Homogeneous immunoconjugates for boron neutron-capture therapy ...

Proc. Natl. Acad. Sci. USA Vol. 95, pp. 13206–13210, October 1998 Medical Sciences

Homogeneous immunoconjugates for boron neutron-capture therapy: Design, synthesis, and preliminary characterization LUFENG GUAN*, LETITIA A. WIMS†, ROBERT R. KANE*‡, MARK B. SMUCKLER*, SHERIE L. MORRISON†, AND M. FREDERICK HAWTHORNE*§ *Department of Chemistry and Biochemistry, University of California, Los Angeles, CA 90095; and †Department of Microbiology, Immunology, and Molecular Genetics and the Molecular Biology Institute, University of California, Los Angeles, CA 90095

Contributed by M. Frederick Hawthorne, September 2, 1998

by each immunoprotein carrier (assuming '106 targetable antigenic sites per cell). It is widely accepted that the only practical method for achieving this massive boron loading is via the conjugation of a limited number of boron-rich macromolecules to each immunoprotein. Although the most common approach to this problem has been the conjugation of heterogeneous macromolecules to numerous indeterminate lysine residues on the targeting protein (5–8), our group has recently focused on the development of methods for the assembly of completely homogeneous boron-rich immunoconjugates. We have reported on the precise stepwise synthesis of oligomeric boron-rich peptides (9, 10) and oligophosphates (11–14) as reagents for the assembly of homogeneous immunoconjugates. The homogeneous boron-rich oligophosphate reagents are extremely hydrophilic and can be fitted readily with unique reactive sites (including primary amino groups and thiol residues). Experiments directed toward the sitespecific modification of immunoproteins with these reagents were focused initially on their conjugation to thiol residues on F(ab9) fragments derived from intact IgG molecules (15). Unfortunately, both the production and conjugation of F(ab9) immunoprotein fragments have proven to be problematic in our hands, prompting our search for alternative immunoprotein carriers and conjugation methods. Genetic-engineering techniques provide methods for the development of new proteins that exhibit novel structures andyor functions. We recently reported on the generation of anti-dansyl (DNS) IgG immunoproteins with unnatural cysteine residues at positions predicted to be solvent-accessible (16). These engineered immunoproteins are attractive partners for the development of homogeneous antibody conjugates. They retain the structure and targeting ability of intact IgG, and they contain unique reactive residues that are introduced, and can be altered, at the genetic level. Furthermore, they do not require extensive chemical andyor enzymatic manipulation before their site-specific modification with reactive conjugation partners. Herein, we report the results of preliminary studies of the conjugation of boron-rich oligophosphates to an engineered immunoprotein.

ABSTRACT The application of immunoprotein-based targeting strategies to the boron neutron-capture therapy of cancer poses an exceptional challenge, because viable boron neutron-capture therapy by this method will require the efficient delivery of 103 boron-10 atoms by each antigenbinding protein. Our recent investigations in this area have been focused on the development of efficient methods for the assembly of homogeneous immunoprotein conjugates containing the requisite boron load. In this regard, engineered immunoproteins fitted with unique, exposed cysteine residues provide attractive vehicles for site-specific modification. Additionally, homogeneous oligomeric boron-rich phosphodiesters (oligophosphates) have been identified as promising conjugation reagents. The coupling of two such boron-rich oligophosphates to sulfhydryls introduced to the CH2 domain of a chimeric IgG3 has been demonstrated. The resulting boron-rich immunoconjugates are formed efficiently, are readily purified, and have promising in vitro and in vivo characteristics. Encouragingly, these studies showed subtle differences in the properties of the conjugates derived from the two oligophosphate molecules studied, providing a basis for the application of rational design to future work. Such subtle details would not have been as readily discernible in heterogeneous conjugates, thus validating the rigorous experimental design employed here. Boron neutron-capture therapy is a binary approach to cancer therapy based on the selective accumulation of significant concentrations of the stable isotope boron-10 (10B) in a tumor target, followed by the irradiation of the target tissue with thermal neutrons. The boron-neutron reaction is efficient and results in the emission of high energy, cytotoxic particles according to the reaction 10B 1 1n 3 7Li 1 4He 1 2.4 MeV. Because the heavy ions emitted in this reaction have a limited translational path (approximately one cell diameter), boron neutron-capture therapy-mediated cytotoxicity is limited to sites that contain elevated 10B concentrations during irradiation with thermal neutrons. A variety of methods for the selective targeting of 10B to tumors have been proposed, and promising reagents include boron-modified porphyrins, monoclonal antibodies, nucleosides, amino acids, and liposomes (1–3). In 1952, Bale suggested that boron-conjugated antitumor antibodies might afford an effective means for delivering 10B for neutron-capture therapy (4). However, it is clear that the implementation of immunoprotein-based targeting strategies poses a special challenge for boron neutron-capture therapy; a simple calculation, based on the delivery of 109 10B atoms per targeted cell, indicates that '103 10B atoms must be delivered

MATERIALS AND METHODS Preparation of IgG3L309C. IgG3L309C (a chimeric antiDNS IgG3 with an exposed cysteine introduced into CH2 by the point mutation Leu-309 to Cys) was purified by DNScoupled affinity chromatography as reported (16). The concentrations of purified antibodies were determined by a bicinchoninic-acid-based protein assay (Pierce). Abbreviations: DNS, dansyl; MBS, maleimidobenzoic acid Nhydroxysuccinimide ester; SRBC, sheep red blood cell. ‡ Present address: Department of Chemistry, Baylor University, Waco, TX 76798. §To whom reprint requests should be addressed. e-mail: mfh@chem. ucla.edu.

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked ‘‘advertisement’’ in accordance with 18 U.S.C. §1734 solely to indicate this fact. © 1998 by The National Academy of Sciences 0027-8424y98y9513206-5$2.00y0 PNAS is available online at www.pnas.org.

13206

Medical Sciences: Guan et al. Preparation of the Oligomeric nido-Carboranyl Phosphodiesters. The pentameric nido-carboranyl phosphodiesters (CB)5 and (G4)5 were synthesized on automated DNA-synthesis instruments as described (12, 13) with no modification to the standard reagents or procedures and purified by HPLC. The purity of the oligomers was analyzed by PAGE (28.5% polyacrylamide gelsy3.5 M urea; ref. 17). Preparation and Characterization of (CB)5- and (G4)5IgG3L309C Conjugates. IgG3L309C (300 mg) at the concentration of 3 mgyml in phosphate buffer, pH 7.4, was treated with 100 ml of 500 mM cysteiney50 mM ammonium citrate buffer, pH 7.5y2 mM EDTAy100 mM NaCl at 37°C for 4 h. The (CB)5 and (G4)5 oligomers were also dissolved in 50 mM ammonium citrate buffer, pH 7.5y2 mM EDTAy100 mM NaCl to a final concentration of 2 mM. Bifunctional linker maleimidobenzoic acid N-hydroxysuccinimide ester (MBS; Pierce) was dissolved in ice-cold dimethylformamide to a final concentration of 200 mM. MBS solution (100 ml) and 100 ml of either (CB)5 solution or (G4)5 solution were added to a 1-ml screw-cap vial. The mixture was incubated at room temperature for 4 h with constant stirring. The excess unreacted MBS was removed by ten 300-ml extractions with water-saturated ethyl acetate. The aqueous phase, containing MBS-activated (CB)5 or (G4)5, was added to the cysteine-reduced antibody. The mixture was incubated at 37°C for 4 h. The antibody conjugates were purified by CENTRI-SEP column (Princeton Separations, Adelphia, NJ) and analyzed by SDSyPAGE (5% sodium phosphate-buffered gels for intact proteins and 12% Trisyglycine-buffered gels for proteins treated to break the interchain disulfide bonds). Gels were stained by either Coomassie blue or silver nitrate (18). The boron content of the antibody conjugates was determined by inductively coupled plasma–atomic emission spectroscopy. Radioiodination of the Antibody Conjugates. The chimeric wild-type IgG3 was labeled with 125I with Iodobeads (Pierce) to a specific activity between 0.2 and 0.3 mCiymg for the Fc receptor-binding assay. For in vivo half-life studies, either (CB)5- or (G4)5-conjugated IgG3L309C or the wild-type IgG3 was labeled with the 125I-labeled Bolton–Hunter reagent (Amersham, DuPont NEN, or ICN) in 0.1 M sodium borate buffery1 M glycine, pH 8.5. The antibody conjugates were purified by size-exclusion chromatography over a 10-ml column of Sephadex G-10. The column was equilibrated and washed with 1% BSA in PBS, pH 7.4, or 0.1% gelatin in 0.1 M sodium borate buffer, pH 8.5. Antigen-Binding Assay. Flat-bottomed microtiter plates (Immulon 2, Dynatech) were coated with 100 ml dansylated BSA in PBS, pH 7.4, and blocked with 3% BSA. Antibody conjugate in PBS with 1% BSA (100 ml) was added to each well of the plate, followed by overnight incubation of the plate at 4°C. After the plate was washed with PBS, 100 ml of goat anti-human IgG kappa-conjugated with alkaline phosphatase in 1% BSA was added to each well. After a 1-h incubation at 37°C, the plate was washed again with PBS. The substrate p-nitrophenyl phosphate (Sigma) was then added (0.6 mgyml in 9.6% diethanolaminey0.24 mM MgCl2, pH 9.8), and the reaction was allowed to proceed at room temperature for 10–30 min, after which the plate was read at 410 nm in an ELISA plate reader. Isotype-Specific ELISA. The isotype-specific ELISA was done as described above with the following modifications; (i) 100 ml of the isotype-specific monoclonal antibody 6050 (specific for the hinge of IgG3; ref. 19) diluted to 1:1000 in PBS with 1% BSA was used as the primary antibody, and (ii) biotinylated anti-mouse IgG (diluted to 1:8000 in PBS with 0.5% Tween) and streptavidin-conjugated alkaline phosphatase were used to detect bound murine IgG. Fc Receptor-Binding Assay. U937 cells (2–8 3 106) were stimulated with 100 unitsyml g-interferon at 105 cells per ml for 2 days. The U937 cells were then incubated with 105–106

Proc. Natl. Acad. Sci. USA 95 (1998)

13207

cells per ml 125I-labeled wild-type antibody and various amounts of unlabeled competitor at 15°C for 3 h in 0.95 ml of binding buffer (0.1 M Hepes free acidy0.12 M NaCly5 mM KCly1.2 mM MgSO4y15 mM HOAcy10 mM glucosey1% BSA, pH 7.4–7.7). The suspension of cells was then layered over 300 ml of binding oil (high temperature silicone oil from Aldrich and paraffin oil from Fisher Scientific at a ratio of 84:16, respectively) in a 1.5-ml microcentrifuge tube and centrifuged at 13,000 rpm (Biofuge 15, Baxter Scientific Products, McGraw Park, IL) for 1 min. The tube was frozen in a dry-iceyethanol bath, and the bottom was cut off for counting. The total radioactivity added to the tube was determined by counting an aliquot of the radioactive protein. In Vivo Half-Life and Biodistribution Studies. Mice were given a 0.1-mgyml solution of KI in place of water for 7 days and then injected i.p. in sets of five with 5 3 106–107 cells per milliliter of 125I-labeled antibody conjugates. For the half-life study, mice were counted in a whole-body counter (William B. Johnson & Associates, Ronceverte, WV) over the course of 500 h maintaining the 0.1-mgyml solution of KI throughout. For the biodistribution study, five mice per protein were used. One mouse of each set of five was killed at 24 h, and a second was killed at 72 h; the organs were recovered, and the residual radioactivity in each was determined. Direct Lysis Assay. Sheep red blood cells (SRBC) were coated with DNS-BSA (0.25 mg/ml DNS-BSAy5% SRBCy150 mM NaCly0.25 mM CrCl3, pH 7.0) for 1 h at 30°C, and then 0.2 ml of packed, antigen-coated SRBCs were loaded with sodium[51Cr]chromate (Amersham) in 2 ml of fresh gelatiney Hepes buffered saline (GelyHBS; 0.01 M Hepes free acidy0.15 M NaCly0.5 mM MgCl2y0.15 mM CaCl2y0.1% gelatin, pH 7.4) at 37°C for 1 h. The free sodium[51Cr]chromate was removed by washing the cells three times in 10 ml of fresh Gel-HBS. Various concentrations of antibody conjugates in 50-ml aliquots of Gel-HBS were added to round-bottomed, 96-well plates (Corning). Then 50 ml of 2% 51Cr-loaded, DNS-BSAcoated SRBC and 25 ml of undiluted normal human serum preabsorbed against DNS-BSA-coated cells were added to each well as the source of complement. The plates were incubated at 37°C for 45 min. Unlysed SRBC were pelleted by centrifugation (IEC, Damon Biotechnology, Needham, MA) of the plate, and 50 ml of supernatant was counted in a g counter. Each sample was assayed in triplicate, and the percentage of lysed cells was calculated.

RESULTS AND DISCUSSION The immunoprotein used in these studies is a genetically engineered chimeric anti-DNS IgG containing a Leu 3 Cys point mutation of residue 309 in CH2 (16). This mutant was designed to provide two unique reactive thiols for conjugation that would be both solvent-accessible and removed from the antigen-binding site. This protein exists primarily as a series of oligomeric forms that can be converted entirely to monomers by mild treatment with cysteine (30–40 molar equivalents, 37°C for 4 h, pH 7.5). Incomplete reduction was seen when the antibody concentration exceeded 4 mgyml. Two alternative reducing agents, tris(2-carboxyethyl) phosphinezhydrochloride and 2-mercaptoethanol, generally overreduced the antibody to sub-IgG-sized fragments. Although excess cysteine could be removed by gel filtration of the reduced antibody, this procedure resulted in a significant reduction in protein concentration caused by dilution of the sample as well as the loss andyor reoxidation of a significant fraction of the antibody. Accordingly, conjugation of the monomeric immunoprotein with the boron-rich oligophosphates was performed customarily on the crude reduction mixture without removal of excess cysteine. Two structurally variant boron-rich conjugation partners were explored. The first oligophosphate, designated (CB)5, contains disubstituted nido-carborane residues as components

13208

Medical Sciences: Guan et al.

FIG. 1.


Structures of the oligomeric nido-carboranyl phosphodiesters (CB)5 (A) and (G4)5 (B).

of the oligophosphate backbone (Fig. 1). Compounds based on this structure have been used in antibody modification experiments in our laboratory (15). The second boron-rich oligophosphate used in this study, designated (G4)5, contains monosubstituted nido-carboranes attached to the oligophosphate backbone via a short ether-based tether. HPLCretention behavior suggests that this oligomer is slightly more hydrophilic than the (CB)5 oligomer. Both boron-rich oligophosphates contain a thymidine residue at the pseudo-39 terminus (which simplifies their construction and provides a UV-active ‘‘tag’’) and a hexylamine residue at the pseudo-59 end. The 59 primary amino group is activated for the thiolselective conjugation reaction by reaction with the Nhydroxysuccinimide functionality of the bifunctional crosslinker MBS. The maleimide residues thus introduced to the boron-rich oligophosphates readily reacted with freshly reduced IgG3L309C immunoprotein (final concentrations of '10 mM protein and 1 mM activated oligophosphate), affording the desired boron-rich immunoconjugates in excellent yield after purification by gel filtration. PAGE showed the almost quantitative formation of homogeneous immunoconjugates (Fig. 2A). Importantly, the immunoconjugates remained intact as fully assembled H2L2 molecules. Evidence of the homogeneity of the conjugates synthesized was provided by PAGE analysis of reduced immunoconjugates. This analysis showed the specificity and completeness of conjugation at the antibody heavy

chain, as would be expected if the engineered thiols efficiently provided the only reactive sites for conjugation (Fig. 2B). The stoichiometry of antibody modification was calculated from each conjugate’s protein and boron concentrations. Both conjugates synthesized exhibited oligophosphateyimmunoprotein ratios very close to the expected stoichiometry (2:1 oligophosphate:conjugate; Fig. 2C). The immunoconjugates thus produced were characterized extensively by their in vitro and in vivo properties. To determine their ability to react with antigen, (CB)5- and (G4)5IgG3L309C at varying concentrations were reacted with antigen-coated microtiter plates, and their binding was detected with an anti-k reagent. In this assay, (G4)5-IgG3L309C was equivalent to IgG3 in its reactivity with antigen, whereas (CB)5-IgG3L309C showed impaired interaction with antigen (Fig. 3). It is unclear whether this altered reactivity reflects changes in the antigen-binding site or whether the attached polyanionic boron-rich oligophosphate interferes with access to antigen attached to the charged surface of the microtiter dish. Several assays probed the characteristics of the constant regions of the conjugated antibodies. Both (CB)5- and (G4)5IgG3L309C showed strong reactivity with an isotype-specific monoclonal antibody that recognizes an epitope in the hinge region (ref. 19; data not shown), showing that conjugation did not lead to significant perturbation of the antibody structure. In marked contrast, both conjugated antibodies were impaired

FIG. 2. SDSyPAGE analysis of IgG3L309C and IgG3L309C conjugates. (A) Unreduced protein analyzed on a 5% phosphate gel. (B) Reduced proteins analyzed on a 12.5% Trisyglycine gel. Lanes 1 and 4, Wild-type IgG3. Lane 2, (CB)5-IgG3L309C. Lane 3, (G4)5-IgG3L309C. (C) Boron and protein concentrations of conjugates and number of boron-rich oligophosphate units per specified conjugate.


FIG. 3. Antigen-binding assay of the wild-type IgG3, (CB)5IgG3L309C, and (G4)5-IgG3L309C. The wild-type IgG3 and the boron immunoconjugates at varying concentrations were bound to DNS-BSA-coated plates, and the bound antibodies were detected with an anti-k reagent.

significantly in effector functions that required access to the Fc region. Although the wild-type IgG3 was able to lyse 43% of antigen-coated SRBC at an antibody concentration of 1.25 mgyml, the (CB)5-IgG3L309C conjugate resulted in only 4% lysis at the same concentration; addition of (G4)5-IgG3L309C did not lead to any lysis. Although both antibodies retained reactivity with the high-affinity Fc receptor, FcgRI (CD64), both showed impaired ability to bind (Fig. 4). It is not surprising that the IgG3L309C immunoprotein shows impaired ability to activate complement and bind FcgRI after conjugation with large molecules. Complement activation requires C1q, a large macromolecule, to bind to an exposed surface in the middle of CH2, which is in the vicinity of residue 309, the conjugation site (Fig. 5; refs. 20–23). Accordingly, conjugation with a large boron-rich molecule most likely inhibits the access of C1q. The site of FcgRI binding is also in CH2, but it has been mapped to a more hinge-proximal location farther from the site of conjugation (24–26). Therefore, although the presence of the boron-rich oligophosphate hinders access to the binding site, binding can still take place with reduced affinity. This continued access to the hinge-proximal portion of CH2 is consistent with our observation that the recognition of (CB) 5- and (G4)5IgG3L309C by antibodies binding in the hinge region is essentially unchanged. If the boron-rich antibody conjugates are to be used for in vivo therapy, it is important that the attached boron does not

F IG . 4. Fc g R I-receptor binding. Wild-type IgG3, (CB) 5 IgG3L309C, and (G4)5-IgG3L309C at varying concentrations were used to inhibit the binding of 125I-wild-type IgG3 to FcgRI, present on interferon-g-stimulated U937 cells.


13209

FIG. 5. Three-dimensional structure of the Fc of human IgG1. Leu-309, which was mutated to Cys in IgG3L309C, is indicated. Glu-318, Lys-320, Lys-322, and Pro-331, which have been implicated in forming the C1 binding site, are shown as space-filling molecules. Pro-331 has also been implicated in forming the FcgRI-binding site, in addition to residues in the hinge-proximal region that were not resolved in the crystal structure. The graphic is based on the coordinates from Diesenhofer (23) and was produced with the program RASMOL MOLECULAR RENDERER (version 2.6; University of Edinburgh and Glaxo).

drastically alter the in vivo properties of the antibody. Both a whole-body half-life study and a biodistribution study were used to evaluate the in vivo characteristics of the boron-rich immunoconjugates in mice. The DNS-specificity of the immunoprotein precluded tumor-targeting experiments. For each in vivo experiment, the boron-rich immunoconjugates were radiolabeled by reaction with prelabeled 125I-Bolton–Hunter reagent. For the half-life study, the resulting 125I-labeled (CB)5- and (G4)5-IgG3L309C as well as the wild-type IgG3 were injected in mice, and the percentage of radioactivity remaining was determined at different times with a wholebody g-counter. The results of the whole-body counting are presented as a plot on semilogarithmic axes of percentage cpm remaining vs. time (Fig. 6). All three proteins exhibited similar kinetics of clearance with the wild-type IgG3 and (G4)5IgG3L309C showing b-phase half-lives of '100 h, whereas the b-phase half-life of (CB)5-IgG3L309C was slightly longer, 127 h. The half-life of IgG3 was slightly longer than what had been reported (27) and may reflect the fact that different methods of iodination were used in the two studies. For the biodistribution studies, mice were killed at 24 and 72 h after injection with the labeled immunoconjugates. Selected tissues were harvested, and the radioactivity present was

FIG. 6. Wild-type IgG3, (CB)5-IgG3L309C, and (G4)5-IgG3L309C labeled with 125I by using the Bolton–Hunter reagent were injected i.p. into mice, and the clearance of radioactivity was monitored by whole-body counting of the mice.

13210


Proc. Natl. Acad. Sci. USA 95 (1998) developed here resulted from conjugation reagents containing subtle structural variation. Because of the homogeneity of the immunoconjugates, discernible effects of these different conjugation reagents are reflected in the biological properties of the resulting immunoconjugates. We now are in a position to investigate systematically the in vivo properties of antibody conjugates made with structurally diverse boron oligomers. We thank W. F. Bauer at Idaho National Engineering Laboratory for the inductively coupled plasma–atomic emission spectroscopy boron analysis. This work was supported in part by National Institutes of Health Grants CA 53870, CA16858, and AI29470. 1. 2. 3. 4. 5. 6. 7. 8. 9.

FIG. 7. Mice injected i.p. with 125I-labeled wild-type IgG3, (CB)5IgG3L309C, and (G4)5-IgG3L309C were killed 24 and 72 h after injection. Organs were recovered and weighed, and their content of radioactivity was determined. Only selected organs with the greatest differences between percentages of proteins remaining are shown. The carcass is included for comparison with organs showing accumulation.

determined (Fig. 7). Although the immunoconjugates and the unmodified wild-type IgG3 showed similar kinetics of wholebody clearance, the residual radioactivity at 24 and 72 h showed some differences in organ localization. Both immunoconjugates were present in the liver and kidney at higher concentrations than wild-type IgG3, and both were present in the blood but at lower concentrations than wild-type IgG3. Though (CB)5-IgG3L309C showed increased accumulation in the spleen at 72 h compared with the wild-type antibody, (G4)5-IgG3L309C showed reduced accumulation at that site. It will now be of considerable interest to compare the relative efficiency of the conjugated and nonconjugated proteins in tumor targeting. It should be noted that although there is greater liver uptake of the conjugated antibodies than of the nonconjugated, a significant amount of conjugated antibodies remains in the circulation for an extended period of time, and these antibodies should be available to bind to the tumor. These studies show that it is possible to engineer accessible cysteine residues into antibody molecules that can be used for site-specific conjugation. More specifically, the cysteine residues can be modified efficiently with maleimide-containing reagents, including polyanionic boron-rich macromolecules. The potential exists for the extension of this conjugation chemistry to other moieties fitted with maleimide residues. The fact that the resulting immunoconjugates are homogeneous and readily characterized is important, because in vivo use in humans is the ultimate goal. The immunoconjugates

10. 11. 12. 13. 14. 15. 16. 17.

18. 19. 20. 21. 22. 23. 24. 25. 26. 27.

Hawthorne, M. F. (1993) Angew. Chem. Int. Ed. Engl. 32, 950–984. Barth, R. F., Soloway, A. H. & Fairchild, R. A. (1990) Cancer Res. 50, 1061–1070. Hawthorne, M. F. (1991) Pure Appl. Chem. 24, 327–334. Bale, W. F. (1952) Proc. Natl. Cancer Conf. 2, 967–976. Varadarajan, A. & Hawthorne, M. F. (1991) Bioconjugate Chem. 2, 242–253. Paxton, R. J., Beatty, B. G., Varadarajan, A. & Hawthorne, M. F. (1992) Bioconjugate Chem. 3, 241–247. Ferro, V. A., Morris, J. H. & Stimson, W. H. (1995) Drug Des. Discov. 13, 13–25. Barth, R. F., Adams, D. M., Soloway, A. H., Alam, F. & Darby, M. V. (1994) Bioconjugate Chem. 5, 58–66. Varadarajan, A. & Hawthorne, M. F. (1991) Bioconjugate Chem. 2, 242–253. Kane, R. R., Pak, R. H. & Hawthorne, M. F. (1993) J. Org. Chem. 58, 991–992. Kane, R. R., Lee, C. S., Drechsel, K. & Hawthorne, M. F. (1993) J. Org. Chem. 58, 3227–3228. Kane, R. R., Drechsel, K. & Hawthorne, M. F. (1993) J. Am. Chem. Soc. 115, 8853–8854. Drechsel, K., Lee, C. S., Leung, E. W., Kane, R. R. & Hawthorne, M. F. (1994) Tetrahedron Lett. 35, 6217–6220. Kim, Y. S., Kane, R. R., Beno, C. L., Romano, S., Mendez, G. & Hawthorne, M. F. (1995) Tetrahedron Lett. 36, 5147–5150. Chen, C.-J., Kane, R. R., Primus, F. J., Szalai, G., Hawthorne, M. F. & Shively, J. E. (1994) Bioconjugate Chem. 5, 557–564. Smith, R. I. F. & Morrison, S. L. (1994) Biotechnology 12, 683–689. Kane, R. R., Kim, Y. S. & Hawthorne, M. F. (1997) in Chemistry and Biology, Advances in Neutron Capture Therapy, eds. Larsson, B., Crawford, J. & Weinreich, R. (Elsevier, New York), Vol. 2, pp. 113–118. Palfner, K. (1995) BioTechniques 19, 926–929. Hamilton, R. G. & Morrison, S. L. (1993) J. Immunol. Methods 158, 107–122. Duncan, A. R. & Winter, G. (1988) Nature (London) 332, 738–740. Tao, M.-H., Canfield, S. M. & Morrison, S. L. (1991) J. Exp. Med. 173, 1025–1028. Tao, M.-H., Smith, R. I. F. & Morrison, S. L. (1993) J. Exp. Med. 178, 661–667. Deisenhofer, J. (1981) Biochemistry 20, 2361–2370. Duncan, A. R., Woof, J. M., Partridge, L. J., Burton, D. R. & Winter, G. (1988) Nature (London) 332, 738–740. Canfield, S. M. & Morrison, S. L. (1991) J. Exp. Med. 173, 1483–1491. Chappel, M. S., Isenman, D. E., Everett, M., Xu, Y. Y., Dorrington, K. J. & Klein M. H. (1991) Proc. Natl. Acad. Sci. USA 88, 9036–9040. Zuckier, L. S., Georgescu, L., Chang, C. J., Scharff, M. D. & Morrison, S. L. (1994) Cancer 73, Suppl., 794–799.