Drug Targeting Organ-Specific Strategies

Table 11.3. Peptide sequences that have been used as spacers in lysosomotropic drug delivery.

developed for lysosomotropic drug delivery (i.e. delivery with constructs whose final destination is the lysosomes). The enzymes involved in the degradation of the spacers shown in Table 11.3 are normally not present in extracellular areas. For this reason, and since peptide bonds are very resistant towards chemical degradation, the linkages that are formed with these spacers are generally very stable in the circulation.

The peptide spacers shown in Figure 11.3a–b are susceptible to degradation by endopeptidases, which either disrupt the bond between the spacer and drug molecule or, alternatively, one of the internal peptide bonds of the spacer. In a subsequent degradation step the remaining part of the spacer is either attacked by other peptidases or is degraded via non-en- zymatic processes [68,69]. Peptide spacers have been used for the delivery of drug molecules that contain peptide carboxylic acid groups, an example of which is the cytostatic drug methotrexate. It should be noted however that peptide spacers are ill-suited for the linkage of drug molecules which contain non-peptide carboxylic acid groups. Probably, the peptidases involved have high substrate specificities, resulting in the inability to degrade the amide bond that is formed with the drug molecule [63]. Peptide spacers have also been used for the conjugation of drug molecules to a primary amino group (Figure 11.3b). As can be concluded from the listed drugs in Table 11.3, the peptidases involved in the release of these drug molecules tolerate different types of non-peptide primary amino groups.

For drug molecules with carboxylic acid groups, ester bonds seem suitable linkages for lysosomal delivery. The anti-inflammatory drug naproxen has been conjugated to a proteinaceous carrier via an ester linkage by means of an α-hydroxy acid spacer (Figure 11.3c) [70]. Such a spacer, of which L-lactic acid is a typical example, contains in addition to the hydroxyl group a carboxyl group that can be used to link the spacer to the protein. Similar types of linkages can also be formed with a drug molecule containing an hydroxyl group and a spacer with two carboxylic acid groups, such as the succinate spacer (Figure 11.3d). Examples include the conjugates prepared with the corticosteroid dexamethasone and the taxoid paclitaxel [11,71,71,72].

290 11 Development of Proteinaceous Drug Targeting Constructs

The relatively acidic pH of the lysosomes (pH 5) has led to the development of several linkage types which are susceptible to acid-catalysed degradation. These acid-sensitive spacers are relatively stable at the neutral pH of the bloodstream, but become hydrolytically labile at lower pH values. Depending on the type of linkage and the functional group of the drug molecule, three subtypes can be distinguished: cis-aconityl linkers (cis-Aco) (Figure 11.3e–f), Schiff base hydrazones or imino linkages (Figure 11.3g), and phosphamides linkages (Figure 11.3h), respectively. The cis-Aco linker can be used for the conjugation of drug molecules with a primary amino group [73,74]. The amide linkage that is formed between the linker and the drug is chemically destabilized at lower pH values due to the flanking carboxylic acid group (anchiomeric assistance). When the cis-Aco spacer is conjugated via its flanking carboxylic acid group to the protein, the acid-sensitivity of the spacer is lost. In order to prevent such loss of pharmacological activity, several other cis-Aco spacers have been developed that are conjugated via a different chemistry to the protein (exemplified in Figure 11.3f) [75,76]. Often, these spacers contain maleimide and haloalkyl groups for the purpose of the final coupling to the protein, since these groups can be reacted under mild conditions with thiol groups which have been previously inserted into the protein.

The Schiff base hydrazone linkers form an imino-linkage between a carbonyl group of the drug and a hydrazine functionality of the spacer. Such linkages have been formed with drug molecules containing either a keto group (doxorubicin) [77,78], an aldehyde group (streptomycin) [79] or a carboxylic acid group (chlorambucil) [80] (Figure 11.4). Recently, the use of a branched spacer has been reported, which enables the attachment of multiple drug molecules to the spacer [81]. For the conjugation of the hydrazone spacer to a protein, strategies similar to those used in the case of the second generation cis-Aco spacers have been followed.

The third class of acid-catalysed linkages, the phosphamide linkages, are formed between a phosphate group and a primary amino group (Figure 11.3h). This type of linkage has been used for the lysosomotropic drug delivery of nucleotide analogues to the liver and macrophages [22,82,83].

Another type of linkage that is degraded intracellularly is the disulfide bond (Figure 11.3i). Disulfide linkages have been extensively used in the preparation of immunotoxins [84]. For instance, Ricin A immunotoxins are pharmacologically active only when they contain a biodegradable disulfide linkage [85]. With respect to the targeting of small drug molecules, disulfide linkages have been used in conjugates with the angiotensin converting enzyme inhibitor captopril, a drug molecule that contains a free thiol group, and in conjugates with derivatives of colchicine and methotrexate [14, 86–88].

A disadvantage of the disulfide linkage is its relative instability in the bloodstream. Disulfide bonds can be degraded by reducing enzymes or disrupted chemically by thiol-disulfide exchange with free thiol compounds such as gluthathione [89]. Although the latter process will proceed preferably intracellularly, due to the higher intracellular concentration of free thiol compounds, thiol-disulfide exchange can result in premature drug release. Disulfide spacers that contain sterically hindered disulfide bonds showed improved stability towards this non-enzymatic degradation [90,91].

One of the limitations of the use of a proteinaceous carrier is the relative small number of drug molecules that can be conjugated without gross alterations in the structure of the protein. For example, extensive derivatization of an antibody carrier may lead to the loss of its

homing potential if the antigen recognition domain is affected. To circumvent this problem, dextran and poly-glutamic acid (PGA) polymers have been used as bridging molecules for the conjugation of cytostatic drugs (Figure 11.3j–k) [92,93].These polymers were loaded with the drug and subsequently reacted with amino groups or carbohydrate residues of the carrier. This technique enabled the conjugation of about 80 doxorubicin molecules per protein in the case of the dextran bridge, and up to 100 doxorubicin molecules as a result of the PGA linkage. Efficacy studies with tumour cell lines and in vivo tumour xenograft models in mice demonstrated the potential of the above-described conjugates [93].

11.5.2 Extracellular Degradation

In addition to targeting constructs which are endocytosed and which release the active drug substance intracellularly, other constructs are activated outside the target cell. The latter approach is not appropriate for target tissues in which the extracellular fluid is rapidly removed by perfusion, since this would result in a reduction of the time that the drug remained at the target site and consequently systemic redistribution of the drug would occur. Thus, extracellular drug release is preferred for use in compartments or tissues where the rate of perfusion is low. Conditions of slow perfusion are associated with most solid tumours due to their poor lymphatic drainage. In addition, many tumour cells secrete proteolytic enzymes that are capable of degrading extracellular matrix in the process of tumour growth and metastasis. If such enzymes are present in the tumour and in minimal amounts in the extracellular fluid of other tissues, their presence can be exploited for the selective release of the drug in the tumour tissue. Examples of such enzymes include cathepsins, that are normally only present in the lysosomes, and matrix-degrading enzymes such as collagenase or plasminogen activators [67]. These enzymes are all peptidases, and therefore peptide linkers are feasible spacers for use with this approach. In the case of a secreted lysosomal enzyme, the same spacer sequences shown in Table 11.3 can be used for the linkers.

An attractive approach to drug targeting is the delivery of the drug-regenerating enzyme instead of the actual drug substance to the target site. This approach, also referred to as ADEPT (antibody directed enzyme pro-drug therapy), is based on a two-step targeting principle. In the first step, an enzyme is selectively delivered to the target site by means of an an- tibody–enzyme conjugate. In the second step, small-molecule pro-drugs are administered, which will subsequently be activated by the targeted enzyme [94,95].

A point to be noted regarding ADEPT relates to the plasma half-life of the enzyme–anti- body construct. Generally, antibody–enzyme conjugates, are slowly cleared from the central circulation.Their sustained presence in the bloodstream will lead to non-target site activation of the pro-drug. Thus, the enzyme and pro-drug should be consecutively administered after a well-chosen time interval when the concentration of the antibody-enzyme conjugate is still high in the target tissue while being low in the central circulation and non-target tissues.

292 11 Development of Proteinaceous Drug Targeting Constructs

11.6 Recombinant DNA Approaches

The importance of recombinant DNA techniques for the synthesis of drug targeting constructs is rapidly increasing. This approach offers, in theory, the possibility of generating all three components of a drug targeting preparation as outlined in Figure 11.1. A carefully chosen cloning strategy results in a uniform end-product with optimum positioning of the different components. To obtain such a fully genetically engineered drug targeting construct, all three components must be peptides or proteins.With respect to the active drug substance this is likely to be an exception rather than the rule. Some constructs, such as immunotoxins and immunocytokines, that do fulfil these requirements have been studied extensively in drug targeting and will be described in detail.

By using recombinant DNA techniques, modifications in the protein backbone, such as additions, deletions and alterations of amino acids, are easily achieved. These modifications can contribute to improved pharmacokinetic properties of the construct.Additions may consist of the introduction of residues that allow covalent conjugation of drug molecules. Deletions of amino acids can employed to remove membrane-bound regions of a protein, thereby increasing its solubility. Single amino acid modifications can be used to minimize antibody responses and alter the binding specificity and/or the three-dimensional structure of a certain protein.

The final requirement of a recombinant DNA approach to the preparation of a drug delivery construct, is the ability to produce large amounts of the protein. This can by achieved by bacterial, fungal, insect and mammalian expression systems.The choice of system depends on how the expression of the protein is regulated, the required purity and yield of the protein, and whether the protein is toxic to certain types of producer cells. Furthermore, individual scientists may prefer a particular type of expressing system, depending on laboratory facilities, safety considerations and production costs. The possibilities and limitations of different expression systems will be discussed and general guidelines which need to be taken into account when choosing an appropriate strategy, will be mentioned briefly. Thereafter, with the aid of several examples, the development and applications of drug delivery constructs obtained using recombinant DNA technology will be described.

11.7 Recombinant DNA Expression Systems

11.7.1 Heterologous Gene Expression in Escherichia coli

The Gram-negative bacterium E. coli is probably the most widely used host for heterologous protein production. An obvious advantage of this system is its simplicity. The genetics are well characterized, the cells grow fast allowing rapid production and analysis of the expressed protein, and transformation is simple and requires minimal amounts of DNA.

In E. coli foreign genes are normally cloned using inducible promoters such as the lac promoter that is regulated by the lac repressor and induced by isopropyl β-D-thiogalactopyra- noside (IPTG). This controls gene expression and prevents loss of the gene in situations where production of the protein might be toxic to the cells. Stronger synthetic promoters, derived from the lac system, tac and trc promoters, are commercially available. Other common-

294 11 Development of Proteinaceous Drug Targeting Constructs

manipulation is not difficult. In contrast to E. coli, fungi are able to carry out post-transla- tional modifications, such as glycosylation, proteolytic processing, folding and disulfide bridge formation. By applying fermentation technology, clinically and industrially important proteins have been successfully expressed in fungi [100–102].

The most commonly used filamentous fungi for heterologous gene expression belong to Aspergillus and Trichoderma species. The transformation system is based on complementation of auxotrophic mutants or on dominant selection marker genes and results in the integration of the foreign gene into the host genome [103]. Filamentous fungi have effective secretory machinery, allowing for accumulation of proteins in the culture medium. In Aspergillus, expression cassettes consisting of the foreign gene fused to an endogenous glucoamylase and separated by a KEX-2 proteolytic site, have resulted in elevated expression levels [102,104]. The KEX-2 site is effectively cleaved by an endopeptidase in the endoplasmic reticulum during secretion resulting in the correctly processed protein accumulating in the medium.

Several yeast species have been engineered for heterologous protein production but the most commonly used for these purposes are the baker’s yeast Saccharomyces cerevisae and the methylotropic yeast Pichia pastoris. Yeast systems utilize both integrated and extrachromosomal (non-integrated) vectors.

Strong inducible yeast promoters used for protein production in S. cerevisae, include GAL1, GAL5, and GAL7 promoters which are induced by galactose, and repressed by glucose. The wealth of information on its genetics and molecular biology has made S. cerevisae an excellent model organism for protein–ligand and protein–protein interactions. However, for an abundant expression of heterologous proteins, other yeast systems such as P. pastoris with its strong and highly regulated alcohol oxidase (AOX1) promoter, have been more successful. In the last 16 years expression of more than 300 foreign proteins have been reported using this system [105]. Examples of more than 100-fold higher protein yield of recombinant single chain antibodies in P. pastoris compared to E. coli have been reported [106,107].

In summary, fungal expression systems are often an excellent choice for eukaryotic protein expression. However, like any other system, fungi have their own limitations, including the inability to carry out the same types of glycosylation as higher eukaryotic organisms and, in some cases, problems related to accurate protein folding resulting in degradation of the protein.

11.7.3 Baculovirus Expression Systems

Baculoviruses are members of a large group of double-stranded DNA viruses which only infect invertebrates, including insects. The restricted host range makes baculoviruses safer than mammalian expression systems. The most widely used baculoviruses are Autographa californica nuclear polyhedrosis virus and the Bombyx mori nuclear polyhedrosis virus. The host cell most commonly used is Sf9, derived from the fall armyworm Spodoptera frugiperda.

Typically, the foreign gene is placed under the control of the extremely strong polyhedrin promoter, allowing for a highly efficient secretion of the heterologous protein into the insect cell culture medium. Glycosylation and other post-translational modifications occur in the insect cells. Up to 1998, more than 500 different heterologous proteins had been produced by

the baculovirus expression vector, of which more than 95% had the correct post-translation- al modifications [108]. No doubt, this number has rapidly increased since. However, a limitation of the baculovirus system is that optimal expression levels require high-quality growth media, careful culturing and the expression of the foreign protein during the phase in which the producing cells are dying.

11.7.4 Stable Transformations of Insect Cells

Although not as popular as the baculovirus-system, stable transformations of insect cells can be used to circumvent the problems mentioned in the last paragraph. Common hosts include the fruitfly and mosquito. The expressed genes are often under the control of the Drosophila metallothionein promoter. Genes coding for resistance to antibiotics such as hygromycin and neomycin are used as selection markers.

11.7.5 Expression Using Mammalian Cells

The mammalian cell expression system contains all the necessary regulatory machinery for accurate and efficient processing and secretion of eukaryotic proteins, although there may be species differences. Foreign DNA is introduced into the cells either via virus infection or directly, employing chemical (for instance lipocomplexes or calcium phosphate) and physical (electroporation or microinjection) methods. The transcriptional control elements (enhancers and promoters) are complex and vary between mammalian cell types. However, simian virus 40 (SV40) and human cytomegalovirus (CMV) promoters are active in many cell types and are therefore commonly used. Obtaining stable transfected cell lines can be time consuming and therefore a transient expression system is often used for initial analysis. Typically COS (African green monkey kidney) and CHO (Chinese hamster ovary) cells are used for this purpose. An obvious advantage of mammalian cell expression is the possibility of advanced glycosylation. Generally, the yield of heterologous proteins produced in a mammalian cell system is much less than in other expression systems. However, for some proteins, the use of mammalian cells may solve the problems observed in prokaryotic and lower eukaryotic organisms with regard to accurate folding and modifications.

11.7.6 Expression Systems: Concluding Remarks

The choice of an expression system for the production of a drug delivery construct is of vital importance but at the same time a difficult task. Several general considerations when choosing an appropriate expression system are outlined in Table 11.4. The use of microorganisms (bacteria and fungi) results in high yield of the product and they are therefore often preferred by researchers. However, for highly specific therapeutic applications, the use of microorganisms is less favourable since they are unable to carry out the post-translational modifications necessary for activity of the protein. For instance, glycosylation is not possible in E. coli and although possible in fungi, it differs from that in mammalian systems. Use of insect

Figure 11.5. Schematic representation of genetically engineered antibody constructs for drug targeting. Intact antibodies consisting of two heavy and two light chains (a) can be converted into divalent F(ab′)2 fragments (b) or to monovalent Fab fragments (c). These fragments are stabilized via disulfide bridges. Alternatively, the variable heavy and light chain fragments are linked via a flexible linker resulting in a monovalent ScFv (d). Di-, triand tetravalent scFv fragments can be constructed by connecting two, three or four scFv fragments with peptide linkers (e–g) or by introducing a S–S bridge between the individual scFv fragments (h). Non-covalent interactions between scFv fragments are created by introducing leucine zipper sequences into the construct (i) or via streptavidin–biotin interactions (j).

combinant protein size, and thereby in complexity of the construct can be achieved when both the carrier and homing device functions are intrinsic properties of one protein.

Antibodies make up a group of proteins which can be considered to have the properties of both a carrier and a homing device and, as a result, have been used in many drug targeting studies [18,109]. However, the relatively large size (150 kDa) of whole IgG molecules hampers tissue penetration of these molecules. Several modifications of the original antibody structure can be carried out to reduce the size of the IgG molecule. For instance, in natural antibodies (Figure 11.5a) the Fc-region is necessary to activate T-cells of the immune system. Since this function of the antibody is not required in most drug targeting constructs, these domains have been removed by recombinant cloning techniques [110]. The resulting F(ab′)2, and Fab fragments, with molecular weight of around 100–110 and 50–55 kDa respectively, are