Dextran Modification of a Fab'−β-Lactamase Conjugate Modulated by

Bioconjugate Chem. 1996, 7, 150−158

150

Dextran Modification of a Fab′-β-Lactamase Conjugate Modulated by Variable Pretreatment of Fab′ with Amine-Blocking Reagents Stephen D. Mikolajczyk,*,† Damon L. Meyer,†,‡ Roberto Fagnani,† Michael S. Hagan,† Kevin L. Law,§ and James J. Starling§ Hybritech Inc., P.O. Box 269006, San Diego, California 92196-9006, and Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, Indiana 46285. Received May 1, 1995X

The physical and pharmacological properties of proteins can be altered by chemical modification with polymers. Preliminary studies showed that attachment of oxidized dextran to the bacterial protein, β-lactamase (βL) effectively reduced in vivo immunogenicity in mice with no loss of enzymatic activity. This report describes a general method for differentially dextran modifying the Fab′ component of a Fab′-β-lactamase conjugate by the use of amine-blocking reagents. Methyl acetimidate (MeAcm) and the N-succinimidyl derivative of (methylsulfonyl)ethyl carbonate (NHS-Msc), reagents which can reversibly block primary amines, were used in model studies to modulate the level of available reactive amines on the F(ab′)2 fragments of both the anti-carcinoembryonic antigen antibody, ZCE025, and the antitumor-associated glycoprotein-72 antibody, CC49. MeAcm had little or no effect on immunoreactivity and was maximally effective in modulating dextran attachment, while NHS-Msc was much less effective. A comparison of NHS-Msc and MeAcm is described. Treatment of F(ab′)2 with 5-300 mM MeAcm prior to dextran treatment showed a proportional decline in the level of dextran attachment as well as intramolecular cross-linking of the protein by the dextran polymers (6 kDa or 33-mer). A conjugate of βL coupled to MeAcm-treated ZCE025 Fab′ [reduced F(ab′)2] was constructed under standard conditions using sulfosuccinimidyl N-[(4-carboxycyclohexyl)methyl]maleimide. After dextran modification, this conjugate maintained good immunoreactivity and enzymatic activity. Biodistribution studies in tumor-bearing nude mice of dextranated and nondextranated conjugate showed comparable overall distribution profiles except that the clearance of the dextranated conjugate from both blood and tumor was delayed about 48-72 h.

INTRODUCTION

In previous work, bioconjugates of the bacterial protein β-lactamase coupled to antibody Fab′ fragments of ZCE025 and CC49 were shown to be efficacious in the tumorspecific delivery of oncolytic drugs (1). While effective in the mouse model studies, extension of this work in human clinical trials required consideration of the immunogenic response to such conjugates, especially since multiple injection protocols would likely be necessary to achieve maximum efficacy (1-3). The immune response to protein-based injectables, while somewhat variable, is a significant problem for most in vivo applications (2-4). In the case of antibody-based bioconjugates, multiple strategies have been devised to minimize immunogenic response, including chimeric or humanized expression of antibodies (5-7) or the expression of small Fv fragments (8, 9). Because such strategies are ongoing in our laboratory as well, the primary focus of the current study has been directed towards βL,1 a bacterial protein for which no good recombinant expression alternatives are currently available for amelioration of the immunogenic response. Chemical methods of reducing the immunogenicity of proteins have generally focused on the modification of * Author to whom correspondence should be addressed: Hybritech Inc., P.O. Box 269006, San Diego, CA 92196-9006. Phone: (619) 535-8754. Fax: (619) 457-5308. † Hybritech Inc. ‡ Current address: NeoRx Corporation, 410 W. Harrison, Seattle, Washington 98119-4007. § Eli Lilly and Company. X Abstract published in Advance ACS Abstracts, January 1, 1996.

1043-1802/96/2907-0150$12.00/0

the protein surface with polymers of poly(ethylene glycol) or oxidized dextran. While both techniques involve covalent attachment to the reactive free amines on proteins, the fundamental difference between these techniques is that PEG is generally only attached through a single “activated” terminus of methyl-PEG (10-12) while an oxidized form of the dextran polymer is used which permits attachment to the protein at multiple sites throughout the polymeric dextran chain (13, 14). Dextran polymers have been shown in other work to be immunogenic (15), though even this is markedly reduced at polymer molecular masses of less than 50 kDa (16). The oxidized form of the 6 kDa dextran used in the current procedure no longer contained the intact ring structure of native dextran and has been shown previously to be nonimmunogenic (13). The primary goal of the current project was to study the effect of oxidized dextran attachment on the Fab′βL conjugates we have employed in previous studies (1, 17, 18). Previous work showed that dextran modification had no effect on the immunoreactivity of the T101 antibody (13) but was inhibitory to the anti-CEA antibody intended for use in this work (19). βL activity was 1 Abbreviations: PEG, poly(ethylene glycol); MeAcm, methyl acetimidate; NHS-Msc, N-succinimidyl (methylsulfonyl)ethyl carbonate; Mr, relative molecular weight; βL, β-lactamase; Fab′βL, covalent conjugate of Fab′ with βL; anti-CEA, anticarcinoembryonic antigen; anti-TAG-72, antitumor-associated glycoprotein-72; NEM, N-ethylmaleimide; DMSO, dimethyl sulfoxide; HIC, hydrophobic interaction chromatography; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; RT, room temperature; sulfo-SMCC, sulfosuccinimidyl N-[(4-carboxycyclohexyl)methyl]maleimide; TNBS, trinitrobenzenesulfonic acid; SEC, size exclusion chromatography; ELISA, enzymelinked immunoabsorbent assay.

© 1996 American Chemical Society

Dextran Modification of Protein Conjugates

unaffected by dextran treatment. Therefore, the objective of the first phase of this investigation was 2-fold: (1) to examine the ability of amine-blocking reagents to modulate the level of dextran attachment on the Fab’s to determine if conditions could be found which would preserve the immunoreactivity and (2) to use the existing murine Fab’s as model proteins to establish the parameters necessary to create a conjugate with fully dextranated βL but with modulated levels of dextran on the Fab′. Thus, the current study involves the development of general methods for modulation of the dextran attachment in a protein-protein conjugate where the individual protein components may be differentially affected by the extent of polymeric attachment. Since both the proteinprotein coupling procedures (20) as well as the dextran attachment (13) utilized the same lysine -amino groups, it was not feasible simply to dextranate the βL prior to coupling to the Fab′. Without the availability of specialized, protected linkers, it was necessary to form the Fab′βL conjugate prior to dextran treatment. The reversible amine-blocking reagents methyl acetimidate (MeAcm) (21-23) and the amino-reactive Nsuccinimidyl derivative of (methylsulfonyl)ethyl carbonate (NHS-Msc) (24) have been widely used for blocking amines on polypeptide chains. NHS-Msc and MeAcm were tested both for their direct effects on Fab′ immunoreactivity and for their ability to selectively block key lysine residues and thus modulate the subsequent attachment of oxidized dextran. The results of this work demonstrate the feasibility of construction of a proteinprotein conjugate with modulated availability of amino groups for subsequent modifications. EXPERIMENTAL PROCEDURES

Reaction of MeAcm and NHS-Msc with F(ab′)2. F(ab′)2 fragments of ZCE025 antibody were prepared by incubation of intact antibody with 3% pepsin (w/w) at pH 3.5 in 100 mM citrate buffer at 37 °C for 90 min (25, 26). CC49 fragments were obtained by incubation in the same buffer at pH 3.9 for 4 h. F(ab′)2 fragments in both cases were purified to homogeneity by cation exchange chromatography using Pharmacia S Sepharose Fast Flow resin. Buffer A was 170 mM sodium acetate (pH 4.5) and buffer B the same with 0.5 M NaCl. With a linear gradient, F(ab′)2 fragments elute at about 50% buffer B. Concentrations of F(ab′)2 used in these experiments were 0.1% from 7-10 mg/mL using an 280 of 1.8. A 2 M solution of MeAcm was freshly prepared by dissolving solid methyl acetimidate hydrochloride (Pierce Chemical Co., Rockford, IL) in 100 mM borate buffer (pH 10), on ice, and adjusting the pH to about pH 9 with 6 N NaOH (as determined with pH paper). All reactions with MeAcm and protein were conducted at 4 °C (on ice). An aliquot of this MeAcm solution was then immediately added to F(ab′)2 which had been dialyzed versus 100 mM borate buffer (pH 10) (pH 9 in some experiments, see Results) to make a final concentration from 5 to 300 mM MeAcm. In a typical experiment, 77 mg of MeAcm HCL was added to 225 mL of borate buffer (pH 10) plus 57 mL of 6 N NaOH to yield about 350 mL of 2 M MeAcm pH (∼9). Dilutions of the 2 M MeAcm were sufficiently large so as not to affect the pH of the final F(ab′)2 solution. The samples were allowed to react for 2 h on ice and then dialyzed versus 160 mM sodium phosphate buffer at 4 °C to remove small molecules. Determination of reactive amines after treatment with MeAcm was determined with TNBS as described by Habeeb (27). The NHS-Msc was dissolved in DMSO at a concentration of 1 M. Typical F(ab′)2 samples were dialyzed versus

Bioconjugate Chem., Vol. 7, No. 1, 1996 151

100 mM borate buffer (pH 8) (pH 7.5 in some experiments, see Results). Aliquots of this DMSO solution were added directly to F(ab′)2 solutions and the samples incubated for 2 h at RT. Samples were then dialyzed in phosphate buffer as above. Deprotection conditions for the MeAcm-treated F(ab′)2 consisted of addition of 200 mL of the sample above to 500 mL of 15:1 concentrated ammonia/glacial acetic acid (21, 23) which was allowed to incubate at RT for 90 min. Samples were then dialyzed versus 160 mM phosphate buffer (pH 6.7) at 4 °C in preparation for dextran treatment. NHS-Msc-treated samples were deprotected by incubation at pH 10 in borate buffer for 4 h at RT prior to dialysis in phosphate buffer. Dextran Attachment to F(ab′)2 and Fab′-βL Conjugate. Oxidized, lyophilized dextran (average molecular weight of 6000 ( 500), prepared by oxidation with sodium periodate as described previously (13), was freshly prepared by dissolving solid oxidized dextran in 160 mM phosphate buffer at 600 mg/mL. An aliquot of this solution was added to the protein at a concentration of 12 mg of oxidized dextran per milligram of protein. Conjugate concentrations were calculated using an 0.1% 280 of 2.0. Samples were incubated at RT for 15 min, and then a freshly prepared solution of 1 M NaBH3CN was added in 10-fold molar excess over the dextran and the sample incubated for an additional 2 h at RT. To remove excess dextran and other small molecules, samples were then applied to an FPLC Pharmacia Superose 12 HR10/30 column equilibrated in 100 mM phosphate (pH 7.0), and the main protein peak was collected. Typically about 10% of the A280 eluted at a retention time consistent with protein dimer (confirmed by SDS-PAGE), the result of intermolecular dextran cross-linking, and this peak was discarded. The flow rate was 0.7 mL/min. After peaks were collected from the Superose column, the samples were immediately cooled on ice and an aliquot of freshly prepared 2 M NaBH4 was added to a final concentration of 100 mM and the sample allowed to set at 4 °C overnight. Samples were then dialyzed versus PBS prior to further analysis. The size exclusion chromatography data shown in Results was performed with 2 Zorbax G-250 columns in series, eluted with 100 mM phosphate buffer (pH 7) at a flow rate of 1 mL/min. SDS-PAGE was performed on the Pharmacia PhastSystem with 4 to 15% gradient gels under reducing or nonreducing conditions as described in Results. Native PAGE was also performed on the PhastSystem utilizing native buffer strips. Immunoreactivity and Biodistribution. The immunoreactivity of (Fab′)2 and conjugate samples was determined using a competitive assay format. Intact ZCE025 antibody was iodinated to a specific activity of 3.8 × 106 cpm/µg by the solid state Iodobead (Pierce) procedure according to the method of Markwell (28). Serial dilutions of the samples from 2.5 to 0.125 µM were combined with 100 000 cpm of 125I-labeled ZCE025 and diluted to a total volume of 25 µL in PBS plus 10% γ-globulin-free horse serum (AHS) (Gibco) which was then added to duplicate wells of microtiter plates containing dried LS174T cells. After incubation for 60 min at room temperature, the cells were washed with PBS plus 10% AHS and bound counts per minute in the wells was determined by γ-counting. In addition, the immunoreactivity of conjugate samples was tested by solid-phase RIA against antigen(+) LS174T and antigen(-) M14 tumor cells. A specific activity of 1.3 × 107 cpm/µg was obtained for the MeAcm control, while a specific activity of 1 × 107 cpm/µg was reached for the dextranated conjugate. 125I-labeled conjugate was

152 Bioconjugate Chem., Vol. 7, No. 1, 1996

Mikolajczyk et al. Scheme 1. Reaction of Protein Lysine Residues with Amine Reactive Reagents, Methyl Acetimidate (MeAcm) and N-Succinimidyl (Methylsulfonyl)ethyl Carbonate (NHS-Msc)a

Figure 1. Immune response in mice to βL with and without attached dextran. βL (20 mg) was injected into six mice at time 0 and at 4 weeks, as indicated by the arrows on the graph. Mouse serum titers were determined on pooled samples at each time point using microtiter plate ELISA assays as described in Experimental Procedures. Titers were defined as the dilution of sera resulting in 50% of the maximum response against unmodified βL. Open squares are the serum titers of mice injected with unmodified βL, and open circles are the titers for dextran-modified βL.

diluted in PBS plus AHS and added to duplicate wells at serial dilutions from 4 × 105 to 5 × 104 cpm in 25 µL aliquots. After incubation for 60 min at RT, samples were washed and counted as above. Evaluation of Immunogenicity. The immunogenicity of βL was evaluated by iv injection of 20 mg of dextranated and nondextranated βL in groups of 6 mice each and measurement of the resulting anti-mouse response in pooled serum samples, using a modification of the ELISA assay previously reported (13). Briefly, microtiter plate wells were coated by exposure to 0.25 mg of unmodified βL in 10 mM sodium phosphate buffer (pH 7.0) for 24 h and then washed thoroughly. Nonspecific binding sites were blocked with a 1 mg/mL solution of BSA in the same buffer. Wells were incubated with serial dilutions of mouse sera for 24 h and washed thoroughly. Wells were then exposed to goat anti-mouse immunoglobulins conjugated to HRP (TAGO Inc., Burlington, CA), and color was developed with the addition of 1 mg/mL o-phenylenediamine in 0.1% (v/v) hydrogen peroxide. Color development was quantified by measuring the absorbance at 490 nm with an ELISA reader. Titers were defined as the reciprocal of the highest dilution of mouse sera resulting in 50% of the maximum response. RESULTS

In Vivo Immune Response to βL ( Dextran Modification. In vivo studies in mice were performed to confirm that the reduction in immunogenic response to dextran-modified βL was comparable to that seen for other proteins (13, 14). Figure 1 shows that the immunogenic response to the strongly antigenic bacterial protein was virtually eliminated after dextran modification. Approximately 4 mol of dextran was attached per mole of βL as determined by size exclusion chromatography, SDS-PAGE, and the anthrone assay (29). This represents the maximum level of dextran attachment under the conditions used in this work. Effect of MeAcm on ZCE025 F(ab′)2 Dextran Attachment. Scheme 1 shows that the reaction product of MeAcm with the -amino group of lysine resembles arginine in structure and has a pK similar to that of

a

R ) the remainder of the polypeptide chain on the protein.

Figure 2. Cation exchange chromatography of ZCE025 F(ab′)2 samples. (A) Solid line, native F(ab′)2; dashed line, F(ab′)2 treated with 300 mM MeAcm. (B) Solid line, native F(ab′)2; dotted line, F(ab′)2 treated with 100 mM NHS-Msc. The native F(ab′)2 was well retained on the column, while the sample treated with NHS-Msc eluted in the void volume, indicating that the protein had insufficient positive charges to be retained on the column.

arginine at 12.5 (21, 23). As such, the pK is 2 orders of magnitude higher than that of lysine, and the product is essentially nonreactive with lysine-reactive reagents at pHs of 8 and below. Figure 2 shows that the retention time of F(ab′)2 by cation exchange chromatography was unaffected by reaction with MeAcm, indicating that the overall charge of the protein was not substantially altered under the column buffer conditions at pH 4.5. Native PAGE of F(ab′)2 treated with 300 mM MeAcm was, however, able to discriminate the slight charge difference from native F(ab′)2 in that such samples had somewhat slower mobility on the gel, indicating a more positively charged molecule than native F(ab′)2 (gel not shown). The immunoreactivity of MeAcm-treated samples was not appreciably different from that of native F(ab′)2 (Figure 3, curve 6). Subsequent treatment of this sample under standard dextranation procedures (see Experimental Procedures) showed no reduction in the immunoreactivity (curve 8), whereas dextran attachment to native F(ab′)2 showed almost complete loss of immunore-

Bioconjugate Chem., Vol. 7, No. 1, 1996 153

Dextran Modification of Protein Conjugates

Figure 3. Immunoreactivity by competitive assay of F(ab′)2 samples treated with 300 mM MeAcm or 100 mM NHS-Msc, with and without subsequent dextran attachment. F(ab′)2 samples indicated in the plot legend show increasing immunoreactivity in approximately the following order: 1, native F(ab′)2 treated with dextran; 2, NHS-Msc at pH 8; 3, NHS-Msc at pH 7.5; 4, NHS-Msc at pH 7.5 followed by dextran treatment; 5, NHS-Msc at pH 8 plus dextran; 6, MeAcm at pH 10; 7, 300 mM MeAcm at pH 9; 8, 300 mM MeAcm pH 10 plus dextran; 9, 300 mM MeAcm at pH 9 plus dextran; 10, F(ab′)2 control with no treatment. The immunoreactivities fell into three main categories. (1) Control dextran-modified with dextran was clearly most inhibited, while the immunoreactivity after NHSMsc treatment alone at pH 8 was also severely impaired. (2) Intermediate levels were obtained with NHS-Msc at pH 7.5 and after dextran modification of samples after treatment with NHSMsc at either pH 7.5 or 8. (3) Near control values were obtained with all MeAcm-treated samples.

activity (curve 1). Partial removal of the acetimidyl groups by limited treatment with ammonium hydroxide/ acetic acid (pH 11) (21, 23) for 90 min at RT showed no adverse effect on the immunoreactivity of the F(ab′)2 and no change in the ion exchange retention time shown in Figure 2A. Subsequent dextran treatment of these samples did, however, restore about half the dextran attachment as compared to native F(ab′)2. In addition to the standard MeAcm reaction conditions at pH 10, reaction at pH 9 was also performed in order to test the pH sensitivity and selectivity of this reagent at a lower pH. Figure 3 shows no appreciable difference in immunoreactivity after dextranation whether the acetimidylation pretreatment was performed at pH 9 or 10 (curves 8 and 9, respectively). A slight retention time and peak width increase observed by size exclusion chromatography did, however, suggest that MeAcm derivatization at pH 10 was somewhat more effective in blocking dextran attachment (data not shown). In addition to the F(ab′)2 of ZCE025 described above, the immunoreactivity of the F(ab′)2 fragment of CC49 was also investigated and showed almost identical results under the same reaction conditions (data not shown). Treatment of ZCE025 F(ab′)2 with Variable Levels of MeAcm. The effect of different concentrations of MeAcm pretreatment on subsequent dextran attachment was studied as a means of determining optimal reaction conditions and the effect of different levels of attached dextran on the immunoreactivity. Figure 4 shows the increase in retention time by SEC, indicating a decrease in Mr, as the level of MeAcm pretreatment was increased prior to dextran modification. Table 1 shows the calculated values for the moles of dextran attached per mole of F(ab′)2 based on the retention times in Figure 4. In addition, Table 1 shows the number of reactive amines available after MeAcm treatment as determined by TNBS. An average of 3-4 amines per mole of oxidized

Figure 4. Size exclusion chromatography (SEC) overlay of F(ab′)2 samples treated with increasing levels of MeAcm prior to dextran modification. Increased retention times indicate a lower apparent Mr. Peak 7 shows native F(ab′)2 with no MeAcm or dextran treatment. MeAcm treatment did not affect the retention time of native F(ab′)2. Peaks 1-6 show the retention times of samples modified with dextran subsequent to the following pretreatment with MeAcm: 1, native F(ab′)2 with no pretreatment; 2, 5 mM; 3, 10 mM; 4, 20 mM; 5, 50 mM; 6, 300 mM. The 100 mM peak was not plotted for clarity but had the same retention time as the 300 mM sample, though with a slightly wider peak width at base line. Peak 1 represents F(ab′)2 with approximately 10 mol of dextran attached per F(ab′)2. Table 1 MeAcm reactive amines dextran chains Amines per treatment (mM) per F(ab)′2 per F(ab)′2 dextran 0 5 10 20 50 100 300

39 25 17 10 5 3.4 1.4

9.2 7.3 6.2 3.8 1 0.5 0.5

4.2 3.4 2.7 2.6 5 6.8 2.8

dextran was available for attachment within the range of 0-20 mM MeAcm treatment. Levels of 100 and 300 mM MeAcm resulted in identical retention times as seen in Figure 4, indicative of minimal or no dextran attachment. The same general trend in gradually decreasing molecular weight was also seen in SDS-PAGE of the same samples (Figure 5). When run under reducing conditions (Figure 5A), the SDS-PAGE demonstrated another aspect of dextran attachment, intramolecular crosslinking. A single F(ab′)2 molecule is composed of two identical Fab′ molecules held together by three disulfide bonds at the hinge region (25, 26). Each Fab′ molecule is in turn composed of a heavy and light chain held together by a single disulfide bond. Thus, each 100 kDa F(ab′)2 molecule is composed of 4 mol of individual polypeptide chains, visible as two poorly resolved bands of approximately 24 and 26 kDa when run denatured and reduced on SDS-PAGE (Figure 5A, lane 8). Regardless of whether the sample was run under reducing conditions, the dextran-modified F(ab′)2 had the same mobility on SDS-PAGE which indicated that the polypeptide chains of dextranated F(ab′)2 were fully crosslinked (Figure 5, lanes A1 and B1). The Fab′-Fab′ as well as the Fab′ heavy-light chains remained covalently attached despite reduction of the disulfide bonds in lane B1. However, after pretreatment with 5 mM MeAcm, four broad and heterogeneous bands of protein were discernible from the reduced gel, lane A2, consistent with the four possible forms of covalent attachment: intact F(ab′)2

154 Bioconjugate Chem., Vol. 7, No. 1, 1996

Figure 5. Gradient PhastGel (4 to 15%) of F(ab′)2 samples pretreated with increasing levels of MeAcm prior to dextran modification. Gel A was run under reducing conditions with 5% 2-mercaptoethanol, while gel B was run under nonreducing conditions. Lane 8 shows native F(ab′)2 with no treatment with MeAcm or dextran. Identical samples were applied to lanes 1-8 on each gel: lane 1, native F(ab′)2 with no MeAcm pretreatment; lane 2, 5 mM; lane 3, 10 mM; lane 4, 20 mM; lane 5, 50 mM; lane 6, 100 mM; lane 7, 300 mM; lane 8, native F(ab′)2. Gel A demonstrated dextran intramolecular cross-linking of the F(ab′)2 molecule under conditions which reduced the disulfide bonds holding its fragment parts together. Lanes A1 and B1 are identical, indicating that the F(ab′)2 was fully cross-linked by the dextran. Lanes A2-A7 showed a decreasing percentage of cross-linking as the level of MeAcm pretreatment was increased. Gel B shows the increased mobility and less heterogeneous bands of F(ab′)2 as the level of dextran attachment decreased.

as in lane A1, two lower bands consistent with the loss of one and then two of the four component fragment parts, and finally the free heavy and light chains as the fastest running bands. Pretreatment with 10 mM MeAcm (lane A3) showed only the dimeric and single chains remaining. With the pretreatment levels of MeAcm increased to 20 and 50 mM MeAcm (lanes A4 and A5), the level of dextran cross-linking of the F(ab′)2 polypeptide chains reached minimal levels, though some apparent dextran attachment on the individual polypeptides was still evident since the bands ran slightly slower and were more heterogeneous than those of native heavy-light chains seen in lane A8. This is supported by the values in Table 1, where calculated dextran attachment dropped significantly after treatment with 50 mM MeAcm. The immunoreactivity of these samples appeared to be directly proportional to the level of dextran modification as modulated by pretreatment with MeAcm (Figure 6). The higher levels of MeAcm pretreatment which resulted in lowered dextran attachment also showed the highest immunoreactivity. At MeAcm concentrations of 50 mM, where all significant intramolecular cross-linking was blocked, the immunoreactivity approached that of native F(ab′)2. Effect of NHS-Msc on Dextran Attachment to ZCE025 F(ab′)2. Scheme 1 shows the product formed when NHS-Msc reacts with the -amine of lysine. Aside from blocking the -amino group of the lysine, the main effect of the Msc derivative on the protein is neutralization of the positive charge which is normally present on the lysine at typical physiological pHs. This effect was demonstrated by cation exchange chromatography of ZCE025 F(ab′)2 before and after treatment with 100 mM NHS-Msc (Figure 2B). Native F(ab′)2 bound well to the column matrix, but after NHS-Msc treatment, the derivatized F(ab′)2 did not bind to the column, instead eluting in the void volume. Samples on native PAGE also showed greater mobility after NHS-Msc treatment, indicating a more negatively charged protein (gel not shown).

Mikolajczyk et al.

Figure 6. Competitive assay showing the increase in immunoreactivity as the level of MeAcm pretreatment was increased prior to dextran modification. These were the same samples shown in Figures 4 and 5. Pretreatment levels from 0 to 20 mM MeAcm showed clearly impaired immunoreactivity after dextran modification, whereas dextran-modified samples pretreated with 50-300 mM MeAcm showed values comparable to those of native F(ab′)2.

However, NHS-Msc treatment under standard conditions at pH 8 had an adverse effect on the immunoreactivity of F(ab′)2 (Figure 3, curve 2). Reaction with NHSMsc was also tested at a lower pH, pH 7.5 (curve 3), to monitor reagent sensitivity and selectivity at a lower pH. While cation exchange chromatography and native PAGE gels both suggested comparably charged protein with NHS-Msc reaction at either pH 7.5 or 8 (comparative data not shown), NHS-Msc treatment at pH 8 was more inhibitory to the immunoreactivity (Figure 3) which, by inference, suggested higher levels of neutralized lysines. Curiously, dextran attachment after pH 8 NHS-Msc treatment showed slightly improved immunoreactivity over the nondextranated NHS-Msc control (curve 4 versus 2). Another aspect of working with Msc-derivatized F(ab′)2 was anomalously high retention times by size exclusion chromatography on the Zorbax columns and conditions used in this work. While likely due to column matrix effects (since slightly denatured proteins would be expected to have decreased retention times by SEC), the observation was not investigated further. The overall ability of NHS-Msc to block dextran attachment was lower than that attained with MeAcm. After 100 mM NHS-Msc pretreatment, and subsequent dextran attachment, the F(ab′)2 on reduced SDS-PAGE looked much like the 10 mM MeAcm pretreatment seen in lane 3 of Figure 5A (NHS-Msc gel not shown). NHSMsc in higher levels was not soluble. Attempts to use the p-nitrophenyl derivative of Msc as a means to increase the reaction with amines resulted only in precipitation of the protein. F(ab′)2 was also reacted with NHS-Mac at lower concentrations to test the possibility that a particularly reactive amine was one of the key sites responsible for the reduced immunoreactivity. Reaction of F(ab′)2 with 2, 5, and 10 mM NHS-Msc clearly blocked some amines, as indicated by increased mobility on native PAGE, but the immunoreactivity of these samples both before and after dextran attachment was not discernibly different from that of native F(ab′)2 controls. Incubation at pH 10 in borate buffer for 4 h at RT removed about half of the Msc groups as indicated by intermediate binding on the cation exchange column as well as increased dextran attachment, but such treatment failed to restore the immunoreactivity (data not shown).

Dextran Modification of Protein Conjugates

F(ab′)2 fragments of CC49 were also reacted with NHSMsc under the above conditions and showed somewhat greater inhibition of immunoreactivity than observed for ZCE025. Biodistribution of Fab′-βL Conjugate ( Dextran Treatment. βL and ZCE025 Fab′ were coupled using sulfo-SMCC as described previously (17) except that the F(ab′)2 was pretreated with 100 mM MeAcm prior to reduction to Fab′ with 20 mM cysteine. Dextran modification of this conjugate was performed at the same mass ratio of oxidized dextran to conjugate as used for F(ab′)2, i.e., 12 mg of dextran per milligram of conjugate protein. Since the Fab′ half of the conjugate was essentially unreactive with dextran, the ratio of dextran mass to βL protein mass was approximately 2 times higher than that used in the model studies with F(ab′)2. The βL activity of the dextran-modified conjugate was identical to that of native βL on a molar basis. SDS-PAGE of the dextranated conjugate showed a single band under nonreducing conditions which was similar in heterogeneous appearance to the broad band of F(ab′)2 seen in lane 1 of Figure 5B. Band mobility on SDS-PAGE suggested an average of approximately 4 mol of dextran per conjugate, consistent with the nominal value expected for βL alone. The immunoreactivity of this Fab′-blocked, βL-dextranated conjugate was the same as that of the Fab′ used to make the conjugate, as determined by the competition assay formats seen in Figures 3 and 6. To serve as a single-arm control for immunoreactivity studies, an aliquot of the MeAcm-treated Fab′ was removed prior to coupling to βL, and the sulfhydryls were capped with NEM. In addition to the sample used for biodistribution studies above, another sample of the Fab′-βL conjugate was treated at a 10 times higher level of dextran, i.e., 120 mg per milligram of conjugate protein. The observed immunoreactivity, βL activity, and SDS-PAGE mobility for this sample were not discernibly different than those of samples treated at 12 mg of dextran per milligram protein. Prior to biodistribution studies, both dextranated and nondextranated conjugate were also tested versus antigenpositive and antigen-negative cells. In Figure 7, the conjugates show good binding to antigen-positive cells, though the dextranated conjugate was clearly lower than the MeAcm-treated Fab′ or Fab′ control. No nonspecific binding to antigen-negative cells was observed with dextranated or nondextranated conjugate. The biodistribution studies of the dextranated conjugate showed a longer half-life in all tissues, including the blood and the tumor, compared to the MeAcm-treated, nondextranated control conjugate (Figure 8). Typical protocols with Fab′-βL conjugate wait 72-96 h after injection before administration of prodrug to allow conjugate clearance from the blood in order to increase the tumor to blood ratio (1). As expected from other work (14, 30), attachment of polymers to proteins increased the overall circulating half-life. From Figure 8, it can be seen that the clearance of the dextranated conjugate from the blood and tissues is generally delayed about 4872 h over that of the nondextranated control conjugate. The tumor to blood ratio for the dextranated conjugate was lower than for nondextranated conjugate at all time points.

Bioconjugate Chem., Vol. 7, No. 1, 1996 155

Figure 7. Direct measure of the binding of 125I-labeled Fab′βL conjugate to antigen-positive (LS174T) and antigen-negative (M14) cells. Samples are as follows. Circles are untreated Fab′ which serves as a single-arm binding control. This sample was treated with NEM to block hinge region sulfhydryls. Squares are conjugate prepared with Fab′ treated with 100 mM MeAcm prior to coupling to the βL. Triangles are conjugate prepared with 100 mM MeAcm pretreatment of the Fab′ and subsequent dextran modification of the entire conjugate. The solid lines show binding to LS174T cells, and the dotted lines show binding to M14 cells. Negligible nonspecific binding to antigen-negative cells was observed. While reduced binding to antigen-positive cells was seen under these conditions, competitive assays performed as seen in Figures 3 and 6 showed no significant differences between the control, MeAcm-treated, and MeAcmtreated dextranated conjugate.

DISCUSSION

Figure 8. Biodistribution in tumor-bearing nude mice of 125Ilabeled conjugate. Direct binding assay of these samples is shown in Figure 7. (A) Dextran-modified conjugate prepared from Fab′ pretreated with 100 mM MeAcm prior to coupling to βL. (B) Conjugate prepared as above except with no dextran modification. Dextranated conjugate showed longer clearance times from the blood and tissues compared to the control conjugate without dextran.

This work demonstrates the feasibility of construction of Fab′-enzyme bioconjugates with differential levels of polymeric protection. In this case, it was considered

essential that the βL portion of the conjugate be adequately dextran-modified since there are no currently

156 Bioconjugate Chem., Vol. 7, No. 1, 1996

available methods for amelioration of the immunogenic response to this bacterial protein by recombinant technology as exists for antibodies. Dextran modification of the βL was, by itself, an effective method of reducing the immunogenic response as seen in Figure 1. While the two Fab’s tested in this work were of murine origin, and would thus be expected to be unsatisfactory for human trials, a number of humanized fragments are under development in this laboratory as potential candidates for the final conjugate. It was deemed appropriate, however, to establish the parameters and limits on the modulation of dextran attachment in model studies since such methodology should, in principle, be applicable to most proteins of future interest. It was also of some interest to determine if conditions could be found which would allow partial dextran modification while preserving immunoreactivity, thereby opening new possibilities for otherwise unsuitable murine antibody fragments. F(ab′)2 was used to study the amine-blocking reagents since this molecule is stable and can be easily reduced to Fab' for conjugation to βL. Reagent treatment and dextranation of Fab′ was performed in some experiments and showed no discernible differences compared to that of F(ab′)2. The effect of the amine-blocking reagents was evaluated by measuring the changes in dextran attachment, in effect, a measure of available amines with regard to both the moles of dextran per F(ab′)2, as determined by size exclusion chromatography in Figure 4, and the degree of dextran intramolecular cross-linking as indicated in Figure 5. TNBS was used to provide approximate values for the number of reactive amines before and after MeAcm treatment. Though subject to interference when used for absolute quantitation (22), the number of amines determined as being available by this technique gave good proportionality to the number of dextrans attached to the protein (Table 1). The trend showing availability/attachment of 3-4 amines per mole of dextran in the MeAcm treatment range of 0-20 mM seen in Table 1 generally reflects the cross-linking trend seen in Figure 5. After treatment with 50 mM or higher MeAcm, the few remaining amines may have been sterically hindered, though the 50 mM MeAcm sample itself clearly had at least one dextran attached as indicated in Figure 4. The 50 mM sample was also the most heterogeneous as indicated by irregular peak shape and increased peak width in Figure 5 which reflects its transitional status between samples with cross-linked dextran attachment and those samples with essentially no dextran attachment. Of the two amine-blocking reagents tested, MeAcm and NHS-Msc, MeAcm appeared to be the reagent of choice. MeAcm had little effect on the activity of the Fab′ binding by itself, possibly due to the similarity in size and charge of the reaction product to lysine (Scheme 1). Pretreatment of the F(ab′)2 with increasing levels of MeAcm was able to facilitate a relatively orderly decrease in the number of dextrans attached in subsequent dextranation steps (Figure 4). An examination of Figures 4-6 suggests that there may be a level of MeAcm pretreatment at about 20-50 mM which could yield a conjugate with sufficient immunoreactivity but still with some protective dextran on the Fab′. Since even humanized antibodies and fragments can elicit anti-idiotypic immune response (2, 3), it may be desirable in some cases to maximize the level of dextran attachment even to sensitive antibodies/ fragments which may have already been engineered for reduced immunogenicity. Clearly the reaction conditions will be different for different proteins, but this work

Mikolajczyk et al.

establishes a framework within which to determine those parameters. The exact reasons for loss of protein activity, in this case the loss of Fab′ immunoreactivity, after treatment with dextran is not clear. It is speculated that the loss of immunoreactivity after dextranation was likely due, at least in part, to steric hindrance. The antigens for both ZCE025 and CC49, CEA (31) and TAG-72 (32), respectively, are large proteins and as such might be expected to be more sensitive to steric problems in the proximal association with the antibody binding site. Such a relationship with substrate size was demonstrated with chymotrypsin, an enzyme which exhibited differential activity toward substrates of different sizes after it had been modified with poly(ethylene glycol) (33). The ability of the PEG-modified chymotrypsin to act on a denatured form of BSA but not on native BSA suggested that conformation of the substrate played a role as well. This may explain why dextran-modified T101 antibody maintained good immunoreactivity with its antigen (13) even though the antigen is a large cell surface protein (34). Evidence from the current work supports the idea of steric hindrance to some extent in that the immunoreactivity approached native levels only at MeAcm concentrations where intramolecular cross-linking by the dextran was largely blocked (Figure 5). On the other hand, loss of immunoreactivity after treatment with NHS-Msc alone might suggest that reduced immunoreactivity resulted from loss of positive charge, also a consequence of amine reaction with dextran. Disruption of protein conformation or simple charge repulsion of the antigen could also account for loss of immunoreactivity after NHS-Msc treatment. Given that treatment with NHSMsc levels as high as 10 mM had no effect on the immunoreactivity even though the net charge on the protein became more negative, it is at least evident that there was no particularly reactive or sensitive amino group with influence on the binding kinetics. NHS-Msc may have application for limited blocking of key lysines or to deliberately increase anionic character of the protein for purposes of altered biodistribution, but its effectiveness was too limited for the current objectives. Though the main drawback to using NHS-Msc was loss of immunoreactivity, this reagent also showed a more limited ability to block dextran attachment. Levels of NHS-Msc of 100 mM were able to prevent dextran crosslinking only to the level achieved with about 10 mM MeAcm as determined by SDS-PAGE under reducing conditions (data not shown). Given the clear excess of reagent, it would appear that reaction with additional NHS-Msc was sufficiently unfavorable as to preclude further reaction. In fact, attempts to force further reaction with the p-nitrophenyl derivative of Msc simply precipitated the protein. Even with the NHS-Msc reagent, the occasional failure to recover the expected levels of protein suggested that protein stability was at issue. The results from the immunoreactivity and biodistribution studies in Figures 3 and 8, respectively, suggest that removal of the acetimidyl groups on the Fab′ portion of the conjugate may not be necessary to achieve an effective conjugate. From other studies with both dextranated (13, 14) and PEG-modified proteins (12, 30), it was anticipated that the dextran-modified version of the Fab′-βL conjugate would have an increased half-life in the blood compared to nondextranated conjugate. This may be due in part to the increased stability of such conjugates as measured by their resistance to proteases and denaturants (30, 35). Figure 8 shows the clearance half-life of the dextranated conjugate was prolonged

Dextran Modification of Protein Conjugates

about 48-72 h over that of the nondextranated control. For some applications, this approach has been used deliberately to increase blood residence of proteins which would otherwise be cleared too rapidly (14). There is often a relationship between higher circulating levels of antibody and the levels found localized at the tumor site (36), a correlation also observed in Figure 8. At the 48 and 72 h time points, the control levels in the tumor and blood have begun to drop appreciably while the level of dextranated conjugate remained much higher. For therapeutic applications, however, a high tumor to blood ratio is more important than the overall level in the tumor. For the enzymatic, site-specific activation of prodrugs, a higher tumor to blood ratio is desirable for minimization of nonspecific activation of prodrug by circulating conjugate. It is evident from Figure 8 that the tumor to blood ratio is higher for the nondextranated conjugate. While this is not the desired trend, the pharmacological variables are sufficiently complex that no conclusions can be drawn without the therapeutic studies to evaluate the efficacy of this conjugate after increased blood clearance protocols have been implemented. If blood clearance ultimately proves to be a limiting factor, another approach is the addition of agents to artificially remove circulating conjugate from the bloodstream as has been used successfully by others (37, 38). Since dextran modification has been shown in this and previous work to be an effective method of reducing βL immunogenicity, future studies will focus on the immunogenic response to the various forms of conjugates suggested by this investigation, including (1) the current conjugate with partially dextran-modified murine Fab′, (2) conjugate containing humanized Fab′ with and without dextran modification, and (3) conjugates with intermediate levels of dextran attachment on both the βL and the Fab′ portion of the conjugate. ACKNOWLEDGMENT

We gratefully acknowledge Dr. Robin Offord, University of Geneva, for his advice and suggestions during the course of this work and for his critical review of the manuscript. LITERATURE CITED (1) 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-β-Lactamase Conjugates: Regression and LongTerm Growth Inhibition of Human Colon Carcinoma Xenograft Models. Cancer Res. 53, 3956-3963. (2) Hinman, L. M., and Yarranton, G. (1993) Annu. Rep. Med. Chem. 28, 237-246. (3) Shawler, D. L., Bartholomew, R. M., Smith, L. M., and Dilman, R. O. (1985) Human Immune Response to Multiple Injections of Murine Monoclonal IgG. J. Immunol. 135, 15301535. (4) Khazaeli, M. B., Saleh, M. N., Wheeler, R. H., Hustler, W. J., Holden, H., Carrano, R., and LoBuglio, A. F. (1988) Phase I Trial of Multiple Large Doses of Murine Monoclonal Antibody CO17-1A. II. Pharmacokinetics and Immune Response. J. Natl. Cancer Inst. 80, 937-942. (5) Khazaeli, M. B., Saleh, M. N., Liu, T. P., Meredith, R. F., Wheeler, R. H., Baker, T. S., King, D., Secher, D., Allen, L., Rogers, K., Colcher, D., Schlom, J., Shochat, D., and LoBuglio, A. F. (1991) Pharmacokinetics and Immune Response of 131IChimeric Mouse/Human B72.3 (Human g4) Monoclonal Antibody in Humans. Cancer Res. 51, 5461-5466. (6) Phelps, J. E., Beidler, D. E., Jue, R. A., Unger, B. W., and Johnson, M. W. (1990) Expression and Characterization of a Chimeric Antibody with Therapeutic Applications. J. Immunol. 145, 1200-1204.

Bioconjugate Chem., Vol. 7, No. 1, 1996 157 (7) Beidler, C. B., Ludwig, J. R., Cardenas, J., Phelps, J., Papworth, C. G., Melcher, E., Sierzega, M., Myers, L. J., Unger, B. W., Fisher, M., David, G. S., and Johnson, M. J. (1988) Cloning and High Level Expression of a Chimeric Antibody with Specificity for Human Carcinoembryonic Antigen. J. Immunol. 141, 4053-4060. (8) Yokota, T., Milenic, D. E., Whitlow, M., and Schlom, J. (1992) Rapid Tumor Penetration of a Single-Chain Fv and Comparison with Other Immunoglobulin Forms. Cancer Res. 52, 3402-3408. (9) Milenic, D. E., Yokota, T., Filpula, D. R., Finkelman, M. A. J., Dodd, S. W., Wood, J. F., Whitlow, M., Snoy, P., and Schlom, J. (1991) Construction, binding properties, metabolism and tumor targeting of a single-chain Fv derived from the pancarcinoma monoclonal antibody CC49. Cancer Res. 51, 6363-6371. (10) Use of functionalized polyethylene glycols for modifications of polypeptides, pp 347-370, Zalipsky, S., and Lee, C., Eds. (1992) Plenum Press, New York. (11) Delgado, C., Francis, G. E., and Fisher, D. (1992) The Uses and Properties of PEG-Linked Proteins. Crit. Rev. Ther. Drug Carrier Syst. 9, 249-304. (12) Beauchamp, C. O., Gonias, S. L., Menapace, D. P., and Pizzo, S. V. (1983) A New Procedure for the Synthesis of Polyethyelene Glycol-Protein Adducts, Effects on Function, Receptor Recognition, and Clearance of Superoxide Dismutase, Lactoferrin, and R-2-Macroglobulin. Anal. Biochem. 131, 25-33. (13) Fagnani, R., Hogan, M. S., and Bartholomew, R. (1990) Reduction of Immunogenicity by Covalent Modification of Murine and Rabbit Immunoglobulins with Oxidized Dextrans of Low Molecular Weight. Cancer Res. 50, 3638-3645. (14) Wileman, T. E., Foster, R. L., and Elliott, P. N. C. (1986) Soluble asparaginase-dextran conjugates show increased circulatory persistence and lowered antigen reactivity. J. Pharm. Pharmacol. 38, 264-271. (15) Richter, W., and Kagedal, L. (1972) Preparation of DextranProtein Conjugates and Studies of their immunogenicity. Int. Arch. Allergy Immunol. 42, 885-902. (16) Kabat, E. A., and Bezer, A. E. (1958) The effect of variation in molecular weight on the antigenicity of dextran in man. Arch. Biochem. Biophys. 78, 306-318. (17) Meyer, D. L., Jungheim, L. N., Mikolajczyk, S. D., Shepherd, T. A., Starling, J. J., and Ahlem, C. N. (1992) Preparation and Characterization of a β-Lactamase-Fab′ Conjugates for the Site-Specific Activation of Oncolytic Agents. Bioconjugate Chem. 3, 42-48. (18) Mikolajczyk, S. D., Meyer, D. L., Starling, J. J., Law, K. L., Rose, K., Dufour, B., and Offord, R. E. (1994) High Yield, Site-Specific Coupling of N-Terminally Modified β-Lactamase to a Proteolytically Derived Single-Sulfhydryl Murine Fab′. Bioconjugate Chem. 5, 636-646. (19) Fagnani, R., Halpern, S., and Hagan, M. (1995) Altered pharmacokinetic and tumour localization properties of Fab' fragments of a murine monoclonal anti-CEA antibody by covalent modification with low molecular weight dextran. Nucl. Med. Commun. 16, 362-369. (20) Yoshitake, S., Yamada, Y., Ishikawa, E., and Masseyeff, R. (1979) Conjugation of Glucose Oxidase from Aspergillus niger and Rabbit Antibodies Using N-Hydroxysuccinimide Ester of N-(4-Carboxycyclohexylmethyl)-Maleimide. Eur. J. Biochem. 101, 395-399. (21) Ludwig, M. L., Hunter, and M. J. (1967) Amidination. Methods Enzymol. 11, 595-605. (22) Reynolds, J. H. (1968) Acetimidation of Bovine Pancreatic Ribonuclease A. Biochemistry 7, 3131-3135. (23) Wallace, C. J. A., and Harris, D. E. (1984) The preparation of fully N--acetimidylated cytochrome c. Biochem. J. 217, 589-594. (24) Tesser, G. I., and Balvert-Geers, I. C. (1975) The methylsulfonylethyloxycarbonyl group, a new and versatile amino protective function. Int. J. Pept. Protein Res. 7, 295-305. (25) Lamoyl, E. (1986) Preparation of F(ab)2 Fragments from Mouse IgG of Various Subclasses. Methods Enzymol. 121, 652-663.

158 Bioconjugate Chem., Vol. 7, No. 1, 1996 (26) Parham, P. (1983) On the Fragmentation of Monoclonal IgG1, IgG2a, and IgG2b from Balb/c Mice. J. Immunol. 131, 2895-2902. (27) Habeeb, A. F. S. A. (1966) Determination of Free Amino Groups in Proteins by Trinitrobenzenesulfonic acid. Anal. Biochem. 14, 328-336. (28) Markwell, M. A. K. (1982) A new solid-state reagent to iodinate proteins. Anal. Biochem. 125, 427-432. (29) Viles, F. J., Jr., and Silverman, L. (1949) Determination of Starch and Cellulose with Anthrone. Anal. Chem. 21, 950953. (30) Abuchowski, A., McCoy, J. R., Palczuk, N. C., Es, T. v., and Davis, F. F. (1976) Effect of Covalent Attachment of Polyethylene Glycol on Immunogenicity and Circulating Life of Bovine Liver Catalase. J. Biol. Chem. 252, 3582-3586. (31) Kuroki, M., Fernstein, P. D., Wunderlich, D., Colcher, D., Simpson, J. F., Poole, D. J., and Schlom, J. (1990) Serological mapping of the TAG-72 tumor-associated antigen using 19 distinct monoclonal antibodies. Cancer Res. 50, 4872-4879. (32) Johnson, V. G., Schlom, J., Peterson, A. J., Bennett, J., Magnani, J. L., and Colcher, D. (1986) Analysis of a human tumor-associated glycoprotein (TAG-72) identified by monoclonal antibody B72.3. Cancer Res. 46, 850-857. (33) Chiu, H.-C., Zalipsky, S., Kopeckova, P., and Kopecek, J. (1993) Enzymatic Activity of Chymotrypsin and Its Poly(ethylene glycol) Conjugates toward Low and High Molecular Weight Substrates. Bioconjugate Chem. 4, 290-295. (34) Royston, I., Majda, J. A., Baird, S. M., Meserve, B. L., and Griffiths, J. C. (1980) Human T cell antigens defined by

Mikolajczyk et al. monoclonal anitbodies: the 65,000-dalton antigen of T cells (T65) is also found on chonic lymphocytic leukemia cells bearing surface immunoglobulins. J. Immunol. 125, 725731. (35) Liliberte, M., Gayet, J.-C., and Fortier, G. (1994) Surface modification of horseradish peroxidase with poly(ethylene glycol)s of various relative molecular masses (Mr): relationship between the Mr of polyethylene glycol and the stability of horseradish peroxidase-poly(ethylene glycol) adducts under various denaturing conditions. Biotechnol. Appl. Biochem. 20, 397-413. (36) Boxer, G. M., Begent, R. H. J., Kelly, A. M. B., Southall, P. J., Blair, S. B., Theodorau, N. A., Dawson, P. M., and Ledermann, J. A. (1992) Factors influencing variability of localization of antibodies to carinoembryonic antigen (CEA) in patients with colorectal carcinoma--implications for radioimmunotherapy. Br. J. Cancer 65, 825-831. (37) Kerr, D. E., Garrigues, U. S., Wallace, P. M., Hellstrom, K. E., Hellstrom, I., and Senter, P. D. (1993) Application of Monclonal Antibodies against Cytosine Deaminase for the in Vivo Clearance of a Cytosine Deaminase Immunoconjugate. Bioconjugate Chem. 4, 353-357. (38) Sharma, S. K., Bagshawe, K. D., Burke, P. J., Boden, R. W., and Rogers, G. T. (1990) Inactivation and Clearance of an Anti-CEA Carboxypeptidase G2 Conjugate in Blood After Localisation in a Xenograft model. Br. J. Cancer 61, 659-662.

BC950088J