Comparison of Recombinant and Synthetically Formed Monoclonal

Antibody-Directed Enzyme Prodrug Therapy (ADEPT) - Basic Principles and its Practice So Far. Kenneth D. Bagshawe. 2011,169-186 ...
0 downloads 0 Views 89KB Size
Download PDF
1084

Bioconjugate Chem. 1999, 10, 1084−1089

Comparison of Recombinant and Synthetically Formed Monoclonal Antibody-β-Lactamase Conjugates for Anticancer Prodrug Activation David E. Kerr,† Vivekananda M. Vrudhula,‡ Ha˚kan P. Svensson,§ Nathan O. Siemers,| and Peter D. Senter* Bristol-Myers Squibb Pharmaceutical Research Institute, Seattle, Washington 98121. Received June 11, 1999; Revised Manuscript Received August 3, 1999

Conjugates of the L49 monoclonal antibody (binds to the p97 antigen on melanomas and carcinomas) were formed by attaching Enterobacter cloacae β-lactamase (bL) to the L49-Fab′ fragment using a heterobifunctional cross-linking reagent or by linking the enzyme to L49-sFv using DNA recombinant technology. The conjugates thus formed, L49-Fab′-bL and L49-sFv-bL, were designed to activate cephalosporin containing anticancer prodrugs at the surfaces of antigen positive tumor cells. Results from in vitro experiments using two lung carcinoma cell lines demonstrated that the conjugates were equally active in effecting the release of phenylenediamine mustard from the cephalosporin nitrogen mustard prodrug CCM. While treatment with either of the conjugates combined with the maximum tolerated doses of CCM led to cures of established SN12P renal cell carcinoma tumors in nude mice, only the L49-sFv-bL conjugate maintained its ability to do so at 1/4 the maximum tolerated dose of CCM. L49-sFv-bL was also superior to L49-Fab′-bL in the 1934J renal cell carcinoma tumor model and was shown to be quite active in two in vivo models of human lung carcinoma. These results demonstrate that the recombinant fusion protein leads to more pronounced therapeutic windows than the chemical conjugate and is active in an array of human tumor models.

INTRODUCTION

The use of antibody-enzyme conjugates for the activation of anticancer prodrugs has been the subject of a great deal of investigation (reviewed in refs 1-4). This targeting strategy, commonly referred to as antibody-dependent enzyme prodrug therapy (ADEPT), is a form of pretargeting that involves two separate steps to achieve therapeutic efficacy. Initially, a nontoxic monoclonal antibody-enzyme (mAb-enzyme) conjugate is administered that diffuses into solid tumor masses and binds to cell-surface antigens. Once sufficiently high tumor to nontumor ratios are achieved, a relatively nontoxic prodrug is injected. Upon contact with the targeted enzyme, the prodrug is catalytically converted into a cytotoxic drug. There are now many examples demonstrating that this drug-targeting strategy results in high intratumoral drug concentrations, pronounced antitumor activities at well tolerated doses, and to positive responses in early clinical trials (1-4). Several enzymes have been utilized for targeted prodrug activation (reviewed in refs 1-3). We (5-11) and * To whom correspondence should be addressed: Seattle Genetics, 22215 26th Avenue SE, Bothell, WA 98021. Phone: (425) 489-4993. Fax (425) 489-4798. E-mail: [email protected]. † Present address: Biocontrol Systems, 12822 SE 32nd St., Bellevue, WA 98005. ‡ Present address: Bristol-Myers Squibb Pharmaceutical Research Institute, Neuroscience and Genitourinary Chemistry, 5 Research Parkway, Wallingford, CT 06492. § Present address: Xcyte Therapies, 2203 Airport Way South, Suite 300, Seattle, WA 98134. | Present address: Bristol-Myers Squibb Pharmaceutical Research Institute, Department of Applied Genomics, Hopewell Building 3, PO Box 5400, Princeton, NJ 08543-5400.

others (12, 13) have focused on the use of Enterbacter cloacae β-lactamase (bL) as an activating enzyme, due to its high specific activities on an array of cephalosporin containing anticancer prodrugs. Conjugates were formed by linking bL to the L49 and 96.5 mAbs (9-11) which recognize distinct epitopes on the p97 melanotransferrin antigen, present on the surfaces of most human melanoma tumors and on some carcinomas (10, 14). One of the advantages in using bL for prodrug activation is that there is no known human counterpart that can catalyze cephalosporin β-lactam hydrolysis. Consequently, the conjugated enzyme is the principal source of prodrug activation. Therapy experiments in mice with melanoma xenografts have demonstrated that pronounced antitumor activities, including cures and complete regressions of established tumors, were obtained using the combination of 96.5-Fab′-bL with CCM (Figure 1), a prodrug from which phenylenediamine mustard (PDM) is released upon hydrolysis of the β-lactam bond (9). The prodrug in these experiments was injected 3 days post conjugate treatment, a period of time that allowed most of the nontargeted enzyme to clear from the systemic circulation. On the basis of these results, a recombinant L49sFv-bL fusion protein was prepared and expressed in Escherichia coli (11). In contrast to the chemically produced L49-Fab′-bL conjugate, the fusion protein underwent very rapid intratumoral uptake and systemic clearance. It was therefore possible to safely administer the prodrug 12 h after conjugate administration, leading to pronounced therapeutic efficacy. Although several mAb-enzyme fusion proteins have now been described for anticancer prodrug activation (8, 11, 15-19), their effects relative to chemically produced conjugates have not been reported. In this paper, we

10.1021/bc990075w CCC: $18.00 © 1999 American Chemical Society Published on Web 09/09/1999

Comparison of Monoclonal Antibody-β-Lactamase Conjugates

Figure 1. Structures of the cephalosporin mustard prodrug CCM and PDM, the drug released upon bL-catalyzed hydrolysis.

compare the antitumor activities obtained when L49Fab′-bL/CCM and L49-sFv-bL/CCM combinations are administered in mice with renal cell carcinoma xenografts. We show that the fusion protein leads to better therapeutic windows in two renal cell carcinoma models. In addition, we demonstrate that the combination of the fusion protein with CCM is effective, both in vitro and in vivo, in treating models of p97 antigen positive human lung adenocarcinomas. MATERIALS AND METHODS

Cell Lines and Reagents. The 3713 and 3754 lung carcinoma cell lines were developed at Bristol-Myers Squibb (Seattle, WA) from tumor biopsies obtained from patients with lung carcinoma. The SN12P and 1934J renal cell carcinoma lines have been reported earlier (10). The mAb L49 (IgG1) binds to melanotransferrin (p97), a human melanoma associated antigen that is also expressed on a number of carcinomas (10, 14). P1.17 (IgG2a) is a mouse myeloma protein that was used as a nonbinding control. L49-Fab′-bL was prepared as described earlier for other mAb-Fab′-bL conjugates (20). Briefly, the Fab′2 fragment of the L49 mAb was reduced with dithiothreitol and combined with maleimide-substituted bL. The conjugate was subjected to purification on a boronic acid affinity column (hydrophilic type L), followed by gel filtration on Sepahacryl S-300. L49-sFv-bL, containing the r2-1 bL mutant (21), was expressed in E. coli and purified as reported earlier (11). Purification involved a combination of affinity chromatographic steps, first using immobilized p97, followed by immobilized boronic acid. PDM and CCM were synthesized as reported earlier (5). Cell-Surface Expression of the p97 Antigen. The expression of melanotransferrin on SN12P and 1934J cells has been described previously (10). Fluorescein conjugates of L49 and P1.17 were used to assess antigen expression on the lung carcinoma cell lines. Cells were detached from monolayer cultures with trypsin, incubated in culture media with mAb-fluorescein conjugates at 10 µg/mL, washed, then subjected to fluorescence activated cell sorter analysis (FACScan, Becton-Dickinson). The L49/P1.17 binding ratios were determined from the ratios of the mean fluorescence intensities after incubation with the respective mAb-fluorescein conjugates. In Vitro Cytotoxicity. A total of 3713 and 3754 lung adenocarcinoma cells were plated into 96-well microtiter plates (104 cells/well in 0.1 mL of IMDM with 10% fetal bovine serum, 0.06 mg/mL penicillin, and 0.1 mg/mL streptomycin) and allowed to adhere overnight. For

Bioconjugate Chem., Vol. 10, No. 6, 1999 1085

blocking experiments, the cells were incubated with unconjugated whole L49 at 1 µM for 30 min prior to treatment with the L49 conjugates. The cells were then exposed to the L49-bL conjugates at 10 nM. After 30 min at 4 °C, the plates were washed three times with antibiotic free RPMI 1640 medium with 10% fetal bovine serum, and then various amounts of CCM were added. CCM and PDM were also added to cells that were not treated with the conjugates. After 1 h at 37 °C, the cells were washed with IMDM and incubated for approximately 18 h. The cells were then pulsed for 12 h with [3H]thymidine (1 µCi/well) at 37 °C, detached by freezing and thawing, and were harvested onto glass fiber filter mats using a 96-well cell harvester. Radioactivity was counted using a β-plate counter. In Vivo Therapy Experiments. Athymic nude mice from Harlan Sprague/Dawley (10 weeks old) were implanted with subcutaneous 3713 or 3754 tumors, and therapy experiments started 35 days later when tumors reached approximately 120 mm3. Tumors from in vivo passages 3-8 were used. Groups of five mice were injected with L49-sFv-bL iv at 1 mg/kg (10 mL/kg in PBS). This was followed 15-18 h later by iv CCM treatment at various concentrations (10 mL/kg in 0.12 M sodium bicarbonate with 10% DMSO). Control mice were either untreated, or received iv PDM (10 mL/kg in saline) at the doses specified in the figures. This was repeated weekly for a total of three rounds. Tumor volume was computed using the following formula: 1/2[longest dimension × perpendicular dimension2]. Mice were removed from the study before their tumor volumes reached 2000 mm3, at which point the average tumor sizes from the remaining mice were no longer plotted. Maximum tolerated doses led to less than 20% sustained (g2 weeks) weight loss, led to no treatment related deaths, and were within 50% of the dose where such events took place. For the renal cell carcinoma experiments, mice (five per group) bearing subcutaneous 1934J or SN12P tumors were treated with 1.4 mg/kg L49-Fab′-bL or with 1.0 mg/ kg L49-sFv-bL. The prodrug CCM was administered 3 days post-L49-Fab′-bL and 1 day post-L49-sFv-bL. Treatments were administered weekly for a total of three rounds. RESULTS

Expression of the p97 Antigen on Carcinoma Cell Lines. The L49 mAb (IgG1) has previously been shown to bind to the p97 melanotransferrin antigen with subnanomolar affinity (11) Almost all human melanomas tested have been reported to be strongly p97 positive (14). Antibodies against the p97 antigen also react with some carcinomas, although the antigen tends to be expressed at lower levels compared to melanomas (10, 14). The presence of the p97 antigen on a panel of 13 lung carcinoma cell lines was determined by fluorescenceactivated cell sorter analysis using the L49 mAb conjugated to fluorescein. Binding was compared to that of P1.17-fluorescein, a nonbinding control conjugate. As shown in Table 1, most of the lung cell lines expressed the p97 antigen. Using a ratio of 1.3 as a cutoff for positive binding (10), 10 of the 13 lung carcinoma cell lines were p97 antigen positive. The average binding ratio among the positive cell lines was 3.0 ( 1.4, indicating generally low levels of p97 antigen expression. For comparison, the binding ratios for renal cell carcinoma and melanoma cell lines are shown, since L49-bL/CCM combinations have demonstrated pronounced in vivo

1086 Bioconjugate Chem., Vol. 10, No. 6, 1999

Kerr et al.

Table 1. L49 Binding to Cell Lines Relative to the Nonbinding P1.17 Control cell line

tumor type

binding ratioa

2981 2987 3697 3713 3754 3766 3776 3896 3963 4013 4015 4023 4026 3677 SN12P 1934J

lung adenocarcinoma lung adenocarcinoma lung adenocarcinoma lung adenocarcinoma lung adenocarcinoma lung adenocarcinoma lung adenocarcinoma lung adenocarcinoma lung adenocarcinoma lung adenocarcinoma lung adenocarcinoma lung adenocarcinoma lung adenocarcinoma melanoma renal cell carcinoma renal cell carcinoma

1.5 1.0 1.1 4.9 4.4 1.0 1.5 1.6 3.5 2.2 2.7 5.0 2.5 4.2 35b 1.8b

a Cells were incubated with L49-fluorescein or P1.17-fluorescein at 10 µg/mL, washed, and analyzed on a fluorescence activated cell sorter. Binding ratios are the ratios of mean fluorescence intensities of L49-fluorescein to P1.17-fluorescein. A ratio of greater than 1.3 is considered p97 antigen positive. b From ref 10.

activities in these tumor models (8-11). A total of 7 of 13 of the lung carcinomas express relatively higher levels of the p97 antigen than the 1934J renal cell carcinoma, a tumor that underwent regression upon treatment with L49-sFv-bL/CCM. In Vitro Studies. The L49-Fab′-bL conjugate was prepared according to previously described methods (10) by combining sulfhydryl-containing L49 Fab′, derived from reduction of L49 Fab′2, with maleimide-substituted bL. After purification, SDS-PAGE analysis revealed a major protein band at approximately 90 kDa, and several minor bands at higher and lower molecular masses (10). The L49-sFv-bL fusion protein was expressed in E. coli and purified by affinity chromatography (11). In contrast to the chemically formed conjugate, the fusion protein was homogeneous according to SDS-PAGE analysis, and had a molecular mass of approximately 63 kDa (11). Both of these conjugates retained the enzymatic and binding activities of the proteins from which they were derived (10, 11). The equilibrium dissociation constants (KD) for the L49-Fab′-bL and L49-sFv-bL conjugates were 1.3 and 1.0 nM, respectively (11). The cytotoxic activities of L49-Fab′-bL and L49-sFvbL in combination with CCM were determined on the p97 positive 3713 and 3754 lung adenocarcinoma cell lines. The cells were exposed to the conjugates, washed to remove unbound material, and treated with varying concentrations of CCM. Cytotoxic activity was determined by measuring the incorporation of [3H]thymidine into DNA relative to untreated cells. On 3713 cells, PDM was approximately 40 times more cytotoxic than the prodrug CCM (Figure 2, panels A and B). The L49-Fab′bL and L49-sFv-bL conjugates both effectively activated the prodrug, leading to cytotoxicities that were equivalent to PDM. Prodrug activation by the conjugates was immunologically specific, since much lower levels of cytotoxic activity was obtained on cells that were saturated with unconjugated L49 prior to treatment with the conjugate/CCM combinations. Similar trends were obtained with the 3754 cells, but the differences in CCM and PDM cytotoxicity were only about 10-fold (Figure 2, panels C and D). These in vitro experiments demonstrate that L49-Fab′-bL and L49-sFv-bL bind equally well to

the p97 antigen and are indistinguishable in their abilities to induce CCM activation on antigen positive cells. In Vivo Therapeutic Activity. The relative antitumor activities of the L49-Fab′-bL and L49-sFv-bL conjugates in combination with CCM were compared in two models of human renal cell carcinoma. The tumors were implanted subcutaneously, and treatment was initiated at a point when they were well established (approximately 200 mm3) and growing. The conjugates were injected iv at 1.0 mg/kg for L49-sFv-bL and 1.4 mg/ kg for L49-Fab′-bL, and CCM was injected iv 24 and 72 h post-L49-sFv-bL and L49-Fab′-bL administration, respectively. These doses and schedules were considered nearly optimal, based on previously published in vivo efficacy and pharmacokinetic data (7-13), and led to comparable intratumoral conjugate levels (11). The intervals between conjugates and prodrug administrations were chosen since they allow sufficient time for the conjugates to localize in tumors and clear from the systemic circulation. This treatment protocol was repeated weekly for a total of three rounds of therapy. The therapeutic effects were compared to those of PDM at its maximum tolerated dose. Treatment with either L49-sFv-bL (Figure 3A) or L49Fab′-bL (Figure 3B) led to cures of all the SN12P tumors when combined with maximum tolerated doses (defined in Materials and Methods) of CCM. Most likely, the reason that more CCM could be administered to L49sFv-bL (240 mg/kg/injection)- compared to L49-Fab′-bL (180 mg/kg/injection)-treated mice was that at the time of prodrug administration there were lower concentrations of the fusion protein compared to the chemical conjugate in the serum (11). Differences between the conjugates were evident when CCM was injected at levels significantly below the respective maximum tolerated doses. The antitumor effects of the L49-sFv-bL/CCM combination were maintained at 1/4 the maximum tolerated dose of CCM (60 mg/kg/injection), while this amount of CCM in L49-Fab′-bL treated mice (1/3 the maximum tolerated dose) resulted only in a delay of tumor outgrowth. PDM at the maximum tolerated dose had no antitumor activity. Similar experiments were conducted in mice with 1934J renal cell carcinoma xenografts (Figure 3, panels C and D). This tumor model has previously been shown to have 1/4 the total p97 antigens/cell compared to SN12P, and accumulation of L49-Fab′-bL very poor (10). L49-sFv-bL treatment led to slightly better therapeutic efficacy than L49-Fab′-bL when each was combined with CCM. Cures were obtained at both 180 mg/kg CCM (two of five animals) and 120 mg/kg CCM (one of five animals) in the L49-sFv-bL treated mice, while all of the tumors in mice receiving L49-Fab′-bL/CCM eventually progressed. Again, PDM had no therapeutic activity. The in vivo therapy results suggest that L49-sFv-bL effects a higher degree of intratumoral CCM activation compared to the chemical conjugate. To extend its utility, we explored the activities of L49-sFv-bL/CCM combinations in mice with p97 positive lung adenocarcinomas. Complete regressions and long-term growth delays were obtained in L49-sFv-bL/CCM treated animals bearing subcutaneous 3713 (Figure 4A) and 3754 (Figure 4B) tumors. The dosing regimens were well below the maximum tolerated doses, since there were no signs of toxicity or weight loss during and after treatment. The 3713 tumor model was insensitive to treatment with the maximum tolerated dose of PDM (3 mg/kg/injection), while 3754 tumor growth (maximum tolerated dose of

Comparison of Monoclonal Antibody-β-Lactamase Conjugates

Bioconjugate Chem., Vol. 10, No. 6, 1999 1087

Figure 2. Cytotoxic effects of mAb-bL + CCM combinations on 3713 (A and B) 3754 and (C and D) human lung carcinomas. The cells were incubated with the mAb-bL conjugates, washed, and treated with CCM for 1 h. The effects were compared to cells treated with CCM or PDM for 1 h without prior conjugate exposure and to cells that were treated with saturating amounts of unconjugated L49 prior to conjugate treatment. The samples were run in triplicate, and the standard deviations were less than 10%.

PDM was 4 mg/kg/injection) was stable for 8 weeks before eventually growing out. The basis for this difference is unclear, but may be due to increased tolerance of PDM in mice with 3754 compared to 3713 tumor implants. These results provide the first evidence that L49-sFvbL in combination with CCM may be efficacious for treating lung adenocarcinomas. DISCUSSION

A number of recombinant mAb-enzyme fusion proteins are being explored and developed as tools in diagnostics and therapy. In contrast to nearly all chemically formed conjugates, such fusion proteins are molecularly defined, and after purification can consist of single molecular entities. An additional advantage in using recombinant technology for fusion protein production is the ease with which specific amino acid substitutions can be introduced in order to customize the proteins for their intended applications. A few examples of mAbenzyme fusion proteins include mAb-alkaline phosphatase conjugates for immunoassays (22), mAb-toxin (23-25), and mAb-ribonuclease (26) conjugates for cancer and mAb-tissue plasminogen activator conjugates for dissolving blood clots (27). There are also several examples of recombinant mAb-enzyme fusion proteins for the activation of anticancer prodrugs. The first example, a mAb-β-glucuronidase conjugate, was reported by Bosslet and co-workers (15). A related molecule has since been reported by Haisma and co-workers that recognizes the CD20 antigen on lymphomas (16). Carter and co-workers have reported the development of a disulfide stabilized anti-p185HER2-Fv-E. coli bL conjugate for the activation of paclitaxel and doxorubicin prodrugs (17, 18). Our work with recombinant mAb-

enzyme fusion proteins began with L6-Bacillus cereus bL for the activation of a doxorubicin prodrug (19) and has since shifted toward the construction of the L49-sFv-bL conjugate described here and elsewhere (11). The particular bL used was from Enterobacter cloacae but incorporated some mutations in one of the loops to enhance its ability to hydrolyze the β-lactams of nitrocefin and a doxorubicin prodrug (21). The L49 mAb was selected because of its ability to bind with high affinity to human melanomas and carcinomas (10, 11). Prior to the studies reported here, there have been no published comparisons of chemical conjugates and recombinant fusion proteins for prodrug activation. Results from the in vitro studies shown in Figure 2 fail to distinguish between L49-Fab′-bL and L49-sFv-bL, which is not surprising since they have similar binding and enzymatic characteristics (11). In contrast, the in vivo studies show that L49-sFv-bL leads to better therapeutic efficacy compared to L49-Fab′-bL when each conjugate was combined with suboptimal doses of CCM (Figure 3). This is most likely due to differences in conjugate biodistribution at the time of prodrug administration. A detailed pharmacokinetic study in mice with subcutaneous melanoma tumors demonstrated that L49-sFv-bL treatment resulted in a tumor/blood ratio of 105 within 24 h of administration (11). Nontarget tissues took up very small amounts of the fusion protein. In contrast, specific intratumoral uptake of the L49-Fab′-bL chemical conjugate was not nearly as pronounced. L49-Fab′-bL cleared quite slowly from the blood, and reached a 5.6 tumor/blood ratio 72 h post conjugate administration. From these data, it can be predicted that fusion protein treatment leads to higher intratumoral drug concentrations compared to the chemically formed conjugate, since

1088 Bioconjugate Chem., Vol. 10, No. 6, 1999

Kerr et al.

Figure 3. Therapeutic effects of L49-bL/CCM combinations in nude mice with SN12P (A and B) and 1934J (C and D) subcutaneous tumor xenografts. The L49-sFv-bL and L49-Fab′-bL conjugates were injected intravenously at 1.0 and 1.4 mg/kg/injection, respectively. CCM was administered intravenously 24 h after L49-sFv-bL and 3 days after L49-Fab′-bL on the days indicated by the arrows. The effects were compared to those of the maximum tolerated dose of PDM, which was administered on the days indicated by the arrows.

Figure 4. Therapeutic effects of L49-sFv-bL/CCM combinations in nude mice with (A) 3713 and (B) 3754 lung adenocarcinoma tumors. L49-sFv-bL was injected intravenously (1.0 mg/kg/injection). CCM, or PDM in previously untreated mice, was administered at the maximum tolerated dose intravenously 24 h later on the days indicated by the arrows.

more of the prodrug is activated in the tumor rather than in the blood. This might account for the improved activities of the fusion protein when the prodrug is administered well below the maximum tolerated dose. As with other mAb-enzyme/prodrug combinations (7, 28-30), the intratumoral drug concentrations obtained are likely to be higher than that achieved from systemic drug administration. In conclusion, the L49-sFv-bL fusion protein presents advantages over the chemically produced conjugate since it is more homogeneous in composition and leads to better therapeutic efficacy. It is significant that pronounced

effects are obtained in lung tumor models that have low levels of antigen expression. The activities of the L49sFv-bL/prodrug combinations in models of human melanoma (8, 11), renal cell carcinoma (10; Figure 3), and lung adenocarcinoma (Figure 4) provide a compelling basis for clinical development. ACKNOWLEDGMENT

We wish to thank Karl Erik and Ingegerd Hellstro¨m for their support throughout the course of this research and Joseph Francisco for his useful comments on the manuscript.

Comparison of Monoclonal Antibody-β-Lactamase Conjugates LITERATURE CITED (1) Senter, P. D., and Svensson, H. P. (1996) A summary of monoclonal antibody-enzyme/prodrug combinations. Adv. Drug Delivery Rev. 22, 341-349. (2) Niculescu-Duvaz, I., and Springer, C. J. (1995) Antibodydirected enzyme prodrug therapy (ADEPT): A targeting strategy in cancer chemotherapy. Curr. Med. Chem. 2, 687706. (3) Melton, R. G., and Sherwood, R. F. (1996) Antibody-enzyme conjugates for cancer therapy. J. Natl. Cancer Inst. 88, 153165. (4) Bagshawe, K. D. and Begent, R. H. J. (1996) First clinical experiment with ADEPT. Adv. Drug Delivery Rev. 22, 365367. (5) 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-β-lactamase conjugates. Bioconjugate Chem. 4, 334340. (6) Vrudhula, V. M., Svensson, H. P., and Senter, P. D. (1995) Cephalosporin derivatives of doxorubicin as prodrugs for activation by monoclonal antibody-β-lactamase conjugates. J. Med. Chem. 38, 1380-1385. (7) Svensson, H. P., Vrudhula, V. M., Emsweiler, J. E., MacMaster, J. F., Cosand, W. L., Senter, P. D., and Wallace, P. M. (1995) In vitro and in vivo activities of a doxorubicin prodrug in combination with monoclonal antibody β-lactamase conjugates. Cancer Res. 55, 2357-2365. (8) Kerr, D. K., Li, Z., Siemers, N. O., Senter, P. D., and Vrudhula, V. M. (1998) Development and activities of a new melphalan prodrug designed for tumor-selective activation. Bioconjugate Chem. 9, 255-259. (9) Kerr, D. E., Schreiber, G. J., Vrudhula, V. M., Svensson, H. P., Hellstrom I., Hellstrom, K. E., and Senter, P. D. (1995) Regressions and cures of melanoma xenografts following treatment with monoclonal antibody β-lactamase conjugates in combination with anticancer prodrugs. Cancer Res. 55, 3558-3563. (10) Svensson, H. P., Frank, I. S., Berry, K. K., Vrudhula, V. M., and Senter, P. D. (1998) Therapeutic effects of monoclonal antibody-β-lactamase conjugates in combination with a nitrogen mustard anticancer prodrug in models of human renal cell carcinoma. J. Med. Chem. 41, 1507-1512. (11) Siemers, N. O., Kerr, D. E., Yarnold, S., Stebbins, M., Vrudhula, V. M., Hellstro¨m, I., Hellstro¨m, K. E., and Senter, P. D. (1997) Construction, expression, and activities of L49sFv-β-lactamase, a single chain antibody fusion protein for anticancer prodrug activation. Bioconjugate Chem. 8, 510519. (12) 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: regressions and long-term growth inhibition of human colon carcinoma xenograft models. Cancer Res. 53, 3956-3963. (13) Meyer, D. L., Law, L. L., Payne, J. K., Mikolajczyk, S. D., Zarrinmayeh, H., Jungheim, L. N., Kling, J. K., Shepherd, T. A., and Starling, J. J. (1995). Site-specific prodrug activation by antibody-β-lactamase conjugates: preclinical investigation of the efficacy and toxicity of doxorubicin delivered by antibody directed catalysis. Bioconjugate Chem. 6, 440446. (14) Woodbury, R. G., Brown, J. P., Yeh, M. Y., Hellstro¨m I., and Hellstro¨m K. E. (1980) Identification of a cell surface protein, p97, in human melanomas and certain other neoplasms. Proc. Natl. Acad. Sci. U.S.A. 77, 2183-2187. (15) Bosslet, K., Czech, J., Lorenz, P., Sedlacek, H. H., Schuermann, M., and Seemann, G. (1992) Molecular and functional characterisation of a fusion protein suited for tumour specific prodrug activation. Br. J. Cancer 65, 234-238.

Bioconjugate Chem., Vol. 10, No. 6, 1999 1089 (16) Haisma, H. J., Sernee, M. F., Hooijberg, E., Brakenhoff, R. H., Meulen-Muileman, I. H., and Pinedo, H. M. (1998) Construction and characterization of a fusion protein of single-chain anti-CD20 antibody and human beta-glucuronidase for antibody-directed enzyme prodrug therapy. Blood 92, 184-190. (17) Rodrigues, M. L., Presta, L. G., Kotts, C. E., Wirth, C., Mordenti, J., Osaka, G., Wong, W. L., Nuijens, A., Blackburn, B., and Carter, P. (1995) Development of a humanized disulfide-stabilized anti-p185HER2 Fv-beta-lactamase fusion protein for activation of a cephalosporin doxorubicin prodrug. Cancer Res. 55, 63-70. (18) Rodrigues, M. L., Carter, P., Wirth, C., Mullins, S., Lee, A., and Blackburn, B. K. (1995) Synthesis and β-lactamasemediated activation of a cephalosporin-taxol prodrug. Chem. Biol. 2, 223-227. (19) Goshorn, S. C., Svensson, H. P., Kerr, D. E., Sommerville, J. E., Senter, P. D., and Fell, H. P. (1993) Genetic construction, expression, and characterization of a single-chain anticarcinoma antibody fused to β-lactamase. Cancer Res. 53, 2123-2127. (20) Svensson, H. P., Wallace, P. M., and Senter, P. D. (1994) Synthesis and characterization of monoclonal antibody-βlactamase conjugates. Bioconjugate Chem. 5, 262-267. (21) Siemers, N. O., Yelton, D. E., Bajorath, J., and Senter, P. D. (1996) Modifying the specificity and activity of the Enterobacter cloacae P99 β-lactamase by mutagenesis within an M13 phage vector. Biochemistry, 35, 2104-2111. (22) Mousli, M., Goyffon, M., and Billiald, P. (1998) Production and characterization of a bivalent single chain Fv/alkaline phosphatase conjugate specfic for the hemocyanin of the scorpion Androctonus australis. Biochim. Biophys. Acta 1425, 348-360. (23) Reiter, Y., and Pastan, I. (1996) Antibody engineering of recombinant Fv immunotoxins for improved targeting of cancer: disulfide-stabilized Fv immunotoxins. Clin. Cancer Res. 2, 245-252. (24) Matthey, B., Engert, A., Klimka, A., Diehl, V., and Barth, S. (1999). A new series of pET-derived vectors for high efficiency expression of pseudomonas exotoxin-based fusion proteins. Gene 229, 145-153. (25) Francisco, J. A., Gawlak, S. L., Miller, M., Bathe, J., Russell, D., Chace, D., Mixan, B., Zhao, L., Fell, H. P., and Siegall, C. B. (1997) Expression and characterization of bryodin 1 and a bryodin 1-based single-chain immunotoxin from tobacco cell culture. Bioconjugate Chem. 8, 708-713. (26) Newton, D. L., Xue, Y., Olson, K. A., Fett, J. W., and Rybak, S. M. (1996) Angiogenin single-chain immunofusions: influence of peptide linkers and spacers between fusion protein domains. Biochemistry 35, 545-553. (27) Runge, M. S., Bode, C., Haber, E., and Quertermous, T. (1991) Hybrid molecules: insights into plasminogen activator function. Mol. Biol. Med. 8, 245-255. (28) Wallace, P. M., MacMaster, J. F., Smith, V. F., Kerr, D. E., Senter, P. D., and Cosand, W. L. (1994) Intratumoral generation of 5-fluorouracil mediated by an antibody-cytosine deaminase conjugate in combination with 5-fluorouracil. Cancer Res. 54, 2719-2723. (29) Aboagye, E. O., Artemov, D., Senter, P. D., and Bhujwalla, Z. M. (1998) Intratumoral conversion of 5-fluorocytosine to 5-fluorouracil by monoclonal antibody-cytosine deaminase conjugates: noninvasive detection of prodrug activation by magnetic resonance spectroscopy and spectroscopic imaging. Cancer Res. 58, 4075-4078. (30) Bosslet, K., Czech, J., and Hoffman, D. (1994) Tumorselective prodrug activation by fusion-protein-mediated catalysis. Cancer Res. 54, 2151-2159.

BC990075W