Bioconjugate Chem. 1994, 5,262-267
262
Synthesis and Characterization of Monoclonal Antibody-@-Lactamase Conjugates HAkan P. Svensson,’ Philip M. Wallace, and Peter D. Senter Bristol-Myers Squibb Pharmaceutical Research Institute, 3005 First Avenue, Seattle, Washington 98121. Received December 17, 1993”
P-Lactamase from Enterobacter cloacae (PL) was conjugated to the Fab’z fragment of the monoclonal antibody L6 through a thioether linkage. Although LG-Fab’2-PL was capable of activating the antitumor prodrug, 7-(phenylacetamido)cephalosporinmustard, it was impaired in its ability to bind to antigens on the H2981 human lung adenocarcinoma cell line. As a result, studies were undertaken to prepare conjugates with preserved binding activities. LG-Fab’-PL and a dimeric conjugate consisting of two individual L6-Fab’ units linked to a single @Lmolecule (dimeric L6-PL) were prepared by linking L6-Fab’-SH to maleimide-substituted PL. Analysis of these conjugates by SDS-PAGE indicated that the linkage involved heavy-chain thiol groups on L6 that are most likely in the hinge region and are therefore removed from the antigen binding site of the antibody. Cell binding studies revealed that the monovalent conjugate LG-Fab’-PL bound as well as L6-Fab’. Dimeric L6-PL displayed slightly less binding than L6-Fab’z7 but bound substantially better than LG-Fab’z-PL. Lower concentrations of dimeric L6-PL compared to LG-Fab’z-PL were required to convert the prodrug 7-(pheny1acetamido)cephalosporin mustard into the cytotoxic drug phenylenediamine mustard. Localization studies were performed in nude mice with H2981 subcutaneous tumor xenografts. A t 96 h post conjugate treatment, there was no significant difference in tumor concentration between LG-Fab’z-PL and dimeric L6-PL. In contrast, the blood and normal tissue levels of dimeric L6-PL were lower than LG-Fab’z-PL, resulting in improved tumor to blood and tumor to normal tissue ratios. Thus, the conjugation methodology described here may be of use for targeting strategies in which high tumor to nontumor conjugate ratios are required in order to minimize nonspecific toxicity.
INTRODUCTION
The efficacy of monoclonal antibody (mAb’ ) based strategies for the treatment and detection of cancer is often dependent on the ability of the mAb to localize in tumor masses and to achieve high tumor to non-tumor ratios (I). Many studies have demonstrated that valency, avidity, and molecular weight (2-4) can play important roles in mAb in vivo distribution. Consequently, a great deal of research has been directed toward optimizing these properties in order to enhance mAb and mAb-conjugate tumor localization, to accelerate clearance from the blood, and to minimize exposure to normal tissues (5, 6). For certain applications of mAb conjugates (6-81, these factors are critical. A number of recent reports have described the use of mAb-enzyme conjugates for the conversion of relatively noncytotoxic drug precursors (prodrugs) into active anticancer drugs (reviewed in refs 6 and 8). This is a twostep approach to cancer therapy in which a mAb-enzyme conjugate is administered, and after allowing enough time for tumor uptake and systemic clearance to take place, an anticancer prodrug is then given. The targeted enzyme converts the relatively nontoxic prodrug into an active anticancer drug. Considerable evidence has accumulated
* To whom correspondence should be addressed. @
Abstract published in Advance ACS Abstracts, April 15,
1994.
Abbreviations used: PL, Enterobacter cloacae P-lactamase; dimeric mAb-@L, (mAb-Fab’)p-PL; DTT, DL-dithiothreitol; FITC, fluorescein isothiocyanate; IC50, concentration that gives 50% cell kill; LFE, linear fluorescence equivalence; mAb, monoclonal antibody; PBS, phosphate-buffered saline, pH 7.4; SDS-PAGE, sodium dodecyl sulfate polyacrylamide gel electrophoresis; SMCC, N-succinimidyl4-(maleimidomethyl)cyclohexane-1-carboxylate. 1043-1802/94/2905-0262$04.50/0
suggesting that optimal effects require high tumor to blood conjugate ratios (8-1O),which presumably can be affected by the chemistry used for conjugate preparation. Several groups have utilized P-lactamases for the activation of cephalosporin-containing prodrugs (11-13). Recently, we described the in vivo activities of a Fab’z conjugate of Enterobacter cloacae p-lactamase (PL) against a subcutaneous human tumor xenograft in nude mice (12). A significant level of antitumor activity was obtained in spite of the fact that the conjugate was impaired in its ability to bind to cell surface antigens. In this report, we describe new conjugation methodology that preserves mAb binding activity and correlate this with the ability of the conjugate to activate prodrug and specifically localize into subcutaneous tumors in vivo. EXPERIMENTAL PROCEDURES
Materials. Crude E. cloacae penicillinase was obtained from Sigma Chemical Co., St. Louis, MO, and purified according to established procedures, using boronic acid affinity chromatography (12, 14). The synthesis of 7-(pheny1acetamido)cephalosporinmustard has been reported elsewhere (15). Nitrocefin and PADAC were purchased from Beckton Dickinson Microbiology Systems, Beckton Dickinson and Company, Cockeysville, MD, and Calbiochem, La Jolla, CA, respectively. The mAbs L6 and P1.17 are both of the IgGza isotype. L6 was purified by protein A chromatography from the supernatant of an L6-producinghybridoma cell line (16). P1.17 was purified from mouse ascites on a protein A column. L6 binds to antigens on a variety of human carcinomas, including the lung adenocarcinoma cell line H2981 (16). P1.17 shows no detectable binding to these cells. The Fab’n fragments of the mAbs were obtained by digestion with pepsinogen as described previously (15), and purified by affinity 0 1994 American Chemical Society
j3-Lactamase Conjugates
Bioconjugate Chem., Vol. 5, No. 3, 1994 263
Enzyme Activity. Specific Activity. The enzymatic chromatography on a protein A column and size-exclusion activities of the purified enzyme and the mAb conjugates chromatography on Sephacryl S-300. L6-Fab' was obwere determined using either nitrocefin (12,20) or PADAC tained by reduction of the interchain disulfides of L6(18)as substrates. (1)Nitrocefin: PL-containing solutions Fab'2, followedby addition of 5,5'-dithiobis(2-nitrobenzoic (5.0 ng of enzyme/mL) in PBS containing 12.5 pg/mL acid) (17). bovine serum albumin were incubated with nitrocefin (50 Conjugate Preparation. The preparation of mAbpM) in a cuvette at ambient temperature. The rate of Fab'z-OL conjugates has been described previously (12). hydrolysis was estimated from the increase in absorbance Briefly, maleimide-substituted mAb (from reaction with at 490 nm (Ae = 19 500 M-I cm-l) that occurs when the N-succinimidyl 4-(maleimidomethyl)cyclohexane-l-carP-lactam ring is opened. (2) PADAC: PL-containing boxylate, SMCC, Pierce Chemical Co., Rockford, IL) was solutions (25-50 ng of enzyme/mL) in PBS/bovine serum allowed to react with PL that had been modified with albumin (as above) were incubated with PADAC (25 pM) 2-iminothiolane to contain free sulfhydryl groups. The in a cuvette at ambient temperature. Hydrolysis rates conjugate was purified using a two-step procedure that were estimated from the increase of absorbance at 450 nm involved size-exclusion and boronic acid affinity chro(AE = 14 700 M-'cm-l). The specific activity of each matographies. conjugate was compared to the specific activity of the BL MAb-Fab'-PL and (mAb-Fab')n+L (dimeric mAb-PL) sample used for the conjugate preparation. The data conjugates were prepared using a method based on represent the mean specific activity ratio from at least previously published procedures (17, 18). SMCC was three preparations of each conjugate, except in the case dissolved in dimethylformamide at 20.0 mM and added of P1.17-Fab'2-PL, which was only prepared once. to PL solutions at 4-5 mg/mL in phosphate-buffered saline Relative Activity. Fractions from boronic acid and size(PBS)to give a final SMCC concentration of 1.0 mM. The exclusion chromatography were tested for relative enzymixture was incubated for 30 min at 30 "C, followed by matic activities. An aliquot of each fraction was diluted removal of unreacted SMCC and reaction byproducts by 2000-8000-fold in a 0.1 mg/mL bovine serum albumin gel filtration chromatography at 4 "C through Sephadex solution in PBS and incubated at ambient temperature G-25M (PD-10, Pharmacia, Piscataway, NJ), equilibrated with nitrocefin (final concentration of 50 pM) in a 96-well in N2-purged 40 mM sodium phosphate, pH 7.4, containing microtiter plate. The absorbance at 490 nm (using 630 0.6 M NaC1. nm as a reference wavelength) was read 5-10 min after the DL-Dithiothreitol (DTT, Sigma Chemical Co.) was initiation of the reaction on an EL 312 Bio-Kinetics Reader dissolved in PBS at 10.0 mM and added to solutions of (Bio-Tek Instruments, Inc.). Reaction conditions were mAb-Fab'z at 5-10 mg/mL in PBS, containing 15 mM chosen such that the value of the fraction with the highest sodium borate, pH 8.0, to give a final DTT concentration absorbance fell within a range of 0.10-0.25. of 0.5 mM. The solution was incubated for 60 min at 30 Cell Binding. L6-PL conjugates were tested for their "C,followed by purification of the reduced mAb by gel abilities to bind to H2981 cells relative to L6-Fab'z and filtration at 4 "C on a PD-10 column which was equilibrated L6-Fab' in a competition assay (15). H2981 cells (0.5 X inN2-purged 40mM sodium phosphate (pH 7.4) containing 105) in Iscove's modified Dulbecco's medium with 10% 0.6 M NaC1. The number of free sulfhydryl groups fetal bovine serum were incubated with the test sample obtained using this procedure (determined using 5,5'and fluorescein isothiocyanate labeled whole L6 (L6dithiobis(2-nitrobenzoicacid)) (19) was 3.8-4.4. FITC) such that the combined L6-Fab' concentration (test The maleimide-substituted enzyme was added to the sample + L6-FITC) was 800 nM. The ratio of test sample reduced mAb at a 1:2-3 molar ratio of pL:mAb-Fab'. The to L6-FITC ranged from 0 to 1. After the sample was mixture was incubated at ambient temperature for 60 min, incubated for 30 min on ice, the cells were washed and followed by the addition of trans-4,5-dihydroxy-1,2- analyzed on a fluorescence activated cell sorter. The mean channel number of fluorescence was converted into linear dithiane (Sigma Chemical Co.) dissolved at 60 mM in HzO fluorescence equivalence (LFE) and percent of binding (final concentration of 5 mM). Incubation was continued was calculated using the following formula: % binding = for 60 min at ambient temperature and then 18 h at 4 "C. 100 - 100[(LFEicm LG-FITC - LFE,,,id/(LFEicm LG-FITC Subsequent manipulations were all carried out at 4 "C. - LFEm mAb)l* The conjugate was subjected to purification by affinity chromatography on a boronic acid affinity column of the In Vitro Cytotoxicity. H2981 cells in Iscove's modified hydrophilic type (L-type, 14), equilibrated in 20 mM Dulbecco's medium with 10% fetal bovine serum were triethanolamine hydrochloride at pH 7.0, containing 0.5 plated out at 8000 cells/wellinto 96-wellplates and allowed M NaC1, followed by washing of the column with the above to adhere overnight at 37 "C. The cells were incubated buffer until A2m = 0. The bound material was eluted off with varying concentrations of PL or PL conjugates for with 0.5 M sodium borate at pH 7.0, containing 0.5 M 30-45 min at 4 "C. After unbound material was washed off, the cells were treated with 10pM 7-(pheny1acetomido)NaC1. Fractions were analyzed for relative enzymatic cephalosporin mustard for 60 min at 37 "C. The cells activity, and those that eluted with sodium borate and were then washed, and incubation was continued for 18 had high enzymatic activities were pooled, concentrated h at 37 "C. This was followed by a 6-h pulse with 13H]by ultrafiltration, and applied to a Sephacryl S-300 thymidine (1pCi/well). The cells were washed with PBS, (Pharmacia) size-exclusioncolumn (equilibrated in PBS). detached by treatment with trypsin/EDTA, harvested onto The fractions were analyzed by sodium dodecyl sulfate filtermats (LKB WALLAC 1295-001Cell Harvester), and polyacrylamide gel electrophoresis (SDS-PAGE) and by counted on an LKB WALLAC 1205 liquid scintillation their relative enzymatic activities. Two pools, consisting counter. The incorporation of PHIthymidine was calof the purified mAb-Fab'-PL and dimeric mAb-PL culated as the percentage of treated cells relative to conjugates, were concentrated by ultrafiltration, filtered untreated controls. through 0.2-pm filters, and stored at -70 "C. The concentrations of the preparations were determined specIn Vivo Biodistribution. Radiolabeling of Conjugates. LG-Fab'2-PL and dimeric LG-PL were labeled with trophotometrically at 280 nm using an E'% of 14.0 and lz51using Iodogen (Pierce Chem. Co). Both conjugates 15.3 (12) for the mAbs and the enzyme, respectively.
264
Svensson et al.
Bioconjugate Chem., Vol. 5, No. 3, 1994
were separated from low molecular weight material by gel filtration on PD-10 columns equilibrated in PBS. Previous studies with such columns and elution conditions yielded 90% protein recovery, which was therefore used as the assumed recovery of radiolabeled conjugates. The specific activities in terms of radioactivity were 1.6 pCi/pg for L6Fab’2-PL and 0.9 pCi/pg for dimeric L6-PL. Immunoreactivity. The radiolabeled conjugates were tested for their abilities to bind to H2981 cells in a competition assay. Cells were grown to confluence in a 96-well plate in Iscove’s modified Dulbecco’s medium supplemented with 10% fetal bovine serum. Each conjugate was added to the cells at a constant concentration of conjugate, while the ratio of labeled to unlabeled material was varied. The cells were incubated for 30 min at 4 “C,washed with PBS, detached with trypsin/EDTA, and counted with a y-counter. In Vivo Distribution. In vivo studies were performed in 6-10 week old female athymic nu/nu mice (HarlanSprague-Dawley, Indianapolis, IN). The mice were implanted subcutaneously with in vivo passaged H2981 tumors. When tumor growth was established and the tumors had reached a size of 100-175 mg, groups of three to four mice were injected intravenously with radiolabeled conjugates at a dose of 0.5 mg (265-310 pCi) of L6 component/kg(10 mL/kg), using 1.0 mg/mL bovine serum albumin in PBS as the diluent. At 96 h post conjugate treatment the mice were bled and sacrificed. Tissues were collected and weighed immediatelyfollowing excision and then counted with a y-counter. Trichloroacetic Acid Precipitation. The in vivo stabilities of the radiolabeled conjugates in tumors and blood were monitored by trichloroacetic acid precipitation. Tumors were homogenized with a pestle, and protein was precipitated with 500 pL of a 10% aqueous (wkvol)solution of trichloroacetic acid for 30 min at ambient temperature. Plasma proteins were similarly precipitated. The samples were centrifuged, and the supernatant from each of these was removed. The pellets were resuspended in 500 pL of a 5 % solution of trichloroaceticacid. After centrifugation, the supernatants were pooled, and the pellets and supernatants were counted with a y-counter. RESULTS AND DISCUSSION
Preparation and Structures of mAb-flL Conjugates. LG-Fab’2-PL and P1.17-Fab’2-DL conjugates were prepared by linking the two proteins via a thioether bond as previously described (12). Briefly, this involved combining thiol-containing PL with mAb-Fab’2 that had been modified with maleimide groups. The conjugate thus formed was purified by size-exclusion and affinity chromatograpy, using Sephacryl S-300 and immobilized boronic acid, respectively. This conjugation method yielded 1:l adducts (>95% purity) of mAb-Fab’2:PL with an apparent molecular weight of 150 kDa. The overall yield was 13-15 % ,based on the amount of thiol-containing PL used for the preparation. Analysis of LG-Fab’2-PL for binding to L6 antigens on the H2981 human lung adenocarcinoma cell line revealed a reproducible and significant loss in binding activity (12). This was in contrast to similarly prepared conjugates between L6-Fab’2 and Bacillus cereus 6-lactamase (11)(15),in which the binding activity of L6 was preserved (15). Consequently,studies were undertaken to prepare conjugates of PL having preserved binding activities. Conjugates of L6 and PL were prepared by reacting reduced L6-Fab’z with maleimide-substituted enzyme. SDS-PAGE analysis indicated that complete reduction
0.30
A
4’0/ 3.0
0.20
?
I
0.10
-b )r
L !O
25
c
S
O
.c _ > .c _
2 %
0
I
1
10 20 30 40 50 60 70 80
Fraction Number
Figure 1. Purification of PL conjugates. (A) The unpurified conjugation mixture was loaded on a boronic acid affinity column, and bound material was eluted off with a sodium borate buffer (indicated by an arrow). Shaded area of protein peak shows the fractions that were pooled. (B) The conjugates were further purified by size-exclusion chromatography on an Sephacryl S-300 column. Two pools were obtained (shaded).
of interchain disulfides took place when DTT was used as the reducing agent (data not shown). The reduced mAb was reacted with maleimide-substitutedenzyme at an L6Fab’:PL molar ratio of 2-3:l and then subjected to purification using a modification of a previously described two-step procedure (12). In the first step, the reaction mixture was applied to a boronic acid affinity column and eluted with sodium borate (Figure 1A). Analysis of the fractions for enzymaticactivity indicated that most of the activity was present in the fractions that bound to the column and eluted with sodium borate. Fractions that had high enzymatic activity (fractions number 17-21, Figure 1A) were pooled and concentrated. SDS-PAGE analysis (data not shown) indicated the presence of material with distinct apparent molecular weights of 40, 110, and 150 kDa, respectively. Adducts of higher molecular weights were also present. The concentrated pool was applied to a Sephacryl S-300 size-exclusion column (Figure 1B). Three peaks were obtained in the elution of the conjugate mixture, all of which had P-lactamase activity. SDS-PAGE analysis showed that the first two peaks (pools I and 11, Figure 1B) contained fractions with apparent molecular weights of 150 and 110 kDa, respectively. The fractions that eluted immediately prior to pool I contained higher molecular weight adducts. The peak that eluted last had a molecular weight of 40 kDa, which corresponds to unreacted PL. The pooled conjugate-containingfractions were analyzed by SDS-PAGE under nonreducing conditions, which showed that the conjugates were a t least 90 % pure (Figure 2A). LG-Fab’2-PL (Figure 2A, lane 5) had an apparent molecular weight of 150kDa, correspondingto the expected molecular weight for a mAb-Fab’2 fragment attached to a single PL molecule. Pool I had the same apparent molecular weight (Figure 2A, lane 4). Pool I1 (Figure 2A, lane 3) had a molecular weight of approximately 110 kDa,
Bioconjugate Chem., Vol. 5, No. 3, 1994
fl-hctamase Conjugates
A
100“
A L6-Fab‘, 0
80-
94674330-
2014.4-
0
5
60-
s
40
-
20
-
c .M
L6-Fab’
A 0
bo
c
265
0
20
40
60
80
100
% Test Sample
B
1 2 3 4 5
94674 3-
302014.4Figure 2. SDS-PAGE analyses (4-20 % ) of PL conjugates under nonreducing (A) and reducing (B) conditions: lane 1, whole L6; lane 2, L6-Fab’z; lane 3, LG-Fab’-PL; lane 4,dimeric L6-PL; lane 5, LG-Fab’z-PL; lane 6, PL. Molecular weight markers (in kDa) are indicated on the left.
which is close to the theoretical molecular weight (95 kDa) for a monomeric Fab’-PL adduct. Further structural information of the conjugates was derived by subjecting the conjugates to SDS-PAGE under reducing conditions. It was found that the heavy (30 kDa) and light (25 kDa) chains of LG-Fab’a could be distinguished due to their different mobilities on the SDS gel (Figure 2B, lane 2). The two chains stained with approximately the same intensity using Coomassie blue. Under the reducing conditions,both heavy and light chains were released from the LG-Fab’2-PL conjugate (Figure 2B, lane 5). Most, if not all, of the PL was conjugated to the heavy chain, since there is approximatelytwice as much light chain released as heavy chain. The identity of the light chain released from the conjugate rests on the fact that it has the same apparent molecular weight as the light chains from either whole L6 or LG-Fab’a (Figure 2B, lanes 1and 2). Pools I and I1 (Figure 1B) did not appear to contain disulfide-linked heavy chains that could be released upon reduction, indicating that the PL is attached only to the heavy chains of these conjugates. The structure of the conjugate obtained in pool I (Figure 1B) therefore is consistent with the covalent attachment of two individual L6-Fab’ fragments to maleimideson PL. The attachment is most likely through a heavy chain cysteine in the hinge region. Further evidence to support the proposed dimeric structure of this conjugate (chemicallyrepresented as (L6Fab’)2-@L and designated as dimeric L6-PL) is based on the fact that it has the same apparent molecular weight as LG-Fab’2-PL under nonreducing conditions on SDSPAGE (Figure 2A, lanes 4 and 5). Finally, similar analyses of pool I1 (Figure 1B) under both nonreducing (Figure 2A,
Figure 3. Cell binding assay. H2981 cells were exposed to conjugated or unconjugated L6 in various mixtures with L6FITC, while the L6-Fab’ concentration was kept constant a t 800 nM. Fluorescence intensity was determined by fluorescenceactivated cell sorter analysis. Dimeric L6-DL is represented as (L6-Fab’) Z-PL.
lane 3) and reducing (Figure 2B, lane 3) conditons indicate that the conjugate is LG-Fab’-PL, in which the L6 heavy chain is linked to maleimides on the enzyme. Conjugates were also made with the control mAb P1.17 using the same methods as described above for L6. The purities and mobilities by SDS-PAGE were similar to that of the L6PL conjugates. On the basis of the amount of maleimidecontaining PL used in the conjugate preparation, the overall yields were 10-25 % for the mAb-Fab’-PL conjugates and 8-13 % for the (mAb-Fab’)a-PL conjugates. Conjugate Activities. A competition assay was used to assess the effects that the different conjugation methods had on the binding characteristics of the PL conjugates to H2981 human lung adenocarcinoma cell surface antigens (Figure 3). Cells were incubated with conjugates or unconjugated mAbs in various mixtures with whole L6FITC. It was found that dimeric LG-PL was only slightly less effective than LG-Fab’a in competing for binding to cell surface antigens. The monovalent conjugate, L6-Fab’PL, showed a substantially lower level of binding in this assay, but the level was identical to L6-Fab’. Since there was no differences in binding between the L6-Fab’ and the LG-Fab’-PL conjugate, it is likely that the enzyme is conjugated to a site on the L6 such that it does not interfere with the binding to cell surface antigens. In contrast to these two examples, LG-Fab’2-PL exhibited significantly impaired binding to H2981 cells compared to unmodified LG-Fab’2. This is consistent with previous findings (12), and is probably due to either the site a t which E. cloacae PL binds to LG-Fab’2 and/or to interactions between the antigen binding site on L6 with the enzyme. P1.17 has been shown previously to give no detectable binding on H2981 cells. The PL conjugates were also tested for P-lactamase enzyme activites using either nitrocefin or PADAC as substrates. With nitrocefin, the specific activity of PL was 405 f 43 pmol/min/mg, and with PADAC, the activity was 23.1 f 1.9 pmol/min/mg. The results showed that the enzymatic activities of all the conjugates were similar to that of unconjugated PL (Table 1). Thus, it appears that the methods for conjugation and purification had little effect on enzyme activity. This is consistent with previous observations for a variety of P-lactamase conjugates (12, 15, 18). In Vitro Prodrug Activation. To establish the effects that the conjugation methods had on in vitro prodrug activation, cytotoxicity studies were performed with the
Svensson et al.
Bioconjugate Chem., Vol. 5, No. 3, 1994
266
Table 1. Enzymatic Activities of @L Conjugates
~
re1 activity (% ) (no. of preparations) 108 f 12 (5) 107 f 12 (4) 110 f 17 (3) 119 f 24 (4)
conjugate LG-Fab’-BL (L6-Fab’)2-BLb LG-Fab’2-BL P1.17-Fab’-OL (P1.17-Fab’)2-BLb P1.17-Fab’2-@L
M2FA
\
124 f 33 (4) lloc (1)
751%
”1
B
U
A -
0LG-Fab’,-PL El (LG-Fab’),-PL
=The activity of each conjugate is expressed relative to the particular PL sample that was used for the conjugate preparation. The numbers in parentheses indicate the number of independently prepared conjugates that were tested. b These are referred to in the text as dimeric mAb-BL conjugates. No standard deviation was obtained, since this conjugate was only prepared once.
’“r
T
0
h W
p
O LG-Fab’,-PL 0 (LG-Fab’),-PL
3
.I-
VI
0
25
Tissue
Figure 5. In vivo biodistribution. LG-Fab’z-PL and (L6-Fab’)~ and administered to PL (dimeric L6-PL) were labeled with 1251 nude mice with subcutaneous H2981 xenografts. At 96 h post treatment, the mice were bled and sacrificed. The data were calculated as a percentage of the injected dose per gram tissue (A) and as tissue to blood concentration ratios (B).
0.01
0.1
1
10
100
Concentration (nM) Figure 4. In vitro prodrug activation. H2981 cells (L6 positive, P1.17 negative) were exposed to varying concentrations of PL conjugates or unconjugated PL, washed, and treated with 7-(phenylacetamido)cephalosporin mustard (10 pM). Cell viability was measured by [3H] thymidine incorporation and is expressed as a percentage of untreated control cells. The dimeric conjugates are represented as (mAb-Fab’h-PL. conjugates in combination with 7-(pheny1acetamido)cephalosporin mustard, a prodrug that was previously shown to be a substrate for PL (12). Upon enzymatic hydrolysis of the P-lactam ring of the prodrug, the cytotoxic agent phenylenediaminemustard is released (15). H2981 cells were incubated with various concentrations of PLcontaining conjugates or unconjugated enzyme. The cells were washed and then incubated with a nontoxic concentration (10 pM) of 7-(phenylacetamido)cephalosporin mustard. The highest degree of immunospecific activation was obtained with the dimeric mAb-PL conjugates, in which there was an 80-fold difference between the L6 and P1.17 conjugates (Figure 4). In this respect, the dimeric LG-PL conjugate was superior to LG-Fab’-PL and L6Fab’2-PL, which had smaller degrees (20-fold) of immunologically specific activation. The finding that both L6- and P1.17-Fab’2-PL conjugates showed a greater degree of prodrug activation compared to their respective mAb-Fab’-PL counterparts (Figure4) may be attributable to a slightly higher degree of non-specificbinding to H2981 cells.
Since the conjugates had similar enzymatic activities (Table l), the level of prodrug activation was expected to be dependent on the amount of conjugate that bound to cells. Therefore, higher concentrations of conjugateswould be needed to activate the prodrug, if the conjugate is impaired in its ability to bind to cell surface antigens. Dimeric LG-PL effected prodrug activation a t the lowest concentration compared to the other conjugates (Figure 4). The concentration of dimeric L6-PL needed to achieve an IC50 value with the prodrug (0.1 nM) was 4 and 10 times lower than LG-Fab’2-PL and LG-Fab’-PL, respectively (Figure 4A). In Vivo Biodistribution. The in vivo distribution of LG-Fab’z-PL and dimeric L6-PL to H2981 tumors was investigated in mice with subcutaneous H2981 tumor xenografts. The conjugates were radiolabeled (1251) without any loss of immunoreactivity, as measured by a competition cell binding assay. Autoradiograms from SDS-PAGE analysis of the radiolabeled materials suggested that the conjugates were homogeneously labeled and stable to the radiolabeling procedure (data not shown). At 96 h post conjugate administration, tissues were removed and counted for radioactivity. Additional experiments confirmed that >90 % of the radioactivity measured was precipitable with trichloroacetic acid and was therefore associated with protein. Figure 5A shows the concentration of the conjugates in various tissues 96 h post conjugate administration. This particular timepoint was chosen since it coincides with the schedule used in a previously reported in vivo therapeutic efficacy study involving LG-Fab’2-PL and was determined to be an appropriate time for prodrug administration in mice receiving this conjugate (12). Analysis of the tissue to blood concentration ratios (Figure 5B) indicated that dimeric LG-PL had a 50 % higher tumor to blood ratio than LG-Fab’2-PL (Pvalue of 0.03). The tumor levels of dimeric LG-PL and LG-Fab’2-PL were not
P-Lactamase Conjugates
statistically different ( P value of 0.17) even though the blood concentration of dimeric LG-PL was half that of LG-Fab’z-PL (P value of 0.01). In all other tissues examined, the amount of radioactivity associated with dimeric LG-PL was lower than that of LG-Fab’z-PL (Figure 5B). Thus, a modest, but significant, improvement in conjugate tumor to blood ratio is achieved with dimeric LG-PL compared to L6-Fab’2-PL, and this is accompanied with reduced conjugate levels in nontarget tissues. On the basis of this, we are now utilizing dimeric mAb-PL conjugates in combination with anticancer prodrugs for in vivo therapy studies. CONCLUSION
LG-Fab’-PL and dimeric LG-PL conjugates were conveniently prepared by linking one and two individual L6Fab’ units to maleimide-substituted PL, respectively. Analyses of these conjugates by SDS-PAGE indicated that linkage occurred through heavy-chain thiols on L6 that are most likely located in the hinge region of the mAb and are removed from the antigen binding site. A greater degree of control was therefore used in the preparation of these conjugates compared to L6-Fab’2-PL, which was prepared by random modification of L6-Fab’z with maleimide groups. As a result, the dimeric LG-PL conjugate displayed better binding characteristics, increased potency in prodrug activation, and improved in vivo localization characteristics compared to L6-Fab’zPL. Although the absolute amount of intratumoral conjugate was not increased using dimeric L6-PL relative to L6-Fab’2-PL the amount in the blood and in nontarget tissues was reduced. This constitutes an improvement for the targeting strategy described here, in which high tumor to nontumor conjugate ratios are required to minimize toxicities due to adventitious drug release. ACKNOWLEDGMENT
We wish to thank Vivekananda Vrudhula for providing the prodrug used in this study, David Kerr for enzyme purification, Nathan Siemers for his comments on the manuscript, and Karl Erik and Ingegerd Hellstrom for their continued support of the project. LITERATURE CITED (1) LoBuglio, A. F., and Saleh, M. N. (1992) Advances in monoclonal antibody therapy of cancer. Am. J.Med. Sci. 304,
214-224. (2) Pack,P., Kajau, M., Schroeckh,V.,Knupfer,U., Wenderoth, R., Riesenberg, D., and Pluckthun, A. (1993)Improved bivalent miniantibodies, with identical avidity as whole antibodies, produced by high cell density fermentation of Escherichia coli, BiolTechnology 11, 1271-1277. (3) Schott, M. E., Frazier, K. A., Pollock, D. K., and Verbanac, K. M. (1993) Preparation, characterization, and in vivo biodistribution properties of synthetically cross-linked mul-
Bioconjugate Chem., Vol. 5,
No. 3, 1994 287
tivalent antitumor antibody fragments. Bioconjugate Chem. 4, 153-165. (4) Holton, 0. D., 111, Black, C. D. V., Parker, R. J., Covell, D. G., Barbet, J., Sieber, S.M., Talley, M. J., and Weinstein, J. N. (1987) Biodistribution of monoclonal IgG1, F(ab’)z and Fab’ in mice after intravenous injection. J. Immunol. 139, 30413049. ( 5 ) Vitetta, E. S., Thorpe, P. E., and Uhr, J. W. (1993) Immunotoxins: magic bullets or misguided missiles? Immunol. Today 14, 252-258. (6) Bamias, A., and Epenetos, A. A. (1992) Two-step strategies for the diagnosis and treatment of cancer with bioconjugates. Antibody Imm. Radiopharm. 5 , 385-395. (7) Goldenberg, D. M. (1993) Monoclonal antibodies in cancer detection and therapy. Am. J. Med. 94, 297-312. (8) Senter, P. D., Wallace, P. M., Svensson, H. P., Vrudhula, V. M., Kerr, D. E., Hellstrom, I., and Hellstrom, K. E. (1993) Generation of cytotoxic agents by targeted enzymes. Bioconjugate Chem. 4, 3-9. (9) Sharma, S. K., Bagshawe, K. D., Springer, C. J., Burke, P. J., Rogers, G. T., Boden, J.A., Antoniw, P., Melton, R. G., and Sherwood, R. F. (1991) Antibody directed enzyme prodrug therapy (Adept): A three phase system. Dis. Markers 9,225231. (10) Wallace, P. M., MacMaster, J. F., Smith, V. F., Kerr, D. E., Senter, P. D., and Cosand, W. L. (1994) Intratumoralgeneration of 5-fluorouracil mediated by an antibody-cytosine deaminase conjugate in combination with 5-fluorocytosine. Cancer Res., in press. (11) Meyer, D. L., Jungheim, L. N., Law, K. L., Mikolajczyk, S. D., Shepherd, T. A., Mackensen, D. G., Briggs, S. L., and Starling, J. J. (1993) Site-specific prodrug activation by antibody-P-lactamase conjugates: Regression and long-term growth inhibition of human colon carcinoma xenograft models. Cancer Res. 53, 3956-3963. (12) Vrudhula, V. M., Svensson, H. P., Kennedy, K. A., Senter, P. D., and Wallace, P. M. (1993) Antitumor activities of a cephalosporin prodrug in combination with monoclonal antibody-@-lactamaseconjugates. Bioconjugate Chem. 4, 334340. (13) Alexander, R. P., Beeley, N. R. A., O’Driscoll, M., O’Neill, F. P., Millican, T. A., Pratt, A. J., and Willenbrock, F. W. (1991) Cephalosporin nitrogen mustard carbamate prodrugs for “ADEPT”. Tetrahedron Lett. 32, 3269-3272. (14) Cartwright, S. J., and Waley, S. G. (1984) Purification of p-lactamases by affinity chromatography on phenylboronic acid-agarose. Biochem. J. 221, 505-512. (15) Svensson, H. P., Kadow, J. F., Vrudhula, V. M., Wallace, P. M., and Senter, P. D. (1992) Monoclonal antibody-plactamase conjugates for the activation of a cephalosporin mustard prodrug. Bioconjugate Chem. 3, 176-181. (16) Hellstrom, I., Horn, D., Linsley, P. S., Brown, J. P., Brankovan, V., and Hellstrom, K. E. (1986) Monoclonal antibodies raised against human lung carcinoma. Cancer Res. 46, 3917-3923. (17) Goshorn, S. C., Svensson, H. P., Kerr, D. E., Somerville, J. E., Senter, P. D., and Fell, H. P. (1993) Genetic construction, expression and characterization of a single chain anti-carcinoma antibody fused to p-lactamase. Cancer Res. 53, 2123-2127. (18) Meyer, D. L., Jungheim,L. N.,Mikolajczyk, S. D., Shepherd, T. A,, Starling, J. J., and Ahlem, C. (1992) Preparation and characterization of 6-lactamase-Fab‘ conjugates for the site specific activation of oncolytic agents. Bioconjugate Chem. 3, 42-48. (19) Riddles, P. W., Blakeley, R. L., and Zerner, B. (1979) Ellman’s reagent: 5,5’-Dithiobis(2-nitrobenzoicacid)-a reexamination. Anal. Biochem. 94, 75-81. (20) Madgwick, P. J., and Waley, S. G. (1987) 6-Lactamase I from Bacillus cereus. Biochem. J. 248, 657-662.
