234

reviews

Frank M. Huennekens

Tumor targeting: activation of prodrugs by enzyme-monoclonal

antibody conjugates

Selective delivery of lethal levels of drugs to tumors, without concomitant damage

to normal tissues, is a major challenge in cancer chemotherapy. Prodrugs used in

conjunction with enzyme-monoclonal antibody conjugates that can target tumors

and convert prodrugs to their active drug forms in situ, offer exceptional promise

in achieving this objective. Synthesis of prodrugs, acquisition of appropriate enzymes

and monoclonal antibodies, and manufacture of conjugates afford considerable

flexibility in experimental design.

Chemotherapy is one of the primary modalities in the treatment of cancer, yet most regimens are only palli- ative. Drugs that have performed spectacularly in vitro often fail to achieve similar efficacy in animal-tumor models or cancer patients. This unfortunate situation is due largely to the fact that agents can be employed at very high concentrations in cell-culture systems, but attempts to reproduce these conditions in vivo usually result in unacceptable damage to normal tissues. For this reason, drugs are usually administered in sub- lethal doses and, although the tumor burden may be reduced dramatically, the surviving cells, characterized by diabolical mechanisms of drug resistance, soon dominate the population and do not respond to further therapy. Mechanisms of resistance to methotrexate (MTX), for example, include an increased level of its target enzyme (dihydrofolate reductase), enzyme with decreased affinity for MTX, and decreased uptake or polyglutamylation of the drug.

The selective delivery of drugs to tumors is, there- fore, a major goal in cancer chemotherapy. Various strategies have been proposed: (1) encapsulation of drugs in organ-seeking liposomesl; (2) incorporation of toxin genes into tumor-infiltrating lymphocytes2; (3) act~yation of inert prodrugs (i.e. drugs modified chemically to prevent their uptake or cytotoxic action) by tumor-elaborated enzymes3; and (4) conjugation of drugs, toxins or radionuclides to monoclonal anti- bodies (mAbs) specific for tumor-associated antigens 4, or to growth factors directed towards receptors on tumors s. The strategy tmderlying the use ofdrug-mAb conjugates is illustrated in Fig. 1. The conjugates may

F. M. Huennekens is at the Division of Biochemistry, Department of Molecular and Experimental Medicine, The Scripps Research Institute, LaJolla, CA 92037, USA.

be cleaved extracellularly and the released drug taken up by active transport or diffusion, or the intact con- jugate may be internalized and the drug released intra- cellularly. The success achieved with drug-mAb con- jugates, although limited, has been encouraging, but the procedure is hampered by, among other reasons, the limited number of drug molecules that can be car- ried by each antibody, and the necessity of using a link- age that can be cleaved chemically, or by endogenous enzymes, to release the drug. A recent addition to the repertoire of selective-delivery strategies combines the better features of the aforementioned strategies (3) and (4) by using prodrugs in conjunction with enzyme- mAb conjugates (Fig. 1). In this approach, the prodrug is converted extracellularly to the parent drug by an enzyme selected for this purpose, followed by uptake of the drug. The advantage of this approach is that the catalytic power of only a single enzyme molecule attached to the antibody provides a means for gener- ating large quantities of active drug at the tumor site. The prodrug/enzyme-mAb strategy (reviewed in Kefs 6-8) has also been termed 'Antibody-Directed Enzyme Prodrug Therapy' (ADEPT) 6 or 'Antibody- Directed Catalysis' (ADC) 9.

Prodrugs and activating enzymes The preparation ofprodrugs involves transformation

of the parent drugs into structures that are not taken up by cells, or are incapable of binding to and inhibit- ing their intracellular targets. Prodrugs should be able to be readily synthesized, available in quantity, and stable to chemical or enzymic degradation in vivo. Uptake of most drugs occurs by diffusion or carrier- mediated active transport. Diffusion-dependent drugs are usually large, non-ionic heterocycles, such as etoposide (VP-t6), or doxorubicin (adriamycin),

TIBTECH JUNE 1994 (VOL 12) © 1994, Elsevier Science Ltd

235

reviews

which can be rendered non-diffusible by attachment of a charged group such as phosphate. Etoposide, for example, when treated with POCl3, produces, rather surprisingly, only a single phosphorylated product whose structure [determined by elemental analysis, and nuclear magnetic resonance (NM1K) spectroscopy and mass spectrometry] is shown in Fig. 2. By contrast, MTX enters cells via a folate-transport system, and this process can be blocked by converting the (x-carboxyl of the substrate to an amide or ester. Specifically, ot-peptides of MTX, prepared in an overall yield of ~50% by couphng an L-glutamyl-0t-L-amino acid to 4-amino-4-deoxy-10-methylpteroic acid 11, have proved to be suitable prodrugs. Derivatization of a drug to decrease cytotoxicity is illustrated by 5-fluorocyto- sine: the prodrug is non-toxic, but its deaminated product (5-fluorouracil), after conversion to the deoxyribonucleotide, is an inhibitor of thymidylate synthase.

Another consideration in the preparation ofprodrugs is the creation of structures that can be converted enzymically into active drugs. In addition to appro- priate specificity and catalytic activity, the enzymes should have favorable structural features, such as long-term stability under physiological conditions, monomeric form, low molecular weight and the lack of a requirement for cofactors. Less clear are the cri- teria regarding the origin of the enzyme. Mammalian enzymes (murine or human, depending upon the intended use) have the advantage of being non- immunogenic during in vivo studies, but their endogen- ous counterparts may produce undesirable activation of the prodrugs at sites distant from the tumor. Bacterial enzymes, on the other hand, may be innocuous in this latter respect, but they are potentially immunogenic. With regard to enzyme specificity, nature often lends a hand by providing an array of catalysts, some of which display a remarkable tolerance for non-physio- logical substrates. Alkaline phosphatase (AP), for example, readily removes phosphate groups from large organic molecules such as etoposide phosphate 1° (Fig. 2a), and MTX peptides [e.g. methotrexate ot-phenylalanine (MTX-Phe; Fig. 2b)] are hydrolyzed by carboxypeptidase-A (CP-A) al. These reactions may be slow compared with those involving the enzymes' natural substrates, but they are sufficient to provide lethal amounts of active drug during the lengthy division time of tumor cells. There must be a proper balance, however, between the rate at which the drug is generated (governed by its concentration and the K m and kca t values of the enzyme) and the rates at which it is taken up producfvely by the cells or lost by diffusion from the tumor site.

In addition to the etoposide phosphate-AP and MTX peptide-CP-A systems discussed above, a number of other prodrug-enzyme pairs have been developed. These' include: p-N-bis(2-chloroethyl) aminobenzoylglutamic acid, cleaved by carboxypep- tidase-G 2 (Fig. 2c) to yield a benzoic-acid mustarda2; phenoxyacetamide derivatives of doxorubicin and melphalan, activated by penicillin amidase7; mito-

Prodrug

Enz~....._. ~ Drug

mAb

[ Drug

Y

Drug

mAb

Drug mAb

Figure 1 Strategies for targeting drugs to tumor cells using drug-monoclonal antibody con- jugates, or enzyme-monoclonal antibody conjugates with prodrugs. (Enz, enzyme; mAb, monoclonal antibody; - - - , cleavage site.)

mycin phosphate, activated by Ap13; the glucuronide of an aniline mustard, activated by ~-glucuronidase14; 5-fluorocytosine, activated by cytosine deaminase15; 2-(L-et-aminoacyl) derivatives of MTX, activated by aminopeptidase16; and a cephalosporin derivative of vinblastine, activated by [3-1actamase iv. In the latter example, a group at Lilly has utihzed an ingenious mechanism for activation of prodrugs: attack of the enzyme on the amide bond in the [3-1actam ring of the cephem can lead to expulsion of a variety of drugs linked to its C-3' position. Not yet reported, but cer- tainly feasible in principle, would be the creation (for example, by addition of a polar group) of prodrug forms ofnucleosides, such as AraC or AZT, which are used in the chemotherapy of cancer and viral diseases. These nucleoside prodrugs would not be taken up by cells, but could be converted to the active drugs by appropriate enzymes.

Monoclonal antibodies and conjugates Monoclonal antibodies destined for use in pro-

drug/enzyme-mAb regimens should have the follow- ing characteristics:

TIBTECH JUNE 1994 (VOL 12)

236

reviews

a H

Oo o

(

b C P - A

O f

H2N,.~ LN~ /N,~ O C " r N H ~ CH-"( r ~')

N ~.,,,,,"~ N~I " - CH:.-- N "--~" " ] - - C - - N H - - CH CH 3 I X ~ / I

NH:~ CH3 (~H~2 COOH

c C P , G 2 I

O ] COOH ClCH2CH~,\ / 7 - - ¢ II I I

' N t-~C%~-NH--~H O T p ClCH;,CH2 /

, (~H I 2)2 A P COOH

Figure 2 Enzymic activation of (a) etoposide phosphate; (b) methotrexate-cephenylalanine; and (c) p-N-bis(2-chloroethyl)aminobenzoylglutamic acid. (AP, alkaline phosphatase; CP-A, carboxypeptidase-A; CP-G2, carboxypeptidase-G2; - - - , cleavage site.)

• availability in quantity, reproducibility and purity (as determined by gel electrophoresis under non- denaturing and denaturing conditions); • targeting an antigen that is highly specific to tu- mors, present at high density (>105 cell<), and not internalized; • stability under physiological conditions but cleared rapidly from circulation; • penetrability into tumor masses.

As a result of their superiority with respect to stab- ility and penetrability, F(ab')2 fragments (M r ~-- 90 kDa) are attractive as replacements for the parent antibodies (M r -- 150 kDa). Few mAbs, unfortunately, possess all of the above desirable qualities; some of the better ones are proprietary and are thus not publicly known, or are unavailable for general use. Most of the current mAbs, moreover, are murine in origin and, although they are useful for feasibility studies in xenograft models, they are unsuitable for clinical trials with human subjects because of their immunogenicity. As discussed by Bagshawe 6, this situation can be alleviated to some degree by derivatizing the mAb with polyethylene glycol (PEG), or by treatment of the host with the immunosuppressive agent cyclosporin. The use of chi- meric antibodies, which are produced by splicing the DNA sequences of human immunoglobulins to those of mor0se antigen-binding sites via recombinant DNA (rDNA) technology, is another attractive possibility.

Construction of enzyme-mAb conjugates is rela- tively straightforward, as a result of the extensive tech- nology that has been developed for the manufacture of conjugates for diagnostic purposes [e.g. enzyme- linked immunosorbent assay (ELISA)]. Most com- monly, the proteins are derivatized with separate link- ers that are capable of reacting chemically to form a stable bond. In a recent study from our laboratory is, bovine pancreatic CP-A (M r = 35 kDa)was treated with succinimidyl 4-(N-maleimidomethyl)cyclo-

hexane-l-carboxylate; by conducting the reaction at neutral pH, the linker was attached only to the N-terminus of the protein (rather than indiscrimi- nately to multiple lysine residues). The mAb (KS 1/4), targeted to a carcinoma-associated antigen 19, was treated with N-succinimidyl 3-(2-pyridyldithio) pro- pionate, again at pH 7 to avoid compromising the binding activity. In this instance, approximately four linkers were attached to each mAb (i.e. one per pep- tide). Both derivatizations were carried out in good yield (-70%). Reduction of the disulfide bond in the mAb linker produced a thiol group, which added to the maleimide double bond on the enzyme linker to form a stable thioether bond. Purification of the con- jugate was accomplished by passage of the reaction mixture through a sizing column to remove unreacted enzyme, followed by separation of the conjugate from unreacted antibody by ion-exchange chromatography. Examination of the purified conjugate preparation by non-denaturing PAGE revealed the presence of approximately equal amounts of two components (M r > 200 kDa), corresponding to the 1 : 1 and 2 : 1 (enzyme : mAb) complexes. The specific activity of the conjugate (18 U rag-l), compared with that of the enzyme (60 U mg-1), confirmed that the preparation contained an average of 1.5 enzyme molecules per antibody. Manipulation of reaction conditions affords the possibility of synthesizing conjugates with any desired ratio (up to 4: 1). Measurements with 12sI--labeled conjugate indicated that full antigen-bind- ing activity had been retained. Similar procedures for the preparation and characterization ofenzyme-mAb conjugates have been utilized in other studies cited above. A variant preparation of conjugates involves the use ofbispecific antibodies 2° with binding sites for the tumor antigen and the enzyme. Conjugation is accomplished in situ by a two-step procedure (first allowing the antibody to attach to the cells, followed by coupling of the enzyme to the antibody).

TIBTECH JUNE 1994 (VOL 12)

237

reviews

a 1°-8~

g ~:~ 10 -7

10 -6 ' - i I 10 20

CP-A (mU)

b 100

o~ v

t -

5o o (_9

\Conjugate\

\ 7', \

10 -8 10 -6 10 -4

Prodrug or drug (M)

Figure 3 (a) Cytotoxicity of MTX-Ala towards UCLA-P3 cells in the presence of increasing amounts of carboxypeptidase A (CP-A) (Ref. 18). Cells were seeded at an initial density of 1000 per well and, after allowing 4 h for attachment, indicated amounts of CP-A were added; for each level of CP-A, various concentrations of L, L-MTX-Ala were present. Growth was stopped after 120h, and cells were enumerated by the sulfa- rhodamine B procedure. ID5o values for MTX-Ala (i.e. concentrations of drug to achieve 50% cell kill), obtained for each amount of CP-A, were plotted as a function of amount of enzyme. (b) In an experiment of this type, cells pre-treated with an enzyme-monoclonal antibody conjugate would be exposed to various concentrations of prodrugs. As controls, cells not treated with conjugate would be exposed to the prodrug or to the parent drug. After an appropriate time, cells would be enumerated and growth plotted as a function of prodrug- or drug concentration in order to determine ID5o values.

Cytotoxicity o f prodrug-conjugate systems Prodrugs and their enzyme-mAb conjugates are

evaluated for cytotoxic potential first with tumor cells in vitro, and subsequently by studies with tumor xenografts in nude mice. Regimens showing promis- ing results may then be promoted to clinical trials with human subjects. Prior to in vitro testing, however, it is instructive to determine the capability of the enzyme to produce lethal concentrations of drug from the pro- drug. The CP-A-dependent activation of the MTX peptide, MTX-Ala, illustrates this point 18. UCLA-P3 cells (a human lung adenocarcinoma) were propagated in the presence of fixed amounts of enzyme, each with varying concentrations of prodrug, and IDs0 values (i.e. concentrations of drug required for 50% cell kill) were determined. The data (Fig. 3a) reveal that ~10 milliunits (mU) of enzyme are needed to maximize the ID50 for MTX-Ala, i.e. to approach that of the parent drug, M T X (~10-8M). The phenylalanine derivative (MTX-Phe), however, is a much better sub- strate for CP-A, requiring only about 1 mU of CP-A to achieve IDs0 parity with M T X (R.ef. 21). Infor- mation of this type is useful for predicting the effec- tiveness ofa prodrug-conjugate combination against a cell line in which the number of antigen sites per cell (and, hence, the maximum amount of enzyme that can be acquired via bound conjugate) is known.

In vitro cytotoxicity measurements are conducted by treating tumor cells with excess conjugate, followed by extensive washing of the cells and their exposure to varying concentrations of prodrug. Cells not treated with conjugate but exposed to prodrug or parent drug

serve as controls. Additional controls may include cells treated only with conjugate or with conjugate con- taining an irrelevant antibody plus prodrug. Rep- resentative results for this type of experiment can be illustrated with the use of a hypothetical plot (Fig. 3b): The cell-kill curves (growth of cells relative to un- treated controls versus concentration) and ID50 values of the drug and prodrug provide the boundary con- ditions, and the curve for prodrug with conjugate- treated cells will be intermediate. The objective, of course, is to move the latter curve as close as possible to that of the drug. In the case of the MTX peptides, the superiority of MTX-Phe over MTX-Ala was evi- dent from their IDs0 values (6.3 X 10-8M and 1.5 X 10 -6 M, respectively) (unpublished data). In the absence of conjugate, both prodrugs have inherent IDs0 values of~5 X 10-6M, which indicated that they were about two orders of magnitude less toxic than M T X (-5 X 10-aM). Most of the other prodrugs mentioned earlier have displayed the same ideal characteristics, with respect to: (1) ID50 values similar to those of their parent counterparts when utilized with cells pre- treated with conjugates containing the relevant acti- vating enzymes; and (2) inherent IDs0 values con- siderably higher than those of the free drugs. In general, the mechanisms responsible for cytotoxicity of prodrugs in the absence of conjugates have not yet been elucidated, but it is evident that inhibition of these processes, to the extent that it can be accomplished, would improve selectivity of the systems.

Prodrug-conjugate combinations that meet the in vitro criteria of efficacy are tested subsequently in

TIBTECH JUNE 1994 (VOL 12)

238

reviews

animal models. The nude-mouse xenograft system, which allows human rumors to be propagated as readily measured external masses in mice, is especially useful. Optimizing regimens, however, must be carried out empirically (and laboriously). Labeled conjugate is administered in various dose schedules to determine in vivo stability, clearance of unbound material, and targeting to the tumor. The pharmacokinetics and maximum tolerated dose of the prodrug are defined in a similar manner. Guided by this information, treatment is initiated, and tumor growth or regression is noted.

Impressive results have been reported for the use of prodrug-conjugate systems with in vivo models. The Oncogen group 1°, for example, treated a human colon carcinoma (H3347) in nude mice with a phos- phatase--antibody (L6) conjugate on day 6 after tumor implantation, followed by etoposide phosphate (2 mg day -1 - the maximum tolerated dose - on days 8, 13, 16 and 32). Marked suppression of tumor growth was observed (tumor volume, 13% of the untreated con- trol on day 26). The prodrug (without conjugate) or etoposide itself failed to duplicate these results. Simi- larly, Bagshawe and his colleagues 12 used CP-G 2 con- jugated to the F(ab')2 fragment of an antibody (W14) directed toward human chorionic gonadotropin, in conjunction with sequential doses of p-N-bis- (2-chloroethyl) aminobenzoylglutamate, to treat mice with implanted CC3 choriocarcinoma cells. At the maximum tolerated dose ofprodrug (10 mg), marked suppression of growth was achieved.

In vivo model studies provide some guidance for planning clinical studies, but optimal conditions are again determined empirically. Bagshawe 6 has pre- sented a detailed and incisive analysis of the obstacles that must be overcome before the ADEPT strategy will be successful with cancer patients.

Future prospects In considering the future of the prodrug/

enzyme-mAb strategy in cancer chemotherapy, it is useful to consider briefly the strengths and weaknesses of the various components in the regimen.

• Prodrugs: given the experience and ingenuity of chemists, it may be stated with confidence that most drugs in current use can probably be converted into suitable prodrugs. • Activating enzymes: these abound, and the growing ability ofbiotechnology procedures to alter specificity, increase catalytic activity, and produce large quantities of enzymes augurs well for this component. Use of enzymes of human origin would obviate concerns about adverse immune response. • Monoclonal antibodies: at present, this is clearly the weak link. The acute shortage of mAbs that target specifically to tumor cells emphasizes the need to search for new antigens that may provide better selec- tivity. The advent of human antibodies 22 should solve the immunogenicity problem for this component. Microbeads, containing inunobilized enzymes, that could be implanted directly into tumor masses may, in

certain instances, obviate the need for tumor-target- ing antibodies. • Conjugates: well-established techniques are already available for coupling enzymes to mAbs. Alternatively, fusion proteins 23 produced by constructs hnking the genes for the enzyme and antibody offer the possibility of obtaining conjugates in quantity from bacterial expression systems. This is another fertile area for biotechnology. • Regimens: after a number of optimal regimens have been established empirically, it should be possible to develop some rational guidelines to shorten this pro- cedure. Computer-modeling of relevant parameters, such as the number of tumor cells, doubling time of the ceils, number of conjugates per cell, rate of release of drug, and rate of drug uptake, offers the possibility of bringing quantification and predictability to the therapeutic regimens.

In summary, although perecdon is still a distant goal for the prodrug/enzyme-mAb strategy, the barriers, although formidable, are technical rather than con- ceptual. Achieving the goal should elevate cancer chemotherapy to a higher plateau.

Acknowledgements The author is indebted to Karin Vitols for many

helpful discussions of this subject. Experimental work in the author's laboratory was supported by Out- standing Investigator Grant CA-39836 from the National Cancer Institute (NCI), National Institutes of Health (NIH) and by Grant CH-31 from the American Cancer Society. The expert assistance of Carol Fedoryszyn and Cherlyn Pegues (Division of Biochemistry) in preparing the manuscript is grate- fully acknowledged.

References 1 Gabizon, A. (i989) in Drug Carrier Systems (P, oerdinck, F. H. and

Kroon, A. M., eds), pp. 185-211, Wiley 2 Rosenberg, S. A. et al. (1988) New Engl.J. Med. 319, 1676-1680 3 Carl, P. L. (1983) in Development of Target-On'ented Anticancer Drugs

(Cheng, Y-C., ed.), pp. 143-155, Raven Press 4 R_eisfeld, P,.. A. and Schrappe, M. (1990) in Therapeutic Monoclonal

Antibodies (Borrebaeck, C. A. K. and Larrick, J. W., eds), pp. 57-73, Stockton Press

5 Lappi, D. A. andBaird, A. (1991) Prog. Growth FactorRes. 2, 223-236 6 Bagshawe, K. D. (1989) Br.J. Cancer60, 275-281 7 Senter, P. D. (1990) FASEBJ. 4, 188-193 8 Esswein, A., Haenseler, E., Montejano, Y., Vitols, K. S. and

Huennekens, F. M. (1991) Adv. Enzyme Regulat. 31, 3-12 9 Jungheim, L. N., Shepherd, T. A. and Meyer, D. L. (1992) J. Org.

Chem. 57, 2334-2340 10 Senter, P. D. et al. (1988) Proc. Natl Acad. Sci. USA 85, 4842-4846 11 Kuefner, U., Lohrman, U., Montejano, Y., Vitols, K. S. and

Huennekens, F. M. (1989) Biochemistry 28, 2288-2297 12 Bagshawe, K. D. et al. (1988) Br, J. Cancer58, 700-703 13 Senter, P. D., Schreiber, G. J., Hirschberg, D. L., Ashe, S. A.,

Hellstrom, K. E. and HeUstrom, I. (1989) Cancer Res. 49, 5789-5792 14 Wang, S-H. et al. (1992) Cancer Res. 52, 4484-4491 15 Senter, P. D. et al. (1991) Bioconjugate Chem. 2, 447-451 16 Cheung, H. T. A., Dong, Z., Escoffier, L., Smal, M. and Tattersall,

M. H. N. (1993) in Chemistry and Biology of Pteridines and Folates (Ayling, J. E., Nair, M. G. and Bangh, C. M., eds), pp. 457-460, Plenum Press

IBTECH JUNE 1994 (VOL 12)

239

reviews

17 Meyer, D. L., Jungheim, L. N., Mikolajczyk, S. D., Shepherd, T. A., Sterling, J. J. and Ahlem, C. N. (1992) Bioconjugate Chem. 3, 42-49

18 Haenseler, E. et al. (1992) Biochemistry 31,891-897 19 Fernsten, P. D., Pekny, K. W., Reisfeld, R. A. and Walker, L. E.

(1990) Cancer Res. 50, 4656-4663 20 Sabra, U. et al. (1990) CancerRes. 50, 6944q5948

21 Vitols, K. S., Haenseler, E., Montejano, Y., Baer, T. and Huennekens, F. M. (1992) Pteridines 3, 125-126

22 Borrebaeck, C. A. K., Danielsson, L., Ohlin, M., Carlsson, J. and Caisson, tk. (1990) in Therapeutic Monoclonal Antibodies (Borrebaeck, C. A. K. and Larrick, J. W., eds), pp. 1-I5, Stockton Press

23 Goshom, S. C., Svensson, H. P., Kerr, D. E., Somerville, J. E., Senter, P. D. and Fell, H. P. (1993) CancerRes. 53, 2123-2127

New approaches to drug delivery through the blood-brain barrier

William M. Pardridge

The development of recombinant proteins, monoclonal antibodies, or antisense

oligonucleotides as pharmaceuticals for the brain will require the parallel

development of practical strategies for delivery of these pharmaceuticals in vivo

through the endothelial wall of capillaries in the brain, the blood-brain barrier. The

brain and spinal cord constitute the only organ to be perfused by capillaries having

such a barrier, which excludes the uptake into the brain of circulating molecules

that do not have access to several specialized transport systems within the barrier.

The current challenge for biotechnology is to develop effective drug-delivery

strategies to the brain in parallel with the ongoing drug-discovery programs for this

organ.

The principal products of biotechnology [recombi- nant proteins, monoclonal antibodies (mAbs) and anti- sense ohgonucleotides] will probably not be effective pharmaceuticals for the brain until safe and effective strategies are developed for transporting these drugs through the brain capillary endothelial wall, which constitutes the blood-brain barrier (BBB). The capil- laries perfusing the brain and spinal cord of all ver- tebrates are endowed with specialized epithelial-like tight junctions that eliminate the normal pathways of free diffusion across capillary barriers existing in peripheral organs (Fig. la,b; P,.ef. 1). Owing to the tight junctions (which eliminate paracellular transport) and to the minimal pinocytosis (which restricts trans- cellular transport) in endothelial cells of brain capil- laries, circulating molecules can gain access to inter- stitial fluid in the brain only via one of two main pathways: free diffusion and facilitated transport 2. Free diffusion is proportional to the lipid solubility of small molecules; accordingly, recombinant proteins, mAbs

147. M. Pardridge is at the Department of Medicine, Brain Research Institute, U C L A School of Medicine, Los Angeles, C A 90024, USA.

or antisense ohgonucleotides cannot enter the brain by this pathway. Catalyzed transport includes carrier mediation and receptor-mediation pathways (Fig. 1 c,d).

Carrier-mediated transport at the BBB involves the activity of one of several independent transporters that mediate the flux of glucose, amino acids, purine bases or nucleosides, or other nutrients (Fig. lc). Examples of receptor-mediated transport processes for peptides at the barrier are given in Fig. ld. Circulating insulin gains access to brain interstitial fluid by receptor- mediated transcytosis through the brain capillary endothelium 2. Transcytosis comprises three sequential steps: (1) receptor-mediated endocytosis at the lu- menal or blood side of the BBB; (2) diffusion through the endothelial cytoplasm (a distance of 300 nm); and (3) exocytosis at the ablumenal or brain side of the barrier. Some receptor-uptake systems at the barrier only mediate the process ofendocytosis into the brain capillary endothelium, and the prototype for such receptor-uptake systems is the scavenger receptor, which mediates the endocytosis of acetylated low- density lipoprotein (LDL) (R.e£ 3). A process similar to receptor-mediated transcytosis is absorptive- mediated transcytosis. Lectins, such as wheat-germ

© 1994, Elsevier Science Ltd TIBTECH JUNE 1994 (VOL 12)