Anti-transferrin receptor antibody and antibody-drug conjugates cross the blood-brain barrier. (drug delivery/methotrexate). PHILLIP M. FRIDEN*, LEE R.
Proc. Natl. Acad. Sci. USA Vol. 88, pp. 4771-4775, June 1991 Medical Sciences
Anti-transferrin receptor antibody and antibody-drug conjugates cross the blood-brain barrier (drug delivery/methotrexate)
PHILLIP M. FRIDEN*, LEE R. WALUS, GARY F. Musso, MARJORIE A. TAYLOR, BERNARD MALFROY, AND RUTH M. STARZYK Alkermes, Inc., 26 Landsdowne Street, Cambridge, MA 02139
Communicated by Hilary Koprowski, February 19, 1991
ABSTRACT Delivery of nonlipophilic drugs to the brain is hindered by the tightly apposed capillary endothelial cells that make up the blood-brain barrier. We have examined the ability of a monoclonal antibody (OX-26), which recognizes the rat transferrin receptor, to function as a carrier for the delivery of drugs across the blood-brain barrier. This antibody, which was previously shown to bind preferentially to capillary endothelial cells in the brain after intravenous administration (Jefferies, W. A., Brandon, M. R., Hunt, S. V., Williams, A. F., Gatter, K. C. & Mason, D. Y. (1984) Nature (London) 312, 162-163), labels the entire cerebrovascular bed in a dose-dependent manner. The initially uniform labeling ofbrain capillaries becomes extremely punctate =4 hr after injection, suggesting a time-dependent sequestering of the antibody. Capillary-depletion experiments, in which the brain is separated into capillary and parenchymal fractions, show a timedependent migration of radiolabeled antibody from the capillaries into the brain parenchyma, which is consistent with the transcytosis of compounds across the blood-brain barrier. Antibody-methotrexate conjugates were tested in vivo to assess the carrier ability of this antibody. Immunohistochemical staining for either component of an OX-26-methotrexate conjugate revealed patterns of cerebrovascular labeling identical to those observed with the unaltered antibody. Accumulation of radiolabelod methotrexate in the brain parenchyma is greatly enhanced when the drug is conjugated to OX-26. The levels of various substances in the blood, such as hormones, amino acids, and ions, undergo frequent small fluctuations that can be brought about by activities such as eating and exercise (1). If the brain were not protected from these variations in serum composition, the result could be uncontrolled neural activity. The blood-brain barrier (BBB) functions to ensure that the homeostasis of the brain is maintained. Specialized characteristics of the endothelial cells that form brain capillaries are responsible for this barrier (1, 2). Brain capillary endothelial cells are joined together by tight intercellular junctions that form a continuous wall against the passive movement of substances from the blood to the brain (3, 4). These cells lack continuous gaps or channels connecting the luminal and abluminal membranes, which, in other endothelial cells, allow relatively unrestricted passage of blood-borne molecules into tissue. The isolation of the brain from the bloodstream is not complete; were this the case, the brain would be unable to function properly due to a lack of nutrients and because ofthe need to exchange hormones and other compounds with the rest of the body. The presence of specific transport systems within the capillary endothelial cells, such as those for amino acids, transferrin, glucose, and insulin (2, 5-8), assures that The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
the brain receives, in a controlled manner, all of the compounds required for normal growth and function. For the transferrin receptor, it has been reported (9) that brain endothelial cells are selectively labeled by anti-receptor antibodies in the rat after in vivo administration, presumably due to a high density of transferrin receptors on the surface of brain capillary endothelial cells. A problem posed by the BBB is that, in the process of protecting the brain, it also excludes many potentially useful therapeutic agents. Currently, only substances that are sufficiently lipophilic or are recognized by an existing transport system can readily penetrate the BBB (1, 2). The lipidization of drugs to enhance brain uptake is very nonspecific, in that it will increase the ability of the particular compound to cross all cellular membranes, and is not possible for many compounds. A carrier system that has some degree of organ selectivity and that could be used for a wide range of compounds would have a significant advantage over current methods for circumventing the BBB. Such a carrier system could be developed by exploiting a transport system known to deliver compounds across the BBB, such as that for transferrin. We report experimental results that suggest that antitransferrin receptor antibodies may have utility as drug delivery carriers for the brain. This delivery system takes advantage ofthe high density oftransferrin receptors on brain endothelial cells as well as the ability of these receptors to shuttle molecules across the BBB. Both qualitative and quantitative experiments show that, in the rat, the antitransferrin receptor antibody OX-26 and antibody-methotrexate (MTX) conjugates bind to capillary endothelial cells in the brain in a dose- and time-dependent manner. In addition, we present data that indicate that this antibody and antibody-drug conjugates cross the BBB.
MATERIALS AND METHODS Preparation of Antibody and Antibody Conjugates. OX-26 antibodies were purified from supernatants harvested from cultures of the OX-26 hybridoma cell line [provided by Alan F. Williams, Medical Research Council, Oxford, U.K. (10)] by using protein A-Sepharose column chromatography. The control IgG2a antibody (UPC10) was purchased from Sigma. A hydrazone-linked conjugate between MTX and OX-26 was synthesized by incubating MTX t-hydrazide with antibody that had been oxidized with sodium periodate to form aldehydes from the carbohydrate groups located on the Fc portion of the protein (G.F.M., S. Abelliera, and A. Morrow, unpublished data and ref. 11). The conjugate, which consisted of approximately six MTX molecules per antibody, was analyzed as described (12). The same procedure was Abbreviations: BBB, blood-brain barrier; MTX, methotrexate. *To whom reprint requests should be addressed.
4771
4772
Medical Sciences: Friden et al.
followed for synthesizing [3H]MTX-OX-26 conjugates, using [3',5',7-3H]MTX (Amersham). Immunohistochemistry. Female Sprague-Dawley rats (100-125 g) were anesthetized with halothane for tail vein injections. At sacrifice, the animals were perfused with phosphate-buffered saline (PBS) to remove blood from the vasculature. The brains were frozen in liquid nitrogen for cryostat sectioning. Sections (-30 Im) were fixed in acetone at room temperature and stored at -20TC. Immunolocalization of injected OX-26 in the brain sections was performed using a horse anti-mouse IgG Vectastain ABC kit from Vector Laboratories by following the manufacturer's protocol. MTX was detected using a rabbit anti-MTX antisera from Western Chemical Research and a biotinylated goat antirabbit IgG antisera from Vector Laboratories. All sections were counterstained with methyl green (0.05% in PBS). Capillary Depletion. Antibodies were labeled with either [14C]acetic anhydride or [3H]succinimidyl propionate (Amersham) essentially as described (13, 14). For 14C, the number of acetyl groups per antibody was in the range of 5 to 10; for 3H, less than one propionyl group was added per antibody. Average specific activities were -10 Ci/mmol and 280 mCi/ mmol (1 Ci = 37 GBq) for 3H and 14C, respectively. Routinely, 1 x 106 dpm of 3H and 5 x 105 dpm of 14C were injected per animal. For all experiments, the radiolabeled compounds were injected as a 400-Al bolus into the tail vein of female Sprague-Dawley rats (100-125 g) under halothane anesthesia and the animals were sacrificed at the appropriate time after injection using a lethal dose of anesthetic. A radiolabeled IgG2a control antibody (14C for 3H-labeled OX-26, or vice versa) was co-injected with the labeled OX-26 to serve as a control for nonspecific radioactivity in the brain due to residual blood.
Proc. NatL Acad. Sci. USA 88 (1991)
Capillary depletion experiments were performed essentially as described by Triguero et al. (15). This method removes >90% of the vasculature from the brain homogenate. The entire capillary pellet and samples of the homogenate and supernatant were solubilized overnight with 2 ml of Soluene 350 (Packard) prior to liquid scintillation counting. All data were collected as dpm by using a Beckman model TD5000 liquid scintillation counter. Data are expressed as percent of the injected dose per brain of the antibody in either the parenchyma or capillary fractions. Blood samples were centrifuged to pellet erythrocytes (which did not display significant binding of radiolabeled materials) and the radioactivity in a sample of serum was determined using liquid scintillation counting.
RESULTS Dose- and Time-Dependent Localization of an AntiTransferrin Receptor Antibody to Brain Capillaries in Vivo. Immunohistochemistry was used to localize the antitransferrin receptor antibody OX-26 in the rat brain vasculature after intravenous injection by the tail vein as shown by Jefferies et al. (9). One hour after injection of 0.5 mg of purified antibody, uniform intense staining of the capillaries was observed throughout the brain (Fig. LA). Larger blood vessels as well as capillaries showed the presence of bound antibody. This staining of the brain vasculature was clearly detected at doses as low as 50 ,g per rat and appeared to saturate at a dose of -0.5 mg per rat (data not shown). As reported by Jefferies et al. (9), specific staining of capillaries was not observed in any of the other organs that were examined (data not shown).
FIG. 1. Immunohistochemical detection of OX-26 in brain sections. Sections were cut from brains taken from rats sacrificed 1 hr (A) and 4 hr (B) after injection of 0.5 mg of OX-26 into the tail vein. The tissue sections were stained with horse anti-mouse IgG. The brown reaction product outlines the cerebral vasculature. The punctate staining pattern is clearly visible 4 hr after injection (B). A methyl green counterstain was used to label cell nuclei. (A, xlO00; B. x 400.)
Proc. Natl. Acad. Sci. USA 88 (1991)
Medical Sciences: Friden et al. To determine how early after injection the anti-transferrin receptor antibody could be detected in the brain capillaries as well as how long it persisted, a time course experiment was performed. Rats were injected with 0.5 mg of OX-26 per animal and sacrificed at various times after the injection. Antibody was detected immunohistochemically within the brain capillaries as early as 5 min after injection (data not shown). The staining observed at this time was uniformly distributed along the capillary endothelium. By 1 hr after injection, the observed staining for antibody along the brain capillaries had become more intense (Fig. LA) and, as for the 5-min time point, was quite uniform. By =4 hr after injection, the staining pattern of OX-26 in the brain vasculature changed dramatically and became very punctate (Fig. 1B), suggesting sequestration of the antibody in some manner. The pattern of localization was still punctate 8 hr after injection but returned to a uniform, fainter staining by 24 hr after injection (data not shown). To ensure that the observed staining was unique to the OX-26 antibody, a control antibody (UPC10) of the same subclass (IgG2a) was injected into rats for immunohistochemical localization. No staining of brain vasculature was observed with the control IgG2a at doses and times comparable to those used with OX-26 (data not shown). Distribution of OX-26 in Brain Parenchyma and Capillaries. The above results show qualitatively that the anti-transferrin receptor antibody OX-26 is present in the brain vasculature after i.v. administration. To demonstrate quantitatively that the anti-transferrin receptor antibody accumulates in the brain parenchyma, homogenates of brains taken from animals injected with radiolabeled OX-26 were depleted of capillaries by centrifugation through dextran to yield a brain tissue supernatant and a capillary pellet. A comparison of the relative amounts of radioactivity in the different brain fractions as a function of time should indicate whether the labeled monoclonal antibody has crossed the BBB. Because the vasculature remains intact during the capillary depletion procedure, the antibody that reaches the parenchyma has traversed the basement membrane and perivascular cells associated with the capillaries. As a control for nonspecific association with the brain due to residual blood contamination, a 14C-labeled IgG2a control antibody was co-injected with the 3H-labeled OX-26. The amounts of OX-26 in the brain parenchyma fraction and in the brain capillary fraction plotted as a function of time after injection are shown in Fig. 2. Initially, the amount of OX-26 associated with the brain capillaries rises sharply, reaching a level of -0.6% of the injected dose by 1 hr after injection. This level then decreases over time, dropping to 0.13% of the injected dose by 24 hr after injection. As the amount of antibody associated with the capillaries decreases, the amount associated with the brain parenchyma increases, reaching a value of 0.44% of the injected dose by 24 hr after injection. This redistribution of the radiolabeled OX-26 from the capillary fraction to the parenchyma fraction is consistent with the time-dependent migration of the anti-transferrin receptor antibody across the BBB. To address the question of stability of the radiolabeled carrier in the circulation, total IgG was extracted from serum followed by polyacrylamide gel electrophoresis and autoradiography. This analysis did not reveal detectable levels of OX-26 degradation as late as 48 hr after injection (data not shown). Localization of OX-26-MTX Conjugates in the Brain Vasculature. If an anti-transferrin receptor antibody is to function as a drug carrier, it must retain its ability to bind to and be internalized by brain capillary endothelial cells after conjugation with drug. To test for targeting and binding to brain capillaries in vivo, an OX-26-MTX conjugate (-6 MTX per 1 antibody) was prepared using a hydrazone linkage. One hour
4773
00 '0
0.2
0
10
20
Time (hours) FIG. 2. Time-dependent changes in the disposition of 3H-labeled OX-26 between brain parenchyma and vasculature. Capillary depletion was performed on homogenates prepared from brains taken from animals 1, 4, and 24 hr after injection of radiolabeled OX-26. The percent injected dose of antibody per brain in the parenchyma fraction (o) and vascular pellet (o) are shown. Similar results were obtained using 14C-labeled OX-26 (data not shown). The values shown are mean ± SEM (n = 3 rats per time point). These results are representative of other studies that have been done.
after its injection into the rat tail vein, the brain was removed and sectioned for immunohistochemistry as for the studies described above. Both the antibody carrier and the MTX "passenger" were visualized in the brain using either antimouse IgG or anti-MTX antisera. A staining pattern similar to that seen with OX-26 alone was revealed when sections from the conjugate-injected animals were stained for the carrier antibody (Fig. 3A). A similar pattern was seen when these sections were stained with anti-MTX antisera (Fig. 3B), indicating colocalization of OX-26 and MTX in the brain vasculature. When equivalent amounts of free MTX were injected into rats, no staining of the brain vasculature was observed using the anti-MTX antisera, indicating that localization of the drug to the brain capillaries is dependent on OX-26 (data not shown). Also, no staining was observed when the anti-MTX antiserum was used on sections containing only OX-26. Distribution of the OX-26-MTX Coajugate in Brain Parenchyma and Capillaries. Capillary-depletion studies identical to those described above for OX-26 were performed with an OX-26-MTX conjugate in which the MTX moiety was labeled with 3H at the 3', 5', and 7 positions. As with unconjugated antibody, the amount of label in the capillary fraction at an early time after injection (30 min for these experiments) was greater than that in the parenchyma fraction (Fig. 4). This distribution changed over time such that by 24 hr after injection, -5-fold more of the labeled MTX was in the brain parenchyma than was in the capillaries. These results are consistent with those obtained with unconjugated antibody and suggest that the OX-26-MTX conjugate crosses the BBB. To ensure that these results were not due to contaminating amounts of free [3H]MTX or [3H]MTX that had been cleaved from the conjugate after injection, a mixture of labeled drug and antibody was injected into rats and a capillary-depletion experiment was performed. The amount of [3H]MTX in the different brain fractions was significantly lower for the mixture as compared to the conjugate (as much as 47-fold in the capillary fraction 30 min after injection; Fig. 4). The
4774
Proc. Natl. Acad. Sci. USA 88 (1991)
Medical Sciences: Friden et al.
A
-~~~~~~~~~~~~~~~~~~~~A
iAjf 4.'~ ;'~'' i
A
FIG. 3. Immunohistochemical detection of an OX-26-MTX conjugate in brain sections. Horse anti-mouse IgG (A) and rabbit anti-MTX (B) antisera were used to detect the carrier antibody and the passenger drug, respectively, in the brain vasculature after i.v. administration of 0.2 mg of OX-26-MTX conjugate. Immunohistochemistry was as described in Fig. 1. (x100.)
[3H]MTX in the mixture also did not show the change in distribution of the label between the various brain fractions over time as was seen with the antibody-MTX conjugate or antibody alone (Figs. 2 and 4).
cell layer (16). The time frame in which the OX-26 punctate staining occurs is consistent with uptake by perivascular cells, indicating BBB penetration. Although suggestive, the above results do not demonstrate that the antibody crosses the BBB as opposed to simply
DISCUSSION The results presented herein indicate that an antibody which recognizes an extracellular epitope ofthe transferrin receptor crosses the BBB and can be used to deliver the chemotherapeutic drug MTX to the brain after intravenous administration. Initially, immunohistochemistry was used to corroborate and elaborate on earlier studies by Jefferies et al. (9) demonstrating that anti-transferrin receptor antibodies bind preferentially to vascular endothelial cells within the brain. This phenomenon is most likely attributable to a higher density of these receptors on the endothelial cells that make up the BBB as compared to those in other capillary beds. Physiologically, this is consistent with the fact that the primary, if not only, pathway for iron to enter the brain is by the transcytosis of iron-loaded transferrin across the BBB (2, 7, 8). Immunohistochemical staining of brain sections taken from animals sacrificed at various times after injection shows that the localization of OX-26 in the brain vasculature changes with time. A very punctate pattern of staining is observed from o2 to 4 hr to -8 to 16 hr after injection as compared to the more uniform pattern observed before and after this period. This most likely represents endocytosis and concentration of the antibody in an intracellular compartment followed by eventual exocytosis. One hour after injection by the carotid artery, ferrotransferrin-horseradish peroxidase conjugates have been observed to label perivascular clefts and cells, apparently as a result of crossing the endothelial
*
01.6 -
e
T Uw
Brain tiomogenate: Conjugate Brain
Eiomnogenate: Co-miix
Parenchvma Fraction: Coniugate
0.4 -
._.
tEe
0.2 -
0.0 --
24 hours
30 minutes
Time
FIG. 4. Enhanced delivery of MTX across the BBB using OX-26 carrier. Capillary depletion experiments were used to compare the distribution of an OX-26-MTX conjugate to that of a mixture of antibody and drug. The data are expressed as percent of the injected dose of 3H-labeled MTX per brain in unfractionated homogenate, the parenchyma fraction, and the capillary fraction. The values shown are mean + SEM (n = 3 rats per time point). These results are representative of other studies that have been done. as a drug
Medical Sciences: Friden et al. binding to the luminal surface of the capillary endothelial cells. To address these concerns as well as to obtain more quantitative data, OX-26 was radioactively labeled and its distribution within the brain was determined using the technique of capillary depletion as developed by Triguero et al. (15). The time-dependent changes that were observed in the distribution of radiolabeled OX-26 within the vascular and parenchymal fractions of the brain are consistent with models for the transcytosis of blood-borne proteins across the BBB (2, 17). This pathway consists of (i) binding of the bloodborne compound to a receptor on the luminal side of the capillary endothelial cell, (it) engulfment of occupied receptors into endocytic vesicles and transport of the vesicular contents to the abluminal surface of the cell, and (iOh) exocytosis of the internalized materials on the abluminal surface. There is precedence for the binding of anti-transferrin receptor antibodies to the receptor on the cell surface with subsequent internalization of the antibody-receptor complex (18), suggesting that steps i and ii of the transcytosis pathway described above can occur. With regard to the release of the antibody on the abluminal surface of endothelial cells, it has been demonstrated that certain antibodies to the low density lipoprotein receptor-related protein undergo internalization, escape degradation in the lysosome, and recycle to the cell surface where they dissociate from the receptor in an acidprecipitable form (19). A similar pathway may be followed by the anti-transferrin receptor antibodies in traversing the brain capillary endothelial cells. The use of radiolabeled antibody in the capillary depletion experiments also allows the quantitation of the amount of carrier in the different brain fractions at specific times. However, because the brain is a dynamic system with uptake from the blood and clearance to the cerebrospinal fluid and the blood occurring concurrently, these calculations most likely underestimate the total amount of antibody that reaches the brain parenchyma from the circulation. Conjugates of OX-26 and MTX were synthesized and tested in vivo to examine the ability of this antibody to deliver drugs to the brain. The results presented suggest that the targeting of OX-26 to brain capillary endothelial cells is not markedly affected by the attachment of "passenger" drug molecules. However, although we have clearly demonstrated enhanced delivery of MTX to the brain by using OX-26 as a carrier, the amount that reaches the brain parenchyma is lower than that of antibody alone (0.27 and 0.44% of the injected dose, respectively, 24 hr after injection). It appears that this difference may be a reflection of the physicalchemical properties of the passenger compound, as we have found that some conjugates of OX-26 exhibit enhanced uptake into the brain parenchyma relative to antibody alone (L.R.W. and P.M.F., unpublished data). The use of radiolabeled MTX in the capillary depletion experiments clearly
Proc. NatL. Acad. Sci. USA 88 (1991)
4775
demonstrates that it is the drug which is delivered to the brain parenchyma. The results presented demonstrate the feasibility of delivering drugs across the BBB by using an anti-transferrin receptor antibody as a carrier. Although the present study establishes delivery of a small drug molecule, neuroactive peptides and neurotrophic factors whose effective brain concentrations are in the nanomolar range (or lower) are also potential candidates for delivery using this system. OX-26 has been shown to deliver proteins as large as 40 kDa to the brain and to achieve concentrations in the brain estimated to be in the 10-100 nM range (L.R.W., P.M.F. and M.A.T., unpublished data). Thus, therapeutic levels of neuroactive compounds may be achieved in the brain using an antitransferrin receptor antibody. We thank Biba Tehrani, Joseph Eckman, Susan Abelleira, and Anne Morrow for their excellent technical support. We also thank Floyd Bloom, Timothy Curran, John Kozarich, Paul Schimmel, and Michael Wall for their support and insightful comments. We appreciate the many helpful comments made by our colleagues at Alkermes, Inc. 1. 2. 3. 4.
Goldstein, G. W. & Betz, A. L. (1986) Sci. Am. 255, 74-83. Pardridge, W. M. (1986) Endocr. Rev. 7, 314-330. Brightman, M. W. (1977) Exp. Eye Res. 25, 1-25. Reese, T. S. & Karnovsky, M. J. (1967) J. Cell Biol. 34,
207-217. 5. Pardridge, W. M. (1983) Physiol. Rev. 63, 1481-1535. 6. Pardridge, W. M., Eisenberg, J. & Yang, J. (1985) J. Neurochem. 44, 1771-1778. 7. Fishman, J. B., Rubin, J. B., Handrahan, J. V., Connor, J. R. & Fine, R. E. (1987) J. Neurosci. Res. 18, 299-304. 8. Pardridge, W. M., Eisenberg, J. & Yang, J. (1987) Metabolism 36, 892-895. 9. Jefferies, W. A., Brandon, M. R., Hunt, S. V., Williams, A. F., Gatter, K. C. & Mason, D. Y. (1984) Nature (London) 312, 162-163. 10. Jefferies, W. A., Brandon, M. R., Williams, A. F. & Hunt, S. V. (1985) Immunology 54, 333-341. 11. Kralovec, J., Spencer, G., Blair, A. H., Mammen, M., Singh, M. & Ghose, T. (1989) J. Med. Chem. 32, 2426-2431. 12. Endo, N., Takeda, Y., Umemoto, N., Kishida, K., Watanabe, K., Saito, M., Kato, Y. & Hara, T. (1988) Cancer Res. 48, 3330-3335. 13. Kummer, U. (1986) Methods Enzymol. 121, 670-678. 14. Montelaro, R. C. & Rueckert, R. R. (1975) J. Biol. Chem. 250, 1413-1421. 15. Triguero, D., Buciak, J. & Pardridge, W. M. (1990) J. Neurochem. 54, 1882-1888. 16. Broadwell, R. D. (1989) Acta Neuropathol. 79, 117-128. 17. Broadwell, R. D., Balin, B. J. & Salcman, M. (1988) Proc. Natl. Acad. Sci. USA 85, 632-636. 18. Lesley, J., Schulte, R. & Woods, J. (1989) Exp. Cell Res. 182, 215-233. 19. Herz, J., Kowal, R. C., Ho, Y. K., Brown, M. S. & Goldstein, J. L. (1990) J. Biol. Chem. 265, 21355-21362.
