Highly Specific in Vivo Tumor Targeting by Monovalent and Divalent Forms of 741F8 Anti-c-erbB-2 Single-Chain Fv1

The in vivo properties of monovalent and divalent single-chain Fv (sFv)based molecules with the specificity of the anti-c-erbB-2 monoclonal an tibody 741F8 were examined in seid mice bearing SK-OV-3 tumor xenografts. 741F8 sFv monomers exhibited rapid, biphasic clearance from blood, while a slightly slower clearance was observed with the divalent 741F8 (sFv')2 comprising a pair of 741F8 sFv' with a C-terminal Gly4Cys joined by a disulfide bond. Following i.v. injection, the 741F8 sFv mono mer was selectively retained in c-eréB-2-overexpressing SK-OV-3 tumor, with excellent tumor:normal organ ratios uniformly exceeding 10:1 by 24 h. The specificity of this effect was demonstrated by the lack of retention of the anti-digoxin 26-10 sFv monomer, as evaluated by biodistribution studies, gamma camera imaging, and cryomacroautoradiography studies. The specificity index (741F8 sFv retention/26-10 sFv retention) of 741F8 monomer binding, measured by the percentage of injected dose per g of tissue, was 13.2:1 for tumor, and 0.8 to 2.1 for all tested normal organs, with specificity indices for tumor:organ ratios ranging from 7.0 (kidneys) to 16.7 (intestines). Comparing divalent 741F8 (sFv')2 with the 26-10 isl-V)... similar patterns emerged, with specificity indices for retention in tumor of 16.9 for the Gly4Cys-linked (sFv')2. These data demonstrate that, following their i.v. administration, both monovalent and divalent forms of 741F8 sFv are specifically retained by SK-OV-3 tumors. This antigenspecific binding, in conjunction with the 26-10 sFv controls, precludes the possibility that passive diffusion and pooling in the tumor interstitium contributes significantly to long-term tumor localization. 741F8 (sFv')2 species with peptide spacers exhibited divalent binding and increased retention in tumors as compared with 741F8 sFv monomers. Since the blood retention of the (sFv')2 is slightly more prolonged than that of the monomer, it was necessary to demonstrate that the increased tumor lo calization of the peptide-linked (sFv')2 was due to its divalent nature. The significantly greater localization of the divalent bismalimidohexane-linked 741F8 (sFv')2 as compared with a monovalent 741F8 Fab fragment of approximately the same size suggests that the increased avidity of the (sFv')2 is a factor in its improved tumor retention. This is the first report of successful specific in vivo targeting of tumors by divalent forms of sFv molecules. The improved retention of specific divalent (sFv')2 by tumors may have important consequences for targeted diagnostic or therapeutic strategies. INTRODUCTION The development of murine MAb3 and their proteolytic fragments has resulted in significant interest in antigen-specific targeting of Received 3/26/93: accepted 6/28/93. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 Supported by National Cancer Institute National Cooperative Drug Discovery Group Grants U01 CA51880 and CA06927. an appropriation from the Commonwealth of Penn sylvania, and the Bernard A. and Rebecca S. Bernard Foundation. 2 To whom requests for reprints should be addressed, at Department of Medical Oncology, Fox Chase Cancer Center, 7701 Burholme Avenue, Philadelphia. PA 19111. ' The abbreviations used are: MAb, monoclonal antibody: sFv. single-chain Fv frag ment; MCA, N.^N«bismaleimidocaproyl amino acid; BMH. bismaleimidohexane; ECD, extracellular domain; CHO, Chinese hamster ovary; HMP, hydroxymethylphenoxyacetic acid; DIC, diisopropylcarbodiimide; DMAP, dimethylaminopyridine: TFA. trifluoroacetic acid; HPLC, high-performance liquid chromatography; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; PBS, phosphate-buffered saline; %ID/g, percent age of injected dose per g of tissue; T:O ratio, tumonnormal organ ratio; IgG, ¡mmunoglobulin G. tumors for diagnostic or therapeutic purposes. However, successful targeting of radionuclides, drugs, or toxins with MAb has been limited by the distribution of these proteins to normal organs (1). Further more, the quantitative delivery of IgO molecules and their fragments to tumors has been quite low, because of such factors as heterogeneity of tumor antigen expression (2) and poor diffusion of MAb from the vasculature into the tumor (3). The recent development of recombi nant sFv molecules has provided agents which may alleviate some of these limitations to the use of MAb in the targeting of neoplasms (4, 5). Radioiodinated sFv administered to tumor-bearing mice ex hibit faster plasma clearance rates and higher tumonnormal tissue ra tios than corresponding IgG or Fab (6, 7). The penetration of sFv into tumor from the vasculature, as determined by microscopic autoradiography, is superior to that of the corresponding IgG, F(ab')2, or Fab' (8). This investigation addresses the relative impact of binding avidity, molecular mass, and rapidity of clearance from the blood pool on sFv targeting of tumors. We have produced tumor-specific sFv species directed against the c-erbB-2 antigen and compared them to control sFv species specific for the drug digoxin. Conventional sFv mole cules, sFv' molecules, and divalent disulfide-bonded (sFv')2 mole cules have been prepared, with the sFv' molecules having carboxylterminal cysteine residues to facilitate site-specific dimerization. We have obtained pharmacokinetic and biodistribution profiles, as well as 7 scintigraphic and macroautoradiographic images of sFv and (sFv')2 administered i.v. to seid mice bearing human tumor xenografts ex pressing c-erbB-2. These results differentiate between the specific binding of 741F8 anti-c-erbB-2 sFv and the nonspecific interstitial pooling of irrelevant 26-10 sFv in tumors and permit the examination of the contributions of sFv valency and size to tumor targeting. MATERIALS AND METHODS In these experiments, results with sFv and sFv' monomers of the same specificity did not significantly differ. Accordingly, for the purpose of clarity, both sFv and sFv' monomers will be referred to as sFv' monomers. Preparation of sFv Molecules. The sFv species utilized in these experi ments were produced at Creative BioMolecules, Inc. (Hopkinton. MA), as described elsewhere (Ref. 9; Footnote 4). All of the sFv and sFv' species were produced in Escherichia coli and were devoid of leaders except for the 741F8 sFv' with a Gly4Cys tail, which had a 9-residue N-terminal extension (ADNKFNKDP) to promote high expression levels. Refolding of the M, 27,000 sFv' analogues used a slight modification of the 3 M urea/glutathione oxidation reduction refolding procedure of Tai et al. (10). Aiiti-digoxin 26-10 and antic-erbB-2 741F8 sFv' species were refolded to yield stable monomers with the C-terminal cysteine in a mixed disulfide with glutathione. Blocked 741F8 sFv' was purified by anion/cation exchange followed by size exclusion chromatog raphy and then converted to (sFv')3 homodimers. Blocked 26-10 sFv' was purified by ouabain-Sepharose affinity chromatography. as described previ ously for 26-10 sFv (4). Monovalent sFv' was deblocked with mild reduction, and (sFv')i dimers were formed through disulfide bonds by oxidation (for fused tails comprising either -SerCys or -Gly4Cys), by MCA-peptide bridging (MCA-GlySer,Gly2Ser3Lys-MCA), or through thioether bonds by BMH 4 M-S. Tai and J. E. McCartney, manuscripts in preparation.