Bioconjugate Chem. 1893, 4, 139-145
138
Synthesis, in Vitro Kinetics, and in Vivo Studies on Protein Conjugates of AZT: Evaluation as a Transport System To Increase Brain Delivery B. Mitra Tadayoni,t Phillip M. Friden, Lee R. Walus, and Gary F. MUSSO* Alkermes Inc., 64 Sidney Street, Cambridge, Massachusetts 02139. Received September 28, 1992
One method to increase the delivery of polar drugs to the central nervous system is via drug-protein conjugates with proteins that interact with and cross brain capillary endothelial cells. As a model for drugs containing a reactive hydroxyl group, AZT was conjugated via a succinate linker to two such protein carriers, the highly cationic histone H1 and an anti-transferrin receptor antibody, OX-26. The protein carriers were selected on the basis of their ability to interact with brain capillary endothelial cells by absorptive or receptor-mediated events, respectively. An in vitro pH profile of the rate of AZT release indicated that the observed hydrolysis proceeds by a specific base-catalysis mechanism. At 37 "C, the release of AZT proceeded at a rate approximately 10-fold faster (&s 8 X lo4 mi+) than expected for a simple ester (AZT succinate; Kobs 1.25 X lo4 min-l). Using simple model systems, product analysis revealed that intramolecular cyclization of the succinate linker accounts for the observed rate enhancement. Drug delivery in vivo was assessed using immunohistochemical techniques and quantitative brain uptake measurements with singly and doubly labeled AZT-OX-26 conjugates. Immunohistochemical staining of brain sections showed the colocalization of AZT and OX-26 in the brain vasculature. Therefore, drug can be linked to the antibody without affecting the targetingproperty of the antibody. Furthermore, an in vivo time course using radiolabeled conjugate showed that AZT is delivered to the brain capillaries but is not transported into the brain parenchyma with the antibody. These results demonstrate that drugs conjugated via a succinate linker to the anti-transferrin receptor antibody OX-26 could provide a useful method of selective drug delivery to brain capillary endothelial cells.
-
INTRODUCTION The delivery of polar drugs to the brain does not generally occur in sufficient levels to achieve a therapeutic benefit due to the presence of a restrictive membrane barrier isolating the brain from the remainder of the body. The anatomical basis for this barrier is the endothelial cell layer which separates the bloodstream from the cellular environment of the brain (2). Because of this blood-brain barrier (BBB), most drugs do not gain access to the brain unless they can utilize a specific transport mechanism or the drug has been sufficiently modified to enhance its lipophilic properties. An alternative method to increase brain delivery of drugs is to tether the drug to a macromolecular carrier via a specific linkage strategy (3). In this scenario, the specific targeting property of the carrier is used to selectively ferry the drug to the brain. In our laboratory, we have shown that, upon intravenous administration in rats, the anti-transferrin receptor antibody OX-26 binds brain capillary endothelial cells and crosses the BBB in a time-dependent receptor-mediated fashion. Approximately 0.5-1 % of the total injected dose of OX-26 was delivered to the brain parenchyma via bolus intravenous administration (6). As it had been reported that cationized proteins display enhanced transport across the BBB via absorptive-mediated transcytosis (1, 7), we chose to explore this phenomenon with an endogenously cationic protein, histone H1. Histone H1 has 25% lysine residues which allowed for high drug loading while maintaining its native cationic character. Our interests were to explore the feasibility of increasing the delivery of 3'-azido-3'-deoxythymidine (AZT) to the brain as a model for other therapeutic drugs containing
* Address correspondenceto this author at Alkermes, Inc., 64 Sidney St., Cambridge, MA 02139. + Current address: ImmunoGen, Inc., 148 Sidney St., Cambridge, MA 02139. 1043-1002l93l2904-0 139$04.00Io
-
hydroxyl groups and to explore the properties of these conjugates for drug release in vitro and in vivo. For this, we developed conjugation strategies with histone and OX26 as drug carriers. I t has been assumed that AZT traverses the cell membrane by a nonfacilitated diffusion mechanism (B), although systemically administered AZT does not penetrate the BBB in high levels (9,10). For conjugation, a succinate linkage was initially selected with the thought that release of AZT could be facilitated through the action of esterases either in the capillary endothelial cells (11)or in the brain parenchyma. AZT was succinylated by treatment with succinic anhydride. Activation of the carbonyl group of 5'-AZT succinate with N-hydroxysuccinimide allowed the facile preparation of drug conjugates. We then explored in vitro and in vivo properties of these conjugates. Here we detail our findings concerning the kinetics and mechanism of the release pathway of the succinate-linked drug. In addition, we describe our findings on the effectiveness of AZT delivery in an in vivo model system. EXPERIMENTAL PROCEDURES Material and Methods. AZT, succinic anhydride, dicyclohexylcarbodiimide (DCC), and N-hydroxysuccinimide were obtained from Aldrich. [l4C1AZT (50 mCi/ mmol) was purchased from Sigma. Nuclear magnetic resonance spectra were recorded with a Bruker AC-250 spectrometer. Infrared spectra were performed with a thin film on NaCl plates using a Mattson Sirius 100 FTIR spectrometer. UV spectra were obtained with a Beckman DU-65 spectrophotometer. High-performance liquid chromatography (HPLC) were performed on Hewlett-Packard 1090M system equipped with a diodearray detector and temperature-controlled autosampler. Elemental analyses were performed by Galbraith Laboratories (Knoxville, TN). The monoclonal antibody OX26 was purified by protein A-agarose chromatography (12). 0 1993 American Chemical Society
140 Bioconjugate chem., Vol. 4, No. 2, 1993
The eluted material was analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and enzyme-linked immunoabsorbent assay (ELISA). Histone H1 (bovine, lysine-rich) was purchased from US. Biochem and checked by SDS-PAGE. Protein concentrations were determined by the method of Lowry (13). Both H1 histone and OX-26 yielded standard curves comparable to bovine serum albumin by this method. All other chemicals and solvents used in this study were reagent or HPLC grade. AZT Succinate, 2. To a solution of 267 mg (1.00 mmol) of AZT (l),31 mg (0.25 mmol) of 44dimethylamino)pyridine (DMAP), and 100 mg of Na2S04 in 10 mL of freshly distilled pyridine was added 120 mg (1.2 mmol) of succinic anhydride. The resulting slurry was stirred under argon at room temperature for 3 h. The reaction mixture was then filtered and the filtrate was concentrated in vacuo to give a yellow oil. The oil was diluted with H20 (45 mL) and applied to a DEAE-Sephedex A-50 column (2.5 X 7.5 cm) using linear gradient of H2O (500 mL) and 0.20 M triethylammonium bicarbonate (500mL). The major band (AZT succinate) was eluted with 0.1 M triethylammonium bicarbonate. The eluate was concentrated and collected on a Sep-pack CIScartridge. After washing with water, the AZT succinate was eluted with 8020 CHsCN/O.l% aqueous trifluoroacetic acid. The title compound was obtained in 80% yield as a flaky powder: mp = 35 "C; 'H NMR (CDC13) 6 1.91 (3 H, s), 2.41 (t, 2 H, J = 4 Hz), 2.5-2.7 (m, 4 H), 4.15 (9,1H), 4.35-4.80 (m, 3 H), 5.98 (t, 1H, J = 9 Hz), 7.47 (s, 1H); IR 3184-2930 (br), 1734-1635, 1472, 1272, 1162, 1099 cm-'. NHS Ester of AZT Succinate, 3. A solution of 180 mg (0.48 mmol) of AZT succinate 2 and 56.0 mg (0.48 mmol) of N-hydroxysuccinimidein 8 mL of freshly distilled tetrahydrofuran (THF) was cooled to 4 "C. To this solution was added 100 mg (0.48 mmol) of dicyclohexylcarbodiimide and the mixture stirred at 4 "C for 2 h. The reaction mixture was concentrated in vacuo to yield a white solid. Purification by flash chromatography (1.25 cm X 15cm silica gel, eluted with 70:20:10 ethyl acetate/hexanes/ methanol) followed by recrystallization from CHzCld hexanes yielded the title compound: lH NMR (CD30D) 6 1.93 (s, 3 H), 2.43 (t, 2 H, J = 6.0 Hz), 3.05 (t, 2 H, J = 6 Hz), 4.10 (m, 1 H), 4.4 (m, 3 H), 6.08 (t, 2 H, J = 6.6 Hz), 7.20 (s, 1 H); IR 3328-2851, 1734, 1646, 1558, 1456, 1271,1205, 1096 cm-'. AZT-histone H1. To a precooled solution (4 "C) of AZT-NHS ester 3 (0.23 mg, 500.0 nmol) in 100 pL of dioxane was added 0.375 mg (25.0 nmol) histone H1, which was dissolved in 1 mL of 0.1 M of N-(2-hydroxyethyl)N-(2-ethanesulfonic acid) sodium salt buffer (HEPES), pH 7.5, and the mixture-stirred at 4 OC for 2 h. The protein conjugate was isolated from the crude reaction mixture by chromatography on Sephadex G-25 (Pharmacia PDlo), equilibrated, and eluted with phosphate/saline buffer, pH 7.4 a t room temperature. Fractions of 1.0 mL were collected and analyzed at 266 nm for the AZT. The desired protein conjugate was collected in fractions 4 and 5 whereas free drug was observed in fractions 10 and 11. The conjugates were adjusted to pH 5.0 with acetic acid and stored at -20 "C until use. Each conjugate was further characterized by its mobility on SDS-PAGE minigels (12 ?& acrylamide). The conjugates migrated as a broad band slightly higher than the native protein standard. AZT-OX-26. For the preparation of AZT conjugates to the antibody OX-26, buffer and reaction conditions similar to those employed for the histone conjugation were used. The amount of the activated AZT derivative was
Tadayoni et ai.
adjusted so as to prepare conjugatesof lower drug to protein ratios, as it was found that high drug loading on lysine residues of this antibody (>12:1) altered its ability to recognize antigen (data not shown). Thus, a maximum of 10 equiv of AZT actived ester 3 was added to the antibody solution for preparation of these conjugates. Conjugates were purified and characterized in a manner analogous to that used for the histone conjugates except that SDSPAGE electrophoresis was conducted using 8%acrylamide gels. The conjugate solutions were adjusted to pH 5.0 with acetic acid and stored at -20 "C until use. [3-[ (Phenethy1amino)carbonylll-oxopropoxylAZT, 4. A solution of 116 mg (0.25 mmol) of NHS ester in 5 mL of freshly distilled THF was cooled to 4 "C. To this solution was added 30.0 pL (0.25 mmol) of phenethylamine and the mixture stirred overnight under Ar. The solvent was removed in vacuo to yield a white solid. Purification by flash chromatography (1.25 cm X 15 cm silica gel, eluted with 5050 ethyl acetate/hexanes) gave the title compound in 60% yield: mp = 105 "C; 'H NMR (CDCl3) 6 1.93 (s, 3 H), 2.46-2.73 (m, 6 H), 2.80 (t, 2 H, J = 7.0 Hz), 3.46 (q,2 H, JI= 5 Hz, J2 = 7.3 Hz), 4.00 (m, 1H), 4.20-4.50 (m, 3 H), 5.71 (t, 1H, J = 5.2 Hz), 6.16 (t, 1H, J = 6.3 Hz), 7.17-7.34 (m, 5 H), 9.2 (s, 1H); IR 33222929,1688 (br), 1545, 1468,1437,1363,1271, 1161,1049 cm-l; HRMS calcd for C22Hz&l&, (M+)470.1914, found 470.1884. N-Phenethylsuccinimide6. A sample of 100mg (1.00 mmol) of succinic anhydride was dissolved in 5 mL of freshly distilled pyridine. To this mixture was added 126 pL of phenethylamine and the mixture was refluxed for 5 h. The resulting yellow solution was concentrated in vacuo to give a yellow oil. The yellow oil was diluted with 10 mL of methylene chloride, washed with 1N HC1, and dried over MgS04. Evaporation of the solvent under reduced pressure gave succinic acid N-phenethylamide 5, which was then dissolved in acetic anhydride and heated to reflux in the presence of DMAP. The solvent was removed in vacuo and the gummy residue was dissolved in saturated NaHC03, extracted with CH2Cl2,washed with 1 N HC1, and dried over MgS04. Concentration of the yellow solution yielded a yellow oil which was purified by flash chromatography (1.25 cm X 15 cm silica gel, eluted with 5050 ethyl acetate/hexanes) to give the title compound as a white solid in 80% yield: mp = 133 "C; lH NMR (CDC13)6 2.65 ( ~ , H), 4 2.88 (t, 2 H, J = 7.5 Hz), 3.75 (t, 3 H, J = 7.6 Hz), 7.19-7.33 (m, 5 H). IR 2973-2919, 1684, 1410, 1311, 1216, 1093, 991, 927, 793 cm-'. Anal Calcd for C12H13N0~:C, 70.90; H, 6.40; N, 6.70. Found: C, 70.67; H, 6.46, N, 6.91. HRMS calcd for C12H13N02 203.0970, found 203.0939. ['4C]AZT-[3H]OX-26. [l4C1AZTsuccinate NHS ester was synthesized using the methods described for 3, starting with [14C]AZT (specific act. = 50 mCi/mmol). TLC analysis of the final NHS derivative of AZT succinate indicated that the final product contained -20% unreacted AZT succinate. As AZT succinate was unreactive in the conjugation procedure, the above material was used without further purification to prepare radiolabeled AZTOX-26 conjugates. Following the conjugation and purification, the specific activity of the conjugate was determined to be 78 mCi/mmol, which corresponds to a 1.6:l drug to protein ratio. In order to obtain a dual-labeled sample, [14C]AZT-OX-26, was tritiated using N-succinimidyl propionate (13)(100 Ci/mmol) to generate [l4C1AZT-i3H1 OX-26 with the final specific activity of 800 mCi/mmol with respect to 3H and 78 mCi/mmol with respect to 14C.
Bioconlugste Chem., Vol. 4,
Protein Conjugates of AZT
Kinetic Studies. The in vitro release of AZT from the protein conjugates was assessed using an HPLC-based assay. Separations were achieved using a Vydac CIS column (catalog no. 218TP54) by applying a linear acetonitrile gradient (0% to 40% over 40 min) in 0.1% trifluoroacetic acid. The conjugates were incubated at 37 OC with various buffers spanning the pH range of 1.1-9.7. Kinetic time courses were developed by time programmed injection/analysis of the conjugate incubations. For the model compounds, the disappearance of the starting material (-dS/dt) as well as the appearance of the product (dP/dt) could be monitored. For the protein conjugates, only the appearance of the product could be accurately measured. Area integration of the AZT eluate (dPldt) was converted to AZT concentration using the calculated linear regression line derived from the data in Figure la. A column wash at the end of the HPLC gradient served to elute bound protein and conjugates. Immunohistochemistry. Immunohistochemicalstudies were performed essentially as described (6). Female Sprague-Dawley rats (125-150 g) were anesthetized with halothane prior to injection of conjugate via the tail vein. At sacrifice, the animals were perfused with either icecold phosphate-buffered saline (DPBS),pH 7.4, or 50 mM sodium acetate, pH 5.0, to wash blood from the vasculature. The brains were removed and frozen in liquid nitrogen. Thirty-micron sections were cut on a cryostat and placed on gelatin-coated slides. The slides were fixed in acetone (lOmin, 23 "C) and stored at -20 "C. OX-26 was detected immunohistochemically using a biotinylated horse antimouse IgG/horseradish peroxidase Vectastain ABC kit from Vector Labs following the manufacturer's protocol. AZT was detected using a rabbit anti-thymidine antisera which cross reacts with AZT (A gift from Dr. D. Stollar, Tufts University) and a biotinylated goat anti-rabbit antisera from Vector Labs. Controls included sections from uninjected animals as well as animals injected with OX-26 alone or AZT alone. Additional controls included the omission of either primary or secondary antibody in the detection on sections from conjugate-injected animals. Capillary Depletion. The procedure of Triguero et al. (15)was followed as described previously(6). The entire capillary pellet and aliquots of the unfractionated homogenate and the parenchyma supernatant fraction were solubilized in Soluene 350 (Packard) at room temperature overnight prior to liquid scintillation counting. All data were collected as dpms using a Beckman TD5000 liquid scintillation counter with quench correction. Data are expressed as percent of the injected dose of radioactivity in either the parenchyma or capillary fractions derived per rat brain. The amount of radioactivity in the brain due to blood contamination was subtracted in all cases. RESULTS AND DISCUSSION
Synthesis of AZT Derivatives. The esterification of the 3'-hydroxyl group of 5' protected nucleotides with a succinate group is a well-documented reaction most commonlyapplied as the attachment point of nucleic acids for automated synthesis on a solid support (18). As AZT bears only a single hydroxyl group at the 5' position, the attachment of a succinate group to the 5' position appeared feasible. It was envisionedthat this linkage would be stable for preparation of drug conjugates, yet allow for sensitivity to esterases or nucleases in vivo (11). Prior to conjugate synthesis, succinic anhydride in the presence of pyridine was used to acylate the 5'-hydroxyl of AZT, thus forming 5'-AZT succinate, 2. Succinate 2 was then activated at the carboxyl group using equimolar amounts of N-hy-
No. 2, 1993 141
Scheme I
0
u
m'03Ni 63 4
/
0
m\ol,,,,,~~~~ I
l
O
3
droxysuccinimide and dicyclohexylcarbodiimideto form the N-hydroxysuccinimido (NHS) ester 3. The synthesis of the AZT derivatives and conjugates is outlined in Scheme I. Synthesis of Conjugates. The resulting NHS ester 3 was used to acylate amine groups on the carrier proteins histone H1 and OX-26, at 4 OC. Using highly purified active ester, the acylation of protein amino groups typically occurred in 70-75% yield. Conjugation Ratio. A combination of several methods was used to determine the drug loading on each carrier. Drug loading on histone H1 was estimated spectrophotometrically (AZT, E266 = 8000, Histone, €266 3000). At drug to protein ratios >5:1, the contribution of histone to the total absorbance at 266 nm approached zero. The absorbance of pooled fractions from gel filtration chromatography was therefore measured at 266 nm. Subsequently, the contribution to A280 from AZT (A2~lA280= 1.67) was calculated from the estimated drug loading to yield the net absorbance at 280 nm, which was then used to calculate the histone concentration (t2m 5200). Typically, 85-90% of the histone was recovered in the conjugatefraction. Histone concentration in the conjugate pool was also determined by the method of Lowry (13) and agreed to within 10% of the value estimated by the above spectrophotometric method. Drug loadings up to 151were achieved by this procedure. Having determined the drug and protein concentration in unknown conjugate samples, a calibration table for AZT concentration versus area percent intergrated at 266 nm was established by HPLC. The integrated area of AZT was linear with respect to AZT concentration in the range of 0.1-4 nmol and is illustrated in Figure l a (r = 0.999). The drug loading on the OX-26 carrier could not be determined using the spectrophotometric technique described above due to the strong absorbance of OX-26 at 280 nm (OX-26 e280 = 210000) and lower drug loading levels. Using a combination of HPLC analysis of fully hydrolyzed conjugate (see In Vitro Release Kinetics; 2 h at pH >9.0) and Lowry protein determination, drug and protein concentrationswere determined. For OX-26, drug loading ranged from 1:l to 7:l. In Vitro Release Kinetics. The detailed HPLC kinetic studies described below were performed on highly drugloaded AZT-histone conjugatesas these conjugatesyielded superior signal to noise ratios. Using the calibration table
-
-
142
Bioconjugate Chem., Vol. 4, No. 2, 1993 '*0°
1
1000
-
800
-
Tadayoni et al.
Table 11. Kinetics of Hydrolyses of AZT Conjugates
/
J
substrate buffer AZT-histone 0.1 N HCI AZT-histone citrate AZT-histone citrate AZT-histone citrate
AZT-histone MOPS
200 600
AZT-OX-26
AZT-histone HEPES AZT-OX-26 HEPES
400
"
MOPS
1
I
I
I
I
0
1
2
3
4
AZT-histone AZT-histone AZT-histone AZT-OX-26 AZT-histone AZT-histone
phosphate
HEPES HEPES
HEPES
pH Kob X lo5 ( m i d ) t l p (min)
1.1 3.1
6.30 0.47
4.1
1.30 1.20
4.9 5.9 6.5 7.4 7.5 7.7 7.9 8.4 8.5
borate 9.7 50 % rat serum
3.20 5.70 78.0 37.0 180.0 330.0 930.0
670.0 13000
310.0
11OOO 147 446 53 307 57 720 21 656 12 158 888 1 925 385 210 74 103 5 233
Scheme I1
:y,
I
,
,
,
.
.
,
0 0
200
400
800
800
Tim (minut..)
Figure 1. (a) AZT standard curve. A plot of AZT concentration
versus area percent integration using HPLC reveals a linear correlation. These data were then used to calculate the concentration of AZT released from the conjugate during the HPLC kinetic analysis of the conjugates. (b) Release of AZT from an AZT-H1 conjugate at pH 8.4 at 37 "C. A plot of integrated area vs time demonstrates that AZT is rapidly released from the conjugate at pH 8.4. Using the standard curve derived in Figure la, all of the conjugated AZT appears released by 200 min. Table I. Chemical Models To Study the Hydrolysis of AZT Conjugates substrate AZT-histone AZT succinate
buffer
pH
HEPES HEPES
7.4 7.4
K 0 b (min-') 7.80 X 10-4 1.25 X 10-4
4
HEPES
7.4
10.20X lo4
t l / (~m i d 888 5544 679
derived for AZT, the kinetic time course of AZT release from a histone conjugate (15AZT:histone, AZT So= 0.190 mM, histone So = 0.0125 mM) was first measured at pH 8.4 at 37 "C as shown in Figure lb. The data show that the release of AZT from the conjugate was rapid at pH 8.4, appeared to be first-order, and leveled off within 2 h. Furthermore, the amount of AZT released from the conjugate correlated to 100% of the amount of AZT spectrophotometrically determined to be on the conjugate, suggesting that the reaction had gone to completion. Next, the release of AZT from the histone conjugate and a model system, AZT succinate, was evaluated at pH 7.4. This data, as shown in Table I, indicated that the rate of AZT release from the conjugate is significantly faster than that of the control, AZT succinate. This observation suggests that the majority of the drug is released from the conjugate via a mechanism other than simple ester hydrolysis. To understand the mechanistic pathway by which AZT is released from the conjugates, a model system was
designed with a simple amine to mimic the t-amino group of protein lysines. Phenethylamine served as good model because it forms an aliphatic amide with active esters and contributes an identifiable UV absorbance (phenyl)to aid in the interpretation of HPLC kinetic analysis. The synthesis of this compound starting from the active ester of AZT succinate is also outlined in Scheme I. A kinetic study of the this model conjugate, 4, yielded rates of AZT release comparable to the histone conjugate as documented in Table I. Analysis of the products from this reaction, separated by HPLC, showed that 4 hydrolyzed into two major compoundsat equivalent rates. These products were identified as AZT and a phenyl derivative eluting later than the nucleotide or starting material. In order to identify the other reaction product, the two possible derivatives of phenethylamine were synthesized, phenethylamido succinate 5, and N-phenethylsuccinimide 6. The HPLC elution characteristics of phenethylamine 5 and 6 indicated that the observed major product in the hydrolysis reaction was indeed imide 6. Diode-array analysis of the reaction product and authentic imide eluted by HPLC confirmed the assignment of this product to the reaction pathway. Next, the experimental measurement of the complete pH-rate profile was determined using the 151AZThistone conjugate and the HPLC kinetic assay. After appropriate pH adjustment, the starting concentration of conjugate solutions was measured by UV spectroscopy. Kinetic analysis was also performed on the OX-26-AZT conjugate at pH 6.4,7.4, and 8.4 and yielded rates comparable to the histone conjugate. The kinetic data from both the histone conjugate and the OX-26 conjugate are summarized in Table 11. The pH-rate profile for the histone conjugate is graphically depicted in Figure 2. The experimental results at pH values surrounding neutrality (6-9) document a 10-fold increase in Kobs with each pH unit increase. This trend is characteristic of a specific base catalysis. Generally, the neutral amide is too weakly nucleophilic toward simple esters to effect catalysis. However, intramolecular acyl transfer, involving the anion of an amide group is a common reaction of suitable substituted esters (16). The reaction involves specific base catalysis followed by intramolecular nucleophilic attack by equilibrated amide anions as depicted in Scheme 11. Under these conditions, the overall first-order
BloConjug?te Chem., Vol. 4, No. 2, 1993 143
Protein Conjugates of AZT
kgKobr.
3
/
-I
PH
Figure 2. Dependence on pH of the rate of hydrolysis AZT-H1 at 37 "C in aqueous buffer. The profile illustrates the linear relationship between drug release and pH above pH 6.0 which is indicative of a specific base catalysis mechanism. The general shape of the pH profile is reflective of succinate esters.
c
z.-
rate constant, Kobe,can be described by eq 1,wherein
a
intramolecular
AZT- H1
V =
-d[AZT-Hl] dt
=
(1) intermolecular
d[AZT]
= ki,t,JAZT] + ki,t,[AZT] dt
refers to the intramolecular imide formation and k i n b r refers to the saponification of the ester in the reaction medium. Further analysis of the HPLC data derived from the hydrolysis of the model compound 6 reveals that Jtintra JZinkr X 10 for this reaction. This suggests that the simple ester hydrolysismechanism accounts for only about 10% of the total reaction pathway. Overall, the kinetic pH profile is reminiscent of that described for the hydrolysis of succinate ester prodrugs (17);however, the rates are 10-fold faster. Succinates have been used in a variety of prodrugs, linkers, and supports over the years. Generally, it was thought that this attachment strategy would provide a bond which could be liberated by either an enzymatic process or saponification. The data here suggest that amide esters of succinates liberate the ester portion via an alternative mechanism and under mild conditions. This observationpotentially impacts a number of commonly employed chemical reactions that have used this bond. These include solid-phase nucleic acid chemistry where an amide-ester bond on a succinate linker serves to tether the growing oligonucleotide to the controlled pore glass support (18) and some commercially available protein cross-linking agents (19). Our data not only indicate that the linkage can be cleaved under mild conditions but also indicate that added nucleophiles are unnecessary (18). The potential lability of this linkage a t neutral pH would warrant a cautionary note about the use of amide esters of succinates in procedures involving extended exposure to neutral or basic aqueous conditions. Localization of Conjugates in the Brain Vasculature. The anti-rat transferrin receptor antibody OX-26 has been shown to bind to brain capillary endothelial cells and to traverse the blood-brain barrier following iv administration into rats via the tail vein (4, 6). The antibody also can function as a carrier for the delivery of
-
t
Figure 3. Immunohistochemical detection of an AZT-OX-26 conjugatein brain sections. horse anti-mouse IgG (a) and rabbit anti-thymidine (b) antisera were used to detect the carrier antibody and the passenger drug, respectively, in the brain vasculature afer iv administration of 0.2 mg of AZT-OX-26 conjugate (XlOO).
methotrexate to the brain (5,s). Similar studies with AZThistone H1 conjugates were unable to detect either component on the brain capillary endothelial cells nor in the brain (20). Therefore, the histone conjugates were not studied further in vivo. Immunohistochemistry was used to determine if, following conjugation with AZT, OX-26 retained the ability to target and bind to brain capillary endothelial cells in vivo. At 1 h postinjection of the conjugate, both the antibody carrier and the AZT "passenger" were visualized in the brain using either anti-mouse IgG antisera or antithymidine antisera, respectively. A staining pattern similar to that reported for OX-26 alone (6) was revealed when sections from the conjugate-injected animals were stained for the carrier antibody (Figure 3a). A similar pattern of immunolocalizationwas obtained when sections were stained with the anti-thymidine antisera (Figure 3b), indicating colocalization of OX-26 and AZT in the brain vasculature. Control experiments indicated that the observed immunostaining was specific for either OX-26 or AZT and that the localization of the drug to the brain vasculature was dependent on the attachment of the drug to the anti-transferrin receptor antibody (datanot shown). Distribution of AZT-OX-26 in Brain Parenchyma and Capillaries. The results of the experiments described above showed qualitatively that the AZT-OX-26 conjugate binds to brain capillary endothelial cells following iv administration. To quantitate the amount of conjugate that reaches the brain as well as assess the vascular and parenchymal distribution, capillary depletion studies were performed using radiolabeled conjugate. This technique
144
Tadayonl et el.
Bioconjugate Chem., Vol. 4, No. 2, 1993 0.4 i
this strategy to prepare antibody carrier-drug conjugates with other drugs offering greater therapeutic potential. ACKNOWLEDGMENT We would like to thank Dr. John Kozarich and Prof. Julius Rebek, Jr. for helpful discussions. We would also like to thank Fariba Tehrani and Joseph Eckman for their technical support. The gift of anti-thymidine and antihistone antisera from Dr. David Stollar, Dept. of Biochemistry, Tufts University, is greatly appreciated. We thank Dr. Ruth Starzyk, Alkermes, Inc., for her data on the in vivo properties of histones. LITERATURE CITED
U.U
I
0
(1) Kumagai,A. K., Eisenberg, J., and Pardridge, W. M. (1987)
. 5
10
IS
20
Time &ours)
Figure 4. Delivery of AZT to the brain using the OX-26 carrier. Capillary deletion data on the AZT-OX-26 conjugate using double-labeled material. The data are expressed as percentage of the injected dosein the parechyma fraction and capillary pellet as a function of time. The values are means i SEM (n= 3 rats , per time point). Legend parenchyma fraction, OX-26 (01AZT ( 0 ) ;capillary pellet, OX-26 (e),AZT (MI. separates the brain into parenchyma and capillaryfractions (6,151.Comparison of the relative amounts of radioactivity in the two fractions as a function of time determines whether the conjugate crosses the BBB. These studies employed a dual-labeled conjugate in which the AZT was 14C-labeledand the antibody carrier was 3H-labeled. This configuration allowed independent measurement of the disposition of both the carrier and the drug within the brain. Throughout the time course studied (24 h), very similar levels of OX-26and AZT were seen in the capillary fraction of the brain. These capillary levels decreased over time, suggesting that the drug and antibody were not being retained by the capillary endothelial cells (Figure 4). As the levels of OX-26 in the capillary fraction decreased,the levels in the parenchyma fraction increased, indicating that the antibody was migrating from the capillaries to the parenchyma in a time-dependent manner. In contrast, the levels of AZT in the brain parenchyma did not rise significantly,suggesting that the majority of the drug was released in the endothelial cells and was not transported across the BBB. Alternatively, the AZT could have been released to the parenchyma, but was rapidly cleared into the cerebrospinal fluid. On the basis of our findings in the in vitro studies, intramolecular hydrolysis of AZT from the conjugate ( t l p < 4 h, 50% rat serum) could be responsible for the time-dependent loss of AZT seen in these studies. SUMMARY The results presented here demonstrate first that AZT can be conjugated to protein carriers to serve as delivery agents using a succinate linkage. Second,that a prominent intramolecular pathway exists for release of ester-linked compound based on the kinetic and mechanistic studies of these conjugates. Third, that drugs linked to macromolecular carriers via a succinate linkage can be used to target compounds to a specific location, such as the brain capillaries with OX-26, with the added advantage of a relatively fast release of drug. We are currently employing
AbsorptiveMediated Endocytosisof CationizedAlbumin and a j3-Endorphin-Cationized Albumin Chimeric Peptide by
Isolated Brain Capillaries: Model System of Blood-Brain Barrier Transport. J. Biol. Chem. 262, 15214-15219. (2) Goldstein, G. W., and Betz, A. L. (1986) The Blood-Brain Barrier. Sci. Am. 255, 70-79. (3) Rodwell, J. D., Ed. (1988) Antibody-Mediated Delivery Systems, Marcel Dekker, Inc., New York, NY. (4) Jefferies, W. A., Brandon, M. R., Hunt, S. V. Williams, A. F., Gatter, K. C., and Mason, D. Y. (1984) Nature 312, 162163. (5) Musso, G., Abelleira, S. M., and Morrow, A. R. Manuscript
in preparation. (6) Friden, P. M., Walus,L., MUSSO, G., Taylor,M., Malfroy, B., and Starzyk, R. (1991) Anti-Transferrin Receptor Antibody
and Antibody-DrugConjugatesCrossthe Blood-BrainBarrier. F’roc. Natl. Acad. Sci. U.S.A. 88, 4771-4775. (7) Pardridge, W. M., Triguero, D., and Buciak, J. (1989)
Transport of Histone Through the Blood-Brain Barrier. J. Pharmacol. Erp. Ther. 251,821-826. (8) Zimmerman, T. P., Mahony, W. B., and P m , K. L. (1987) 3‘-Azido-3‘-deoxythymidine, an UnusualNucleosideAnalogue that Permeates the Membrane of Human Erythrocytes and Lymphocytes by Nonfascilitated Difussion. J. Biol. Chem. 262, 5748-5754. (9) Klecker, R. W., Jr., Collins, J. M., Thomas, R., Jenkins, J. F., Broder, S., and Myers, C. E. (1987) Plasma and Cere-
brospinal Fluid Pharmacokinetics of 3‘-Azido-3’-Deoxythymidine: A Novel Pyrimidine Analogue with Potential Application for the Treatment of Patients with AIDS and Related Diseases. Clin. Pharmacal. Ther. 41, 407-412. (10) Terasaki, T.,Pardridge, W. M. (1988)RestrictedTransport of 3’-Azido-3’-Deoxythymidineand Dideoxynucleosides Through the Blood-Brain Barrier. J. Infect. Dis.158,630632. (11) Lidinsky, W. A., andDrewes, L. R. (1983) Characterization
of the Blood-Brain-Barrier Protein Compositionof the Capillary Endothelial Cell Membrane. J.Neurochem. 41,13411345. (12) Ey, P. L., Prowse, S. J., Jenkin, J. R. (1978) Isolation of
Pure IgGI, IgGa, and IgGzb Immunoglobulins from Mouse Serum Using Protein A-Sepharose. Zmmunochem. 15,429436. (13) Lowry, 0. H., Rosebrough,N. J., Farr, A. L., and Randall, R. J. (1951) Protein Measurement with the Folin Phenol Reagent. J. Biol. Chem. 193, 265-275. (14) Kummer, U. (1986)Tritium Radiolabelingof Antibodiesto High Specific Activity with N-Succinimidyl (2,3-aH)-
propionate. Use in Detecting and Analyzing Monoclonal Antibodies. Methods Enzymol. 121, 670-678. (15) Triguero, D., Buciak, J., and Pardridge, W. M. (1990) CapillaryDepletion Method for quantification of Blood-Brain Barrier Transport of CirculatingPeptides and PlasmaProteins. J. Neurochem. 89-275, 1-7. (16) Bemhard, S. A., Berger, A., Carter, J. H., Katchalaki, M., Sela, M., and Shalitin, Y. (1961) Co-operative Effecta of Functional Groups in Peptides. I. Aspartyl-Serine Derivatives. J. Am. Chem. SOC.84, 2421-2434,
Protein Conjugates of AZT
(17) Anderson, B. D., Conradi, R. A., and Lambert, W. J. (1984) Carboxyl Group Catalysis of Acyl Transfer Reactions in Corticosteroid 17- and 21-Moonoesters. J. Pharm. Sci. 73, 604-611. (18) Matteucci,M.D., andcanthers, M. H. (1980)The Synthesis
Bloconjugate Chem., Vol. 4, No. 2, 1993 145
of Oligodeoxyppimidineson a PolymerSupport Tetrahedron Lett. 21, 719-722. (19) Abdella, R. M., Smith, P. K., and Royer, G. P. (1979) Biochem. Biophys. Res. Commun. 87,734-742. (20) Ruth Starzyk, personal communication.
