent iron uptake and increased iron levels in the cells (Waheed et al., 2002). There is also a hypothesis that HFE has two mutual activities in cell, i.e., inhibition of uptake or inhibition of release of iron, and the balance between Tf saturation and TfR1 concentrations determined which of these functions predominates (Townsend and Drakesmith, 2002). Functional significance of the association of HFE with the TfR1 and the critical regulatory role of HFE in iron metabolism remain unveiled.

View larger version (33K): [in this window] [in a new window]   Fig. 2.   Crystal structure of the complex of HFE and transferrin receptor 1. A ribbon diagram of the complex reveals that two HFE molecules grasp each side of a transferrin receptor 1 homodimer on the same membrane, giving rise to a 2:2 stoichiometry. Noticeably, the complexation of HFE appears to induce conformational changes in transferrin receptor 1, thus to influence the function of transferrin receptor 1 in terms of transferrin binding and internalization (Bennett et al., 2000).

A new TfR-like family member, transferrin receptor 2 (TfR2), has been cloned and sequenced (Kawabata et al., 1999). The TfR2 gene is located on chromosome 7q22 and gives rise to two transcripts of approximately 2.9 and 2.5 kb in length (Kawabata et al., 1999). Amino acid sequence analysis reveals that, like TfR1, TfR2 is a type II transmembrane glycoprotein that shares a 45% identity and 66% similarity in its extracellular domain with TfR1. The -transcript product is primarily expressed in the livers of humans and mice (Kawabata et al., 1999; Fleming et al., 2000) and the TfR2 -transcript distributed widely but is expressed at low levels (Kawabata et al., 1999). Sequence analysis of TfR2 coding and noncoding region reveals that TfR2 does not possess IRE (Kawabata et al., 1999). Therefore, it is likely that expression of TfR2 is not regulated by IRP-mediated feedback regulatory mechanism in response to cellular iron status. Instead, it may be regulated by a different mechanism, probably related to the cell cycle or cellular proliferation status (Kawabata et al., 2000). TfR2 has a similar function to TfR1 with respect to Tf binding and Tf-mediated iron uptake. Both TfR1 and TfR2 interact with Tf in a pH dependent manner; apo-Tf binds to these receptors only at acidic pH and holo-Tf binds at neutral or higher pH (Kawabata et al., 2000). However, the affinity of TfR2 for iron-loaded Tf is 25-fold lower than that of TfR1 for Tf (West et al., 2000).

The expression pattern for TfR2 is distinct from that for TfR1. It has been shown that during liver development, TfR2 was up-regulated and TfR1 was down-regulated; during erythrocytic differentitation of murine erythroleukemia cells induced by dimethylsulfoxide, expression of TfR1 increased, whereas TfR2 decreased (Kawabata et al., 2001a). Therefore, TfR2 appears to serve unique functions involved in iron metabolism, hepatocyte function, and erythrocytic differentiation. In addition, high levels of TfR2 expression were also found in the erythroid cell lines including erythroid leukemia cell line. Levels of expression of TfR2- mRNA were found to be significantly higher in erythroleukemia marrow samples than in nonmaligant control marrow samples, suggesting that TfR2- may be a useful marker of early erythroid precursor cells (Kawabata et al., 2001b). The clinical significance of TfR2- expression in leukemia cells remains to be determined. Mutations in TfR2 have been identified as the cause of a form of hemochromatosis that is not linked to the mutation of HFE (Camaschella et al., 2000; Roetto et al., 2001), which suggests that TfR2 is associated with iron overload and offers a tool for molecular diagnosis of non-HFE related disorders.

The reason for the existence of two TfRs is still not fully understood. The high level of TfR2 expression in the liver suggests a particular role for this receptor, possibly TfR2 contributes substantially to the liver's ability to capture and store iron. In addition, a possibility has recently been suggested that TfR2 might play an important role in modulation of hepcifin expression, thus involving the regulation of dietary iron absorption (Nicolas et al., 2001; Fleming and Sly, 2001, 2002). Deficiency of TfR2 (Camaschella et al., 2000; Roetto et al., 2001; Girelli et al., 2002) has been demonstrated to cause an hereditary hemochromatosis-like phenopyte, implicating TfR2 as a participant in iron homeostasis. On the other hand, the observation that the TfR1 knockout mutation in the mouse leads to an embryonic lethal phenotype demonstrated that TfR2 couldn't fully compensate for the functions of TfR1 (Levy et al., 1999).

	IV. Transferrin Receptor-Mediated Iron Uptake

Top Previous Next References

Iron is vital for almost all living organisms. However, iron concentrations in body tissues must be tightly regulated because excessive iron leads to tissue damage as a result of the formation of free radicals. Acquisition of iron is a challenge in which many proteins participate to ensure that iron uptake is sufficient and appropriate to the needs of cells, but the total number of proteins involved in mammalian iron metabolism is unknown (Andrews, 1999). Recently, several new genes have been identified, which provide insights into the mechanism of iron absorption by the enterocytes of the duodenal mucosa (Jiang and Qian, 2001; Ke and Qian, 2002). Fe³⁺ in food is reduced to Fe²⁺ by duodenal cytochrome b (McKie et al., 2001) and then absorbed into the enterocyte by an iron transporter protein DMT1 (Fleming et al., 1997; Gunshin et al., 1997). Iron subsequently crosses the enterocyte and is exported from its basolateral surface by another iron transporter, ferroportin 1 (Donovan et al., 2000). In the export process the iron is reoxidized to Fe³⁺ by hepaestin (Vulpe et al., 1999) and bound to serum Tf (Aisen, 1998). In humans, failure to maintain appropriate levels of iron is a feature of iron-deficiency anemia, hereditary hemochromatosis, and even certain neurodegenerative diseases (Smith et al., 1997; Qian et al., 1997a; Qian and Wang, 1998; Andrews, 1999; Qian and Shen, 2001). Abnormally high levels of iron have been found in some regions of the brain in neurodegenerative disorders (Qian and Wang, 1998; Aisen et al., 1999; Qian and Ke, 2001), although it is not clear whether iron accumulation in the brain is an initial event that causes neuronal death or is a consequence of the disease process. It is likely that misregulation of iron metabolism is important in the pathophysiology of certain neurodegenerative diseases (Qian and Wang, 1998; Qian and Shen, 2001; Rouault, 2001).

Cells take up iron by using a variety of mechanisms. In high organisms, one principal pathway of cellular iron acquisition is by the receptor-mediated uptake of transferrin-bound iron, which is one of the best understood processes in cell biology. Several review papers have given a detailed description (Qian and Tang, 1995; Morgan, 1996; Qian et al., 1997b; Aisen, 1998; Andrews, 1999, 2000; Lieu et al., 2001). Figure 3 shows the current model of iron uptake from Tf via TfR1-mediated endocytosis. Briefly, the process is triggered by the binding of Fe₂-Tf to a specific cell-surface TfR1 (Morgan and Laurell, 1963; Morgan and Appleton, 1969). After endocytosis via clathrin-coated pits, which bud from the plasma membrane as membrane bound vesicles or endosomes, the Fe₂-Tf·TfR1 complex is routed into the endosomal compartment. Upon maturation and loss of the clathrin coat, the endosome becomes competent to pump protons in a process energized by ATPase, and the endosomal lumen is rapidly acidified to a pH of about 5.5 (Dautry-Varsat et al., 1983; Klausner et al., 1983; Paterson et al., 1984). At this pH, the binding of iron to Tf is weakened, leading to iron release from the protein. The free Fe³⁺ released to endosomes is reduced to Fe²⁺ on the cis-side of the endosomal membrane probably mediated by oxidoreductase (Núñez et al., 1990). Fe²⁺ is subsequently transported out of the Tf cycle endosome by the divalent metal transporter DMT1, i.e., from the endosomal membrane to the cytosol (Fleming et al., 1998; Tabuchi et al., 2000). Once in the cytosol, iron is utilized as a cofactor for aconitase, the cytochromes, RNA reductase, and heme, or stored as ferritin. After release of iron into the endosome, the resultant apo-Tf·TfR1 complex is then recruited through exocytic vesicles back to the cell surface. At extracellular physiological pH, apo-Tf dissociates from its receptor due to its low affinity at pH 7.4, is released into the circulation, and reutilized (Morgan, 1996, 2001; Qian et al., 1997b). ATP-mediated energy is necessary for sustaining TfR-mediated endocytosis and recycling (Podbilewicz and Mellman, 1990; Schmid and Smythe, 1991).

View larger version (38K): [in this window] [in a new window]   Fig. 3.   The cycle of transferrin and transferrin receptor 1-mediated cellular iron uptake. Differic transferrin (holo-transferrin) binds to transferrin receptor 1 on the cell surface. The resulting complex is internalized and acidified through the action of a proton pump. Iron is subsequently released from transferrin and transported out of endosomes via the divalent metal transporter DMT1. HFE appears to inhibit transferrin receptor 1 endocytosis, probably through binding to transferrin receptor 1. Apo-transferrin and transferrin receptor 1 both returned to the cell surface, where they dissociate at neutral pH, and both participate in another round cycle of iron uptake. Intracellular iron is either incorporated into heme or stored in ferritin.

It has been shown that the number of receptors displayed on the cell surface is proportional to iron uptake and that iron deficiency induces TfR1 gene expression (Aisen, 1998), which implies the significance of TfRs in iron uptake. Furthermore, surface display of TfR1 is affected by its total cellular concentration, as well as its distribution and rate of recycling between the cell surface and cell interior. The efficiency of TfR1 function is also influenced by other proteins, including SFT (stimulator of iron transport) and HFE. SFT stimulates iron uptake by both Tf and non-Tf pathways (Gutierrez et al., 1997), whereas HFE appears to negatively module Tf-dependent iron uptake (Parkkila et al., 1997; Roy et al., 1999).

To date, the mechanisms of iron transport across the blood-brain barrier (BBB) have not been completely clarified. The accumulated evidence suggests that the Tf and TfR pathway may be the major route of iron transport across the luminal membrane of the capillary endothelium (Bradbury, 1997; Moos and Morgan, 1998, 2000; Malecki et al., 1999), and that iron, possibly in the form of ferrous iron, crosses the abluminal membrane and enters into the brain, although the molecular events of this process are unknown (Bradbury, 1997; Moos and Morgan, 1998). The evidence shows that the uptake of Tf-bound iron by TfR-mediated endocytosis from the blood into the cerebral endothelial cells is no different in nature from the uptake into other cell types (Bradbury, 1997). As found in other cells, this process also includes several steps: binding, endocytosis, acidification and dissociation, and translocation of iron across the endosomal membrane. Most of the Tf will then return to the luminal membrane with TfR, whereas the iron then crosses the abluminal membrane by an undetermined mechanism (Bradbury, 1997; Moos and Morgan, 1998). As mentioned before, recent studies have shown that ferroportin 1/hephaestin and/or hephaestin-independent iron export systems might play a key role in ferrous iron transport across basal membrane of enterocytes in the gut (Vulpe et al., 1999; Donovan et al., 2000; Kaplan and Kushner, 2000). This process is very similar to what occurred in the BBB cells (capillary endothelium) (Qian and Shen, 2001). Because the form of iron transport across this membrane might be ferrous iron (Bradbury, 1997; Moos and Morgan, 1998), therefore, a ferroxidase such as hephaestin (or ceruloplasmin) might be necessary for ferrous iron to be oxidized to ferric iron, so that iron, after crossing the basolateral membrane of the BBB cells, could be carried away by Tf (Ke and Qian, 2001). Based on the similarity of the transport form of iron across the basolateral membranes (both are ferrous iron), and the existence of ferroportin 1 and hephaestin in the brain (Jiang et al., 2002), it is possible that ferroportin 1/hephaestin or ferroportin 1/ceruloplasmin system might play a role in iron transport across the BBB cells. Another proposed mechanism involved in ferrous iron transport across abluminal membrane is the role of astrocytes. The astrocytes probably have the ability to take up ferrous iron from endothelial cells through their end feet processes on the capillary endothelia (Malecki et al., 1999; Oshiro et al., 2000). In addition to Tf/TfR pathway, it has been suggested that the lactoferrin receptor/lactoferrin and GPI-anchored p97/secreted p97 pathways might play a role in iron transport across the BBB (Faucheux et al., 1995; Qian and Wang, 1998; Malecki et al., 1999). It is also possible that a small amount of iron might cross the BBB in the form of intact Tf·Fe complex by receptor-mediated transcytosis (Moos and Morgan, 1998) (Fig. 4). After the iron has been transported across the BBB, it is likely to bind quickly to the Tf that is secreted from the oligodendrocytes and choroid plexus epithelial cells (Bradbury, 1997; Moos and Morgan, 1998) or other transporters and then transported to where iron is needed (Qian and Shen, 2001).

View larger version (34K): [in this window] [in a new window]   Fig. 4.   The role of Tf and TfR in iron transport across the BBB. The Tf/TfR pathway may be the major route of iron transport across the luminal membrane of the BBB. The Tf-Fe (transferrin-bound iron) uptake by endothelial cells is no different in nature from the uptake into other cell types. DMT1 might play a role in translocation of iron from endosome to cytosol. Then iron (Fe2+) crosses the abluminal membrane probably via FP1 (ferroportin 1)/HP (hephaestin) and/or HP-independent export systems. LfR/Lf and GPI-P97/S-P97 pathways might also involve iron transport across the BBB. (GPI-P97, GPI-anchored P97; Lf/LfR, lactoferrin/lactoferrin receptor; S-P97, secreted P97; ?, to be confirmed).

	V. Transferrin As a Metallodrug Mediator

Top Previous Next References

The transferrins are primarily iron-binding proteins, but in human serum, Tf is only about 30% saturated with iron, so there is a potential capacity for binding to other metal ions that enter the body. Indeed, over 30 metal ions have been reported to bind to Tf with either carbonate, oxalate, or other carboxylates as synergistic anions, although Fe³⁺ has a higher affinity than any other metal ion for which the binding constant has been determined (Aisen, 1998; Sun et al., 1999). Such binding may play an important role in the transport and delivery of medical diagnostic radioisotopes such as ⁶⁷Ga³⁺ and ¹¹¹In³⁺ (Harris and Pecoraro, 1983; Ward and Taylor, 1988; Harris et al., 1994; Bernstein, 1998) and therapeutic metal ions such as Bi³⁺ (Li et al., 1996a; Sun et al., 2001; Zhang et al., 2001), Ru³⁺ (Kratz et al., 1994; Smith et al., 1996) and Ti⁴⁺ (Sun et al., 1998a; Messori et al., 1999).

Electronic absorption spectroscopy is frequently used to detect metal binding to the specific iron sites of transferrins. Apo-transferrin (apo-Tf) is a colorless protein with an intense ultraviolet absorption near 280 nm with ₂₇₈ 93,000 M¹ cm¹, attributable to -* transitions of the aromatic amino acids tyrosine, tryptophan, and phenylalanine. The binding of metal ions to the phenolic groups of the tyrosine residues in the specific metal binding sites of apo-Tf leads to the production of two new absorption bands centered at ca. 240 nm and ca. 295 nm in the UV-difference spectra. This has been widely exploited for metal titration and thermodynamic studies. Typical difference spectra are shown in Fig. 5, in which two new bands centered at 241 and 295 nm appear and increase in intensity with time after addition of 2 mol Eq of Bi³⁺ as [Bi(Hcit)], indicative of a slow complexation of Bi³⁺ in the specific binding sites of apo-Tf. When transition metal ions bind to apo-Tf, there are often additional intense tyrosinate-to-metal charge transfer (LMCT) bands in the visible region of the spectrum (400-500 nm;  ca. 4-9 × 10³ M¹ cm¹) that are also diagnostic of site-specific binding. For example, the Fe³⁺ complex is orange-red with a band at ca. 465 nm, the Cu²⁺ and Co³⁺ complexes are yellow, and the Mn³⁺ complex is brown (Aisen et al., 1969). Electronic absorption spectroscopy also allows determination of the strength of metal binding to Tf (Sun et al., 1999). The strength of binding of metal ions has been found to correlate well with metal ion acidity, and the most readily hydrolyzed (most acidic) metal ions bind most strongly to transferrin (Li et al., 1996b; Sun et al., 1997a). This provides a basis for the prediction of unknown stability constants for metal·Tf complexes and allows the discovery of new metal ion (e.g., Ti⁴⁺) binding to Tf (Sun et al., 1998a). Table 1 lists the binding constants for therapeutic metal ions with Tf.

View larger version (21K): [in this window] [in a new window]   Fig. 5.   Detection of metal binding to transferrin via UV different spectroscopy. Bi(III) citrate (2 mol Eq) is added to apo-transferrin solution at pH 7.4, 5 mM bicarbonate. The slow uptake of Bi3+ by transferrin is evidenced by the appearance of two bands at 241 and 295 nm due to deprotonation of phenol groups of Tyr residues of transferrin (Li et al., 1996a).

Iron binding to Tf induces protein structure changes from an open to a closed conformation, which is thought to be crucial for receptor recognition. Therefore it is important to study the structural changes induced by other metal ions. Although many crystal structures are available for either apo-Tf or holo-Tf, there appears very few for other metal-Tf (Shongwe et al., 1992; Smith et al., 1996). X-Ray crystallographic studies of human lactoferrin have demonstrated that Ru³⁺ coordinates directly to the imidazole nitrogen of His253, one of the Fe³⁺ ligands in the iron binding cleft of the N-lobe, with displacement of a chloride ligand. At least one indazole ligand remains coordinated to the Ru³⁺ (Smith et al., 1996). However, this study could not provide whether the protein would form the "closed" structure with Ru complexes. Modeling by superposition of normal closed structure suggests that certain Ru³⁺ complexes may allow closure of the protein (Smith et al., 1996). Small angle X-ray scattering has also been used to provide direct structural information on the conformational changes induced by metal ions. It was shown that In³⁺ induced the same domain closure as Fe³⁺ (Grossmann et al., 1993b).

Isotopic labeling of Tf can provide assignments of resonances for specific amino acid residues of Tf, which enhances the usefulness of NMR spectroscopy in exploring conformational changes in the protein. By means of two-dimensional NMR and single site mutations, the S-¹³CH₃ resonances of all the five Met residues of the N-lobe and nine Met residues of intact Tf have been tentatively assigned (Beatty et al., 1996; He et al., 1999b). The Met residues are well spread throughout the protein and occupy several environments. Some of the Met resonances (e.g., Met464 and Met109) are sensitive to metal binding even though these residues are far away from the metal binding site (>10 Å). Similar changes in ¹H/¹³C shifts of Met resonances are observed for Fe³⁺, Ga³⁺, and Bi³⁺ indicating that they induce similar conformational changes (Sun et al., 1998b). Ti⁴⁺ also induces a similar structural change as Fe³⁺ (Guo et al., 2000). This approach provides a useful technique for probe protein conformational changes induced by metals under biologically relevant conditions (Sun et al., 2001).

There is increasing interest in the use of metal-containing compounds in medicine, such as the use of platinum complexes in cancer chemotherapy, gold compounds in the treatment of arthritis, gallium and indium in diagnosis and radiotherapy, and bismuth in anti-ulcer medication (Abrams and Murrer, 1993). One concern is how we can ensure that a metal-based therapeutic or diagnostic agent reaches its target site. One way is to incorporate features that are recognized specifically by the target. We can seek to use natural recognition mechanisms such as the Tf/TfR recognition system. The natural Tf cycle for the delivery of Fe³⁺ to cells offers an attractive system for strategies of drug delivery and targeting since TfR can bind strongly to a range of other metal ions apart from Fe³⁺, and many heterometal·Tf complexes are still recognized by the TfR.

1. Ga³⁺ and In³⁺. The mechanisms by which Ga³⁺ is transported into target sites are of fundamental importance since gallium compounds have been used extensively both in the diagnosis and the treatment of human cancers (Tsuchiya et al., 1992; Bernstein, 1998). It has been shown previously that Ga³⁺ binds to Tf in the specific Fe³⁺ binding sites with a similar affinity, attributed to the similarity between these two metal ions (Harris and Pecoraro, 1983). In vivo studies using ⁶⁷Ga find that all gallium in blood is present in plasma (with traces in leukocytes) and is tightly bound to Tf (Clausen et al., 1974). There has been a quite controversy about whether ⁶⁷Ga uptake is a Tf-independent or -dependent process and whether tumors and normal tissues differ in the mechanism of uptake. Early studies have shown that the uptake of ⁶⁷Ga by cultured murine tumor cells can be significantly stimulated by the addition of exogenous transferrin to the tissue culture medium (Harris and Sephton, 1977). In contrast, a decrease in ⁶⁷Ga uptake by certain tumor cells has been noticed following the addition of Tf to the incubation medium (Vallabhajosula et al., 1981). The uptake study of ⁶⁷Ga³⁺ by human leukemic cell line HL60 demonstrated that both a transferrin receptor-dependent and a Tf-independent mechanism exist. HL60 cells incorporated about 1% of the Ga³⁺ dose over 6 h in the absence of Tf. However, the presence of Tf could increase cellular Ga³⁺ uptake approximately 10-fold (Chitambar and Zivkovic, 1987). Anti-Tf receptor monoclonal antibody inhibited Ga³⁺ uptake, and decrease in the density of cellular TfR led to corresponding decreases in the Tf-dependent uptake of Ga³⁺ (Chitambar and Zivkovic, 1987). Cell surface-bound Ga₂-Tf displayed similar kinetics as Fe₂-Tf during the first 10-min uptake, suggesting that the initial internalization of Ga₂-Tf closely resembled that of Fe₂-Tf (Chitambar and Zivkovic-Gilgenbach, 1990). However, unlike ⁵⁹Fe, a small fraction of internalized ⁶⁷Ga was released from cells by an unknown reason. Ammonium chloride inhibited the internalization of both ⁶⁷Ga and ⁵⁹Fe, indicating that ⁶⁷Ga-Tf uptake by HL60 cells involves initial internalization into acidic receptosome and followed by dissociation of ⁶⁷Ga and Tf and subsequent trafficking of each to separate compartments (Chitambar and Zivkovic-Gilgenbach, 1990). Ga₂-Tf disrupts TfR-mediated cellular uptake of iron as results of inhibition of the iron-containing M2 subunit of ribonucleotide reductase (Chitambar et al., 1988, 1991). Transferrin-enhanced uptake of ⁶⁷Ga was also found in other cell lines (Table 1). Therefore, it can be concluded that the TfR pathway is of primary importance in the incorporation of Ga³⁺ into the cytoplasma of cells displaying this receptor. However, ⁶⁷Ga can also enter tumors and other cells by a Tf-independent mechanism, which is probably also used by iron (Chitambar and Zivkovic, 1987; Weiner et al., 1996); this becomes apparent when Tf is in short supply or saturated with iron or other metal ions (Sohn et al., 1993).

2. Bi³⁺, Ti³⁺ and Ru³⁺. Bi³⁺ complexes are in widespread use in the treatment of ulcers (Sun et al., 1997b; Sun and Sadler, 1998; Briand and Burford, 1999; Sadler et al., 1999); Ru³⁺ compounds are potential anticancer agents, which are often active against metastases but not against the primary tumors (Kratz et al., 1994; Clarke et al., 1999); Ti⁴⁺ complexes have been shown to exhibit high antitumor activities against a wide range of murine and human tumors with less toxic side effects than cisplatin (Köpf-Maier and Köpf, 1987; Harding and Mokdsi, 2000). There are two titanium complexes, titanocene dichloride and budotitane, now in clinical trials (Köpf-Maier and Köpf, 1987; Keppler et al., 1991). All these metal ions have been found binding to Tf strongly in the specific iron sites (Table 1). Therefore, Tf may act as a carrier to deliver these therapeutic metal ions into the cells. Cell uptake experiments showed that Ti₂-Tf and Bi₂-Tf can block both membrane binding and cellular uptake of Fe₂-Tf (Guo et al., 2000). These experiments provide evidence that both bismuth and titanium are likely transported via a similar mechanism as iron, i.e., TfR-mediated endocytosis. Recognition of Bi₂-lactoferrin by IEC-6 rat intestinal cells (Zhang et al., 2001) may also have implications for bismuth antimicrobial action. It is likely that bismuth may block the pathway of iron transport into the bacteria and cuts iron supply required by the bacteria for its growth. A recent study showed that Ti⁴⁺ does not bind strongly to DNA bases at physiological pH but forms strong complexes with nucleotides only at low pH values (below 5) (Guo and Sadler, 2000). Therefore, Tf may serve as a carrier to deliver titanium complexes to tumor cells and to prevent hydrolysis of Ti⁴⁺ complexes at neutral pH. Titanium is subsequently released as a result of acidic microenvironment in tumors than in normal tissue (Yamagata and Tannock, 1996), and targets DNA. Ru³⁺ complexes were reported bound to both albumin and Tf with an 80% portion binding to albumin and the remainder to the latter (Messori et al., 2000; Frasca et al., 2001). Injection of Ru³⁺-Tf resulted in high tumor uptake of the metal (Som et al., 1983; Ando et al., 1988; Srivastava et al., 1989), which suggests that Tf uptake appears to be the more important mode of transport of Ru³⁺ anticancer complexes to the tumor. The Ru₂-Tf exhibits a significantly higher antitumor activity against human colon cancer cells than the albumin-bound complex or the Ru³⁺ complex itself (Kratz et al., 1994, 1996), probably attributed to the Tf-mediated uptake mechanism, which may lower ruthenium toxicity by preventing it from other binding or uptake until it has been delivered to the cells.

	VI. Transferrin Conjugates in Site-Specific Drug Delivery

Top Previous Next References

A. General Methods of Preparation of the Conjugates

1. Chemical Linkage. Various therapeutic agents have been chemically linked to Tf. Several studies have been reported to link doxorubicin with Tf via the formation of a Schiff base (Yeh and Faulk, 1984; Barabas et al., 1992; Sizensky et al., 1992; Bërczi et al., 1993a,b; Singh et al., 1998). Glutaraldehyde was frequently used for this purpose. Briefly, certain amounts of Tf and doxorubicin both dissolved in 150 mM NaCl were directly mixed and followed by the addition of glutaraldehyde (in 150 mM NaCl) dropwise. The coupling procedure was stopped by the addition of ethanolamine. The conjugate was subjected subsequently to purification and characterization. The conjugates prepared in such a way were found to exert cytotoxicity (Yeh and Faulk, 1984; Barabas et al., 1992; Bërczi et al., 1993a,b).

2. Protein Engineering. A novel alternative approach to using the Tf uptake pathway for cellular delivery of therapeutic agents is to incorporate the drug into the structure of Tf using recombinant protein engineering (Ali et al., 1999a,b). A therapeutic peptide sequence cleavable by the human immunodeficiency virus type 1 protease has been inserted into various regions of human serum Tf by protein engineering techniques. These insertions were cloned and expressed using a baculovirus expression vector system. The results showed that mutant proteins retained the native Tf function and that the inserted peptide sequence was surface-exposed. Most importantly, two of the mutants could be cleaved by human immunodeficiency virus-1 protease (Ali et al., 1999a,b). In another study, Tf was fused to mouse-human chimeric IgG3 at different positions by protein engineering, and the resulting fusion protein was found to be able to cross the BBB and to target the brain (Shin et al., 1995). These studies have demonstrated the potential of Tf not only as a carrier protein for site-specific drug delivery, but also for developing new therapeutic agents for a broad spectrum of diseases in the future.

B. Cellular Uptake and Efficacy of the Conjugates

1. Transferrin-Doxorubicin. Although doxorubicin (Adriamycin) is an effective and widely used cancer chemotheraputic agent; cardiotoxicity and emergence of resistance tumor cell lines significantly limit its utility in clinical practice (Kovar et al., 2002). Various approaches have been devised to circumvent these limitations, among which is the attachment of cytotoxic drugs to suitable carrier proteins, such as Tf, that accumulate in tumor tissue. The Tf-doxorubicin conjugate has been shown to exhibit greatly increased cytotoxicity relative to unconjugated doxorubicin toward a variety of culture cell lines (Table 2). Cellular uptake experiments revealed that free doxorubicin at concentrations below 1 × 10⁷ M had little effect on K-562 cell, while Tf-doxorubicin conjugate inhibited 75% of cellular activity. When normal peripheral blood mononuclear cells were tested against the conjugate, the 50% inhibitory concentration was found to be 1.4 to 1.7 × 10⁶ M, at which concentration over 85% of K-562 cells were inhibited (Sizensky et al., 1992). In another cytotoxicity assays, K-562 cells were exposed to doxorubicin or Tf-doxorubicin and cultured for 16 to 18 h. It was found that at a concentration of 0.05 µM, 37% inhibition of K-562 observed for the conjugate compared with 5% for the free drug (Bërczi et al., 1993b). In vivo studies showed that the life span of tumor-bearing mice was significantly increased when they were treated with Tf-doxorubicin conjugate (increase in life span 69% versus 39% with doxorubicin); although no long-term survivors was observed, the tumor burden with conjugate-treated mice was much smaller compared with free doxorubicin-treated mice (Singh et al., 1998).