125I-labeled Anti-Epidermal Growth Factor Receptor-vIII Single- Chain ...

Vol. 5, 1539 –1549, June 1999

Clinical Cancer Research 1539

125

I-labeled Anti-Epidermal Growth Factor Receptor-vIII SingleChain Fv Exhibits Specific and High-Level Targeting of Glioma Xenografts1 Chien-Tsun Kuan, Craig J. Reist, Catherine F. Foulon, Ian A. J. Lorimer, Gary Archer, Charles N. Pegram, Ira Pastan, Michael R. Zalutsky, and Darell D. Bigner2 Departments of Pathology [C-T. K., G. A., C. N. P., M. R. Z., D. D. B.] and Radiology [C. J. R., C. F. F., M. R. Z.], Duke University Medical Center, Durham, North Carolina 27710; Ottawa Regional Cancer Centre, Cancer Research Group, Ottawa, Ontario, Canada K1H 8L6 [I. A. J. L.]; and Laboratory of Molecular Biology, National Cancer Institute, NIH, Bethesda, Maryland 20892 [I. P.]

ABSTRACT A single-chain antibody fragment, MR1(scFv), with specific binding to epidermal growth factor receptor-vIII (EGFRvIII), was produced, radiolabeled, and evaluated for biodistribution in human glioma-bearing athymic mice. The mutant receptor EGFRvIII has a deletion in its extracellular domain that results in the formation of a new, tumor-specific antigen found in glioblastomas, breast carcinomas, and other tumors. The scFv molecule, designed as VH-(Gly4Ser)3-VL, was expressed in Escherichia coli in inclusion body form; recovered scFv fragments were properly refolded in redox-shuffling buffer. Size-exclusion chromatography of purified scFv demonstrated a protein monomer of Mr 26,000. Labeling was performed using N-succinimidyl 5-[125I]iodo-3-pyridinecarboxylate (SIPC) or Iodogen to specific activities of 0.5–2.0 mCi/mg, with yields of 35–50% and 45–70%, respectively. The immunoreactive fraction (IRF) of the labeled MR1(scFv) was 65– 80% when SIPC was used and 50 –55% when Iodogen was used. The affinity (KA) of MR1(scFv) for EGFRvIII was 4.3 3 107 6 0.1 3 107 M21 by BIAcore analysis, and it was 1.0 3 108 6 0.1 3 108 M21 and by Scatchard analysis versus EGFRvIII-expressing cells. After incubation at 37°C for 24 h, the binding affinity was maintained, and the IRF was maintained at 60 –70%. The specificity of MR1(scFv) for EGFRvIII was demonstrated in vitro by incubation of radiolabeled MR1(scFv) with the EGFRvIII-expressing U87MG.DEGFR cell line in the presence or absence of competing unlabeled MR1(scFv) or antiEGFRvIII MAbs L8A4 and H10. In biodistribution studies

Received 12/7/98; revised 3/5/99; accepted 3/10/99. 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 This work was supported by NIH Grants CA42324, CA11898, and NS20023. 2 To whom requests for reprints should be addressed. Phone: (919) 6845018; Fax: (919) 681-8337.

using athymic mice bearing s.c. U87MG.DEGFR tumor xenografts, animals received intratumoral or i.v. infusions of paired-label [125I]SIPC-MR1(scFv) and [131I]SIPC-antiTac(scFv) as a control. When given by the intratumoral route, MR1(scFv) retained high tumor uptakes of 85% injected dose per gram of tissue at 1 h and 16% injected dose per gram of tissue at 24 h following administration. Specific: control scFv tumor uptake ratios of more than 20:1 at 24 h demonstrated specific localization of MR1(scFv). The excellent tumor retention of MR1(scFv), combined with its rapid clearance from normal tissues, resulted in high tumor:normal organ ratios.

INTRODUCTION The concept of using antibodies to target radionuclides to cancer cells has been explored extensively in the past several decades, and the technologies of hybridoma production and recombinant DNA generation have resulted in improved approaches to cancer therapy (1– 4). Recently, impressive results have started to emerge in the treatment of lymphomas and tumors that are amenable to non-i.v. routes of delivery, suggesting that, in certain settings, labeled tumor-specific antibody or fragments may play an important role in cancer diagnosis or immunotherapy (5–7). The long-term goal of our group is focused on the RIT of solid intracranial tumors and neoplastic meningitis. We have been developing genetically engineered constructs of optimal size, specificity, and affinity for more uniform and better tumor penetration and shorter plasma halflife to limit normal tissue exposure; in addition, we also have been exploring different routes of administration, using a compartmental approach into spontaneous cysts or surgically created resection cavities or intrathecal administration (8 –10). Because brain tumors usually do not metastasize out of the central nervous system, intratumoral administration by convection-enhanced infusion delivery is especially attractive (11, 12). Fv fragments of immunoglobulins are the smallest functional modules of antibodies required for antigen binding. Their small size makes them potentially more useful than whole antibodies for certain clinical applications involving the selective delivery of radionuclides to tumors (2, 3, 13). Due to the small size and lack of murine antibody constant domains, scFv3 proteins have rapid pharmacokinetics, greater tumor penetration (14), and lower immunogenicity than intact IgG, F(ab9)2, or Fab.

3

The abbreviations used are: scFv, single-chain Fv; MAb, monoclonal antibody; EGFRvIII, epidermal growth factor receptor-vIII; SIPC, Nsuccinimidyl 5-iodo-3-pyridinecarboxylate; IPTG, isopropyl thiogalactoside; HPLC, high-performance liquid chromatography; TCA, trichloroacetic acid; %ID/g, percentage injected dose per gram of tissue; AUC, area under the curve.

1540 Glioma Targeting by Anti-EGFRvIII Single-Chain Fv

The scFv may be assembled from the variable regions of particular MAbs (15, 16) or made de novo from phage display libraries (4, 17). By using bacterial expression systems, many scFv have been produced with good yields, either as insoluble inclusion bodies in Escherichia coli (15) or by secretion into the periplasm (18) or into the culture supernatant (19). The rate of clearance of a scFv from the blood pool and normal tissues is much more rapid than that seen with intact antibody or Fab (20, 21). This offers the possibility of imaging at earlier times after injection and, for therapy, reducing the radiation-absorbed dose to normal tissues. In addition, autoradiographic studies have shown that scFvs penetrate into tumor better and more homogeneously than intact MAb and larger fragments (14), properties that are important for radioimmunotherapy. The rapid tumor uptake and normal tissue clearance of scFvs make them ideally suited for short half-life nuclides, such as 211At (t1/2 5 7.2 h), used for therapy, and 18F (t1/2 5 2 h), used for positron emission tomography imaging. Thus, radiolabeled scFv molecules have emerged as promising candidates for imaging and therapy of malignant tumors. The mutant EGFRvIII has been found in brain neoplasms, where it is present in 60 –70% of glioblastomas and gliosarcomas but at a low frequency in anaplastic astrocytomas (22, 23). It is also present in squamous cell, adenosquamous, and undifferentiated non-small cell lung cancer (24); intraductal and infiltrating ductal breast cancer; and ovarian carcinoma (25, 26). This mutant receptor is characterized by an 801-bp in-frame deletion generating a new glycine codon at the deletion junction (27). EGFRvIII is expressed on the cell surface and contains a tumor-specific sequence near the NH2 terminus of the receptor extracellular domain. Transfection of NIH-3T3 cells with EGFRvIII cDNA results in a transformed phenotype (23, 28), and introduction of EGFRvIII cDNA into human glioblastoma cells enhances the in vivo malignancy of these cells (29). The frequent expression of EGFRvIII in human cancers makes it a promising target for therapeutic applications. We have developed several MAbs that are specific for EGFRvIII and do not react with normal tissues, including those expressing the wild-type EGFR (26). We have previously shown that, when the anti-EGFRvIII MAbs are radioiodinated using the SIPC labeling method, enhanced intracellular retention of radioactivity was observed compared to conventional radioiodination methods (30, 31). A recombinant antibody phage, MR1(scFv), was isolated from a scFv phage display library by panning with successively decreasing amounts of synthetic peptide containing the EGFRvIII mutant-specific sequence (32). An immunotoxin was produced and characterized and shown to exhibit good binding affinity and high cytotoxicity against EGFRvIII-positive cell lines. Here, we have used the same clone to produce a scFv fragment, MR1(scFv), and we have examined its binding characteristics, stability, and specificity after radioiodination as well as evaluated the biodistribution of 125I-SIPC labeled MR1(scFv) in vivo. Because direct intratumoral administration may be a way to increase the tumor-to-normal tissue dose ratios for radiopharmaceuticals, particularly brain tumors (33–35), we examined the biodistribution of [125I]SIPC-labeled MR1(scFv), comparing intratumoral and i.v. routes of infusion. s.c. human U87MG.DEGFR xenografts in athymic mice were used as the tumor target, and the [125I]SIPC-labeled MR1(scFv; antigen-

specific) and [131I]SIPC labeled anti-Tac(scFv; control) antibody fragments were administrated simultaneously to evaluate in vivo preclinical behavior.

MATERIALS AND METHODS Production of MR1(scFv) Cloning of MR1(scFv) and Plasmid Construction. The MR1 scFv gene was isolated and assembled from an scFv phage display library and sequenced as described previously (32). NdeI and EcoRI sites were introduced at the 59 and 39 ends of the scFv gene and used to clone it into an expression vector pUli7 (16) to produce an expression plasmid, pMR1(scFv). Expression, Refolding, and Purification. The scFv recombinant protein representing the antibody fragments against EGFRvIII was designed as the monomeric VH-linker-VL, where the linker is a 15-mer peptide (Gly4Ser)3 (15, 36) and the expression is under the control of the T7 promoter. The recombinant scFv was expressed in IPTG-induced E. coli BL21(lDE3) cells and accumulated in inclusion bodies (16). Bacterial cultures were inoculated into Superbroth containing 100 mg/ml ampicillin and grown at 37°C to an A600 of 2.0 –2.5. IPTG was added to 1 mM, and growth was continued for 90 min. The cells were then sedimented by centrifugation and resuspended in 50 mM Tris-HCl-20 mM EDTA (pH 7.4) for storage at 270°C. Inclusion bodies were prepared from the bacterial cells by washing a total of five times with 50 mM Tris-HCl-20 mM EDTA (pH 7.4). Refolding of the proteins into active molecules was modified according to Brinkmann et al. (16). Inclusion body proteins were solubilized with 7 M guanidine-100 mM Tris-HCl (pH 8.0) and then reduced by adding dithioerythritol to a final concentration of 10 mg/ml in the same solution. This mixture was incubated at room temperature overnight to affect complete reduction of all disulfide bonds. The reduced protein solution was then diluted exactly 100-fold into renaturation buffer containing 2 M urea, 10 mM Tris-HCl, and 0.9 mM oxidized glutathione (pH 10.3) at 10°C with rapid mixing. As soon as this mixing was complete, the solution was incubated at 10°C for 60 –70 h. After renaturation, the solution containing the refolded protein was adjusted to pH 6.5 and filtered via a 0.2-mm Zap-Cap filter. Properly folded scFv protein was then purified by ionexchange chromatography on Fast Flow SP-Sepharose (Pharmacia LKB Biotechnology Inc., Piscataway, NJ) and eluted with a stepwise NaCl gradient. Correctly folded scFv fragments were eluted from SP-Sepharose between 0.15 and 0.3 M NaCl. Fractions were monitored by SDS-PAGE, and those containing the desired molecular weight species were concentrated to ;1 ml using Centriprep-10 concentrators (Amicon). Further purification to homogeneity was accomplished via size-exclusion HPLC with a TSK2000SW column (TosoHaas, Philadelphia, PA) eluted with PBS-0.4 M NaCl at a flow rate of 0.75 ml/min to separate dimers and multimers from monomers. Protein concentrations were determined by the Bradford assay using the Coomassie Plus kit (Bio-Rad, Hercules, CA). SDS-PAGE was conducted according to Laemmli as described (16) under nonreducing conditions. Gels were stained with Coomassie brilliant blue.


Radiolabeling MR1(scFv) was radiolabeled with 125I using Iodogen or SIPC (30, 31) and purified by gel filtration using a 10-cm Sephadex G-10 (Pharmacia) column. A sample of the purified radiolabeled MR1(scFv) was analyzed via gel filtration HPLC using a TSK2000SW column and by SDS-PAGE followed by autoradiography using a phosphor image analysis system. The phosphor screen was scanned using a Storm 860 PhosphorImager (Molecular Dynamics), and the resulting image was analyzed using the ImageQuant analysis program developed by Molecular Dynamics. Assay for Immunoreactivity The immunoreactive fraction of the radioiodinated MR1(scFv) preparation was determined according to Lindmo et al. (37). Approximately 10,000 cpm of radiolabeled MR1(scFv) was incubated in triplicate with varying amounts of magnetic beads conjugated to purified EGFRvIII extracellular domain protein for 45 min at room temperature. As a control for nonspecific binding of MR1(scFv), the experiment was also performed with magnetic beads conjugated to the wild-type EGFR protein. Bound activity was separated from free activity by magnetic separation, and the percentage immunoreactive fraction was determined. Characterization of Radiolabeled scFv Affinity Constant Determination by Surface Plasmon Resonance (BIAcore). Purified EGFRvIII protein was immobilized on the surface of biosensor chips for analysis using the Pharmacia BIAcore. Coupling of antigen is achieved using N-ethyl-N9-(3-dimethylaminopropyl) carbodiimide/N-hydroxysuccinimide according to the manufacturer’s instructions. The running buffer was 10 mM HEPES, 150 mM NaCl, and 3.4 mM EDTA (pH 7.4). The MR1(scFv) samples were passed over the biosensor chip at concentrations from 200 to 1000 nM. The association and dissociation rate constants (kassoc and kdiss) and average affinity were determined using the nonlinear curvefitting BIAevaluation software. KA at equilibrium was calculated as: KA 5 kassoc/kdiss. Thermal Stability Assay. Thermal stability of the MR1(scFv) was determined by incubating samples at 10 mg/ml in PBS at 37°C for 24 h. Analysis of stability was accomplished by using analytical chromatography on a size-exclusion column to identify monomers, dimers, and larger aggregates and by examining the binding affinity via BIAcore analysis as well as by immunoreactivity assay. Stability of the radiolabeled MR1(scFv) was also assayed by incubation with PBS, mouse serum, and human glioma cyst fluid at 37°C for 30 min, 1 h, 2 h, and 4 h. Protein-associated radioactive counts from each sample were then determined by assaying for solubility in 12% TCA. Scatchard Analysis. EGFRvIII-expressing U87MG. DEGFR and NR6M cells (28, 29) were incubated with varying concentrations of serially diluted [125I]SIPC-labeled MR1(scFv; ;2000 ng/ml) for 18 h at 4°C. As a control for nonspecific cell association of radioactivity, cell binding was performed in the presence of a 100-fold excess of unlabeled MR1(scFv). Data were analyzed using the RADLIG 4.0 Equilibrium Binding Data Analysis Software (Biosoft, Cambridge, United Kingdom).

Measurement of Binding Specificity in Vitro. SIPCradiolabeled MR1(scFv) was incubated with U87MG.DEGFR, NR6M (both positive for EGFRvIII), U87MG, and NR6W (both lacking the EGFRvIII receptor) at 80 ng per 4 3 106 cells for 1 h at 4°C. An additional sample of 125I-labeled MR1(scFv) was assayed with the U87MG.DEGFR cells in the presence of excess MAb L8A4 or H10, which are specific for EGFRvIII. MAbs L8A4 and H10 were isolated by our group (26) and were shown to share overlapping epitopes with MR1(scFv) by SPOTs (Genosys Biotechnologies, Inc., Woodlands, TX) peptide mapping analysis.4 Cells were pelleted and washed in PBS, and the pellets and combined supernatants were counted to determine the percentage of MR1(scFv) bound. Biodistribution Studies Four- to 5-week-old female athymic mice (nu/nu genotype, BALB/c background) were used in all experiments and were maintained in the Duke Comprehensive Cancer Center Isolation Facility. The biodistribution target, U87MG.DEGFR, a human glioma transfected to express the EGFRvIII, is maintained as a xenograft by serial passage in athymic mice. Tumors for biodistribution experiments were initiated by the injection of 50 ml of tumor homogenate. For i.v. biodistribution experiments, tumors were placed on the flank, and for direct intratumoral infusion, the tumors were initiated on the right hind limb. All experiments were started 7 days following initiation, when tumors had reached 150 –300 mm3 in size. Two paired-label tissue distribution studies were performed in athymic mice bearing s.c. U87MG.DEGFR xenografts. Both experiments involved scFv fragments labeled using the SIPC method. In the first experiment, 125I-labeled MR1(scFv) (2 mCi, 2 mg) and 131I-labeled anti-Tac(scFv) (2 mCi, 2 mg) were injected in a total volume of 100 ml of PBS via the lateral tail vein. Groups of five animals were killed by halothane overdose 0.5, 1, 4, 12, and 24 h after injection of the labeled compounds. In the second experiment, for direct intratumoral infusion the skin was opened above the tumor and a 33-gauge cannula was inserted into the middle of the tumor. Forty ml of scFv cocktail, 125I-labeled MR1(scFv) (3 mCi, 3 mg), and 131I-labeled anti-Tac(scFv) (3 mCi, 3 mg) were infused at a rate of 6.0 ml per min using a Harvard PhD 2000 infusion pump. The cannula was removed 1 min following the end of the infusion, and the skin was closed over the tumor with surgical staples. Groups of five mice were killed and dissected 1, 2, 4, 8, 16, and 24 h after injection. Tissues of interest were removed, washed, weighed, and counted for 125I and 131I activity using a dual-channel automated gamma counter. The %ID/g for each radionuclide was calculated by comparison to injection dose standards. Tumor:normal tissue radioactivity ratios also were calculated. Statistical analyses were performed using a paired t test.

4

C. J. Wikstrand and D. D. Bigner, unpublished results.


RESULTS Expression and Characterization of MR1(scFv) The cDNA encoding the recombinant single-chain antibody fragment MR1(scFv) from an scFv phage display library was previously isolated by panning with successively decreasing amounts of a synthetic peptide, LEEKKGNYVVTDHSGGK, the first 13 amino acids corresponding to the EGFRvIII mutant-specific sequence (32). In this study, we used the same clone to construct an expression plasmid, pMR1(scFv), and produced an MR1(scFv) protein to examine the binding characteristics, stability, and specificity of radioiodinated MR1(scFv) and to evaluate its biodistribution in tumor-bearing mice. Protein Expression, Refolding, and Purification. The MR1(scFv) construct was designed as a single-chain protein in which the VH and VL domains were connected by a 15-mer peptide linker (Gly4-Ser)3. The MR1(scFv) sequence was used to construct a plasmid in which expression is driven by the T7 promoter (16). E. coli BL21(lDE3) cultures harboring plasmid for expression of the MR1(scFv) were induced in the exponential-growth phase with IPTG for 1.5 h. The recombinant protein accumulated in large amounts (.150 mg/liter culture in shaker flasks) as insoluble intracellular inclusion bodies. These inclusion bodies contained almost pure recombinant protein, but in an inactive form requiring refolding to regain activity. Inclusion bodies were isolated from lysed bacteria by centrifugation. Properly folded scFv fragments were obtained by mixing the solubilized and reduced inclusion bodies in redox-shuffling refolding buffer, as described in “Materials and Methods.” After renaturation, properly folded scFv fragments were purified by SP-Sepharose ion exchange fast-performance liquid chromatography and subsequently by size-exclusion HPLC. The yield of refolded scFv was ; 2–3%, expressed as milligram of protein after TSK2000SW chromatography per milligram of inclusion body protein. The MR1(scFv) protein, after refolding and purification, migrated as a single band of the expected size (Mr 26,000) that was .95% pure on SDS-PAGE (Fig. 1). Analysis of the purified MR1(scFv) by size-exclusion HPLC demonstrated that the scFv fragment was a monomer that was eluted at 13–14 min (Fig. 1). Radioiodination and Immunoreactivity. MR1(scFv) was radioiodinated using two procedures: conventional iodination with Iodogen or SIPC. The radiolabeling yields using each procedure were generally in the ranges of 45–70% using Iodogen and 35–50% using SIPC. The purity of the radiolabeled scFv was analyzed by TCA solubility, size-exclusion HPLC, and SDS-PAGE. Preparations of radiolabeled MR1(scFv) were assayed for solubility in 12.5% TCA, and the radioactivity was determined to be .95% insoluble for all preparations. The HPLC radioactivity chromatogram (Fig. 1A) of radiolabeled MR1(scFv), using SIPC as the labeling method, suggests that ;95% of the radioactivity corresponded to a retention time consistent with the molecular weight of an scFv. There was a slight shoulder associated with the activity peak; however, SDSPAGE analysis followed by autoradiography confirmed the presence of a single radiolabeled species of the appropriate molecular weight (Fig. 1A, inset). Comparable results were obtained for radioiodinated MR1(scFv) using the Iodogen label-

Fig. 1 Radioactivity chromatograms of [125I]SIPC-labeled (A) and Iodogen-labeled (B) MR1(scFv). Approximately 8 mCi of the radiolabeled preparation of MR1(scFv) were analyzed via size-exclusion HPLC. The retention time of the major peak was 13.5 min. Inset, autoradiograph of the same preparation of radiolabeled MR1(scFv) after SDS-PAGE. 14C-labeled molecular weight standards were also run as a reference.

ing method (Fig. 1B). The immunoreactivity of [125I ]SIPClabeled and Iodogen labeled MR1(scFv) were assayed using EGFRvIII extracellular domain-coated magnetic beads. The immunoreactive fraction of the labeled MR1(scFv) was determined to be 65– 80% using SIPC and 50 –55% using Iodogen. For this reason, the SIPC labeling method was used for the remainder of the experiments detailed below. Binding of MR1(scFv) to EGFRvIII and Stability. A kinetic analysis of the interaction of purified scFv fragment with immobilized EGFRvIII extracellular domain by surface plasmon resonance (BIAcore) was conducted to determine the association and dissociation rate constants and calculation of the affinity constants. Determination of the association and dissociation rates from the sensorgrams revealed a kassoc of 9.1 3 104 M21zs21 and a kdiss of 2.1 3 1023 s21 for MR1(scFv), as shown in Table 1. The KA at binding equilibrium, calculated as KA 5 kassoc/kdiss, is 4.3 3 107 21 M . Our data show that recombinant MR1(scFv) exhibits an affinity similar to that reported previously for MR1(scFv)PE38KDEL immunotoxin (32). Only very slight differences were found in the rate constants between nonradiolabeled and radiolabeled MR1(scFv) as shown in Table 1. The stability of scFv fragments at 37°C is crucial to their


Table 1 Protein MR1(scFv) 125

I-MR1(scFv)

Kinetic analysis of MR1(scFv)-EGFRvIII interactionsa

Incubation

kassoc (M21 z S21)

None 37°C/24 h None 37°C/24 h

9.08 3 10 6 0.08 3 10 7.08 3 104 6 0.09 3 104 1.39 3 105 6 0.08 3 105 1.14 3 105 6 0.08 3 105 4

kdiss(s21) 4

23

KA (M21) 23

2.10 3 10 6 0.04 3 10 1.83 3 1023 6 0.04 3 1023 3.51 3 1023 6 0.04 3 1023 3.21 3 1023 6 0.04 3 1023

4.32 3 107 6 0.12 3 107 3.87 3 107 6 0.14 3 107 3.98 3 107 6 0.30 3 107 3.58 3 107 6 0.30 3 107

a Interaction of purified scFv fragments with immobilized EGFRvIII extracellular domain by surface plasmon resonance (BIAcore), as described in “Materials and Methods.” Data were generated and analyzed using the BioEvaluation Version 3.0 software package provided by BIAcore Biosensor. Molarities of the scFv were calculated based on a Mr 26,000 monomer.

usefulness as therapeutic agents and is governed by their tendency toward aggregation at 37°C. In stability studies, a sample of the MR1(scFv) protein retained 90% of its initial binding activity after 3 months of storage in PBS at 4°C. The thermal stability of MR1(scFv) was examined by measuring the amount of aggregation and inactivation in terms of binding affinity and immunoreactivity at 37°C for 24 h. We examined the chromatographic profiles of scFv before and after incubation in PBS for 24 h at 37°C. MR1(scFv) was a monomer before incubation in PBS and remained monomeric for 24 h at 37°C (results not shown). It was noted that ,1% of lower molecular weight catabolites were found after incubation. We also measured and compared the binding affinity of MR1(scFv) via BIAcore analysis before and after thermal incubation (Table 1). The results showed that, after incubation of MR1(scFv) at 37°C for 24 h, the construct still retained 90% of the binding activity. The stability of [125I]SIPC labeled MR1(scFv) was assayed in various media relevant to the potential in vivo applications of this scFv. Both human and athymic mouse serum were used along with PBS to determine the stability of MR1(scFv) over time at 37°C, as defined by TCA solubility of radioactive counts. MR1(scFv) remained relatively stable over 4 h in PBS at room temperature; after 30 min at 37°C, we observed a slight decrease in the protein associated activity in all media. After 4 h at 37°C, we observed 1.8, 2.7, and 1.2% losses in protein-associated activity for mouse serum, human glioma cyst fluid, and PBS, respectively (data not shown). In addition, the affinity of radiolabeled MR1(scFv) after incubation at 37°C for 24 h was still 90% of the initial activity (Table 1). Immunoreactivity of MR1(scFv) was analyzed following radioiodination by incubation in the presence of excess EGFRvIII extracellular domain coupled to magnetic beads. Results determined by Lindmo analysis demonstrated that immunoreactive fraction of MR1(scFv) remained unchanged after 16 h and still was ;60 –70% of the initial value after 24 h incubation in PBS at 37°C (data not shown). Scatchard Analysis of [125I]SIPC-labeled MR1(scFv) versus EGFRvIII-positive Cells. The affinity of radiolabeled MR1(scFv) for the EGFRvIII positive cell lines U87MG.DEGFR and NR6M was determined by Scatchard analysis. Varying amounts of radiolabeled MR1(scFv) were incubated with 106 cells in triplicate at 4°C for 18 h. A second set of samples was incubated with 100-fold excess of unlabeled MR1(scFv) to block the specific binding of the labeled MR1(scFv) and allow for determination of nonspecific association of radioactivity. Fig. 2 shows the Scatchard plots for MR1(scFv) binding to both cell lines. The affinity was determined to be 1.0 3 108 6 0.1 3 108 M21 for U87MG.DEGFR

Fig. 2 Scatchard analysis-defined affinity determination for 125I-labeled MR1(scFv) versus EGFRvIII-positive cells. Scatchard plots for activity bound to U87MG.DEGFR cells (A) and NR6M cells (B).

cells (Fig. 2A) and 2.0 3 108 6 0.4 3 108 M21 for NR6M cells (Fig. 2B). The Scatchard-defined bmax was used to estimate the number of EGFRvIII receptors per cell. The number of receptors per cell was 1.2 3 106 for U87MG.DEGFR and 7.0 3 105 for NR6M, consistent with earlier reports (26, 38).


Fig. 3 Specificity of 125I-labeled MR1(scFv) for EGFRvIII-expressing cell lines. A, the percentage of labeled MR1(scFv) bound. f, binding in the absence of unlabeled MR1(scFv); M, percentage bound in the presence of a 10-fold excess of unlabeled MR1(scFv). B, inhibition of 125 I-labeled MR1(scFv) binding by anti-EGFRvIII MAbs L8A4 and H10. The percentage of labeled MR1(scFv) bound to U87MG.DEGFR cells is shown in the absence and presence of cold MAb as a competitive inhibitor of cell binding. Binding to EGFRvIII-negative U87MG cells is also shown. Columns, means of triplicate measurements; bars, SD.

Specificity of Radioiodinated MR1(scFv) in Vitro. [125I]SIPC-labeled MR1(scFv) was incubated with cell lines containing either the mutant EGFRvIII (U87MG.DEGFR and NR6M) or wild-type EGFR (U87MG and NR6W), in the presence and absence of an excess of unlabeled MR1(scFv). Fig. 3A shows the percentage of radioactive counts bound to cells after a 1 h incubation at 4°C and one wash. There was no detectable binding above background levels (, 3%) to the EGFR wildtype expressing or nontransfected cells in the presence or absence of the unlabeled competitor, whereas binding was observed in both EGFRvIII-expressing cell lines. Specificity was demonstrated by the ability of unlabeled MR1(scFv) to reduce the binding to background levels. A similar experiment was performed using the EGFRvIII-positive cell line U87MG.DEGFR, negative control cell line U87MG for nonspecific binding, 125I-labeled MR1(scFv), and the anti-EGFRvIII

Fig. 4 i.v. injections of 125I-labeled MR1(scFv) and 131I-labeled antiTac(scFv) in U87MG.DEGFR xenografts. A, uptake of radioiodine (% ID/g). B, localization ratios of 125I-labeled MR1(scFv) and 131I-labeled anti-Tac(scFv). Animals in groups of five were given i.v. injections of 125 I-labeled MR1(scFv) and 131I-labeled anti-Tac(scFv) and killed at the time points indicated. Tumor %ID/g values and ratios of MR1(scFv) over anti-Tac(scFv) in tumor were determined. Columns, means; bars, SD.

MAbs L8A4 and H10 as competitive inhibitors. In the absence of anti-EGFRvIII specific MAb, ;63% of the 125I-labeled MR1(scFv) was bound to the U87MG.DEGFR cells. Fig. 3B demonstrates that the MAbs only partially inhibit the binding of radiolabeled MR1(scFv) to the target cells. MAbs L8A4 and H10 demonstrated a 65% and 56% inhibition of binding of the radiolabeled MR1(scFv), respectively. Less than 3% of the activity bound to the EGFRvIII-negative control cells. Biodistribution Studies In the first experiment, an i.v. injection route was used to compare the biodistribution properties of 125I-labeled MR1(scFv) and 131I-labeled anti-Tac nonspecific scFv in athymic mice with U87MG.DEGFR xenografts. Both labeled proteins exhibited rapid clearance of radioiodine activity from all normal tissues. Kidney levels were high at 30 min (82 6 16%ID/g, MR1; 103 6 19%ID/g, anti-Tac) and had declined ;10-fold by 4 h. The tumor uptake of the two radioiodinated scFv as a function of time after i.v. injection are compared in Fig. 4A. Uptake of 125I-labeled MR1(scFv) in these U87MG.DEGFR xenografts was highest, 3.38 6 0.39%ID/g, at 0.5 h and declined to 0.49 6 0.09%ID/g by 4 h. Tumor levels of 131 I-labeled anti-Tac nonspecific scFv were significantly lower


from normal tissues resulted in high tumor:normal organ ratios. For example, at 24 h, the ratios of tumor activity to that in liver, blood, stomach, kidney, spleen, and lungs were 958:1, 902:1, 361:1, 238:1, 430:1, and 66:1, respectively (Fig. 6). Although tumor:normal organ ratios of .1 were also observed for the control scFv, these values were considerably lower than those obtained with MR1(scFv).

DISCUSSION

Fig. 5 Intratumoral infusions of 125I-labeled MR1(scFv) and 131I-labeled anti-Tac(scFv) in U87MG.DEGFR xenografts. A, uptake of radioiodine (% ID/g). B, localization ratios of 125I-labeled MR1(scFv) and 131 I-labeled anti-Tac(scFv). Animals in groups of five were given intratumoral infusions of 125I-labeled MR1(scFv) and 131I-labeled antiTac(scFv) and killed at the time points indicated. Tumor %ID/g values and ratios of MR1(scFv) over anti-Tac(scFv) in tumor were determined. Columns, means; bars, SD.

(P , 0.05) than specific scFv at all time points; however, the MR1(scFv)/anti-Tac(scFv) tumor accumulation ratio remained at 2:1 or lower throughout the course of the experiment (Fig. 4B). With the exception of brain, where tumor:organ ratios up to 25:1 (0.5 h) were observed, tumor:normal tissue ratios were modest, generally less than 3:1. Different results were observed when the two radioiodinated scFv molecules were administered by infusion directly into the U87MG.DEGFR xenografts. In this experiment, levels of 125I-labeled MR1(scFv) in tumor were 84.5 6 47.5%ID/g at 1 h compared with 28.9 6 13.9%ID/g for coadministered 131I-labeled anti-Tac nonspecific scFv. As shown in Fig. 5A, both scFv cleared from these xenografts with time. With MR1(scFv), 15.7 6 12.4% ID/g of the 125I activity still remained in the tumor at 24 h compared with only 0.69 6 0.54% ID/g for 131I-labeled anti-Tac scFv. MR1(scFv)/anti-Tac scFv tumor accumulation ratios were greater than 15:1 by 16 h after injection (Fig. 5B), demonstrating the specificity of MR1(scFv) localization in this EGFRvIII-expressing xenograft model. The excellent tumor retention of MR1(scFv) combined with its rapid clearance

scFv molecules that react specifically with tumor-associated antigens are being pursued as potential molecular carriers for use in the selective delivery of radionuclides to tumors. Because of their rapid clearance from the blood pool and most normal organs, scFvs might be ideally applied in combination with promising radionuclides such as 18F, 211At, and 213Bi, which have short physical half-lives that would confound their clinical application in tandem with intact IgG or larger MAb fragments. An additional advantage of particular import for therapeutic radionuclides emitting short-range radiation is the more rapid tumor penetration and homogeneous distribution of scFv molecules (14) compared with whole MAb, increasing the likelihood that homogeneous tumor radiation dose deposition will be achieved. We have used the MR1(scFv) cDNA clone encoding the antibody variable regions that are specific to the EGFRvIII mutant-specific sequence and produced a recombinant singlechain MR1(scFv) protein in E. coli that exhibited EGFRvIII binding specificity, affinity, and thermal stability and that can be radiolabeled without loss of binding affinity. Because the recombinant MR1(scFv) can be efficiently labeled with radioactivity while retaining most of its immunoreactivity, it could be useful for tumor targeting. The thermal stability of MR1(scFv) provides clinical advantages for radioimmunotargeting. Several reports have demonstrated that the rapid equilibration and elimination phases displayed by scFv proteins in vivo make them very effective targeting agents (2, 13, 39). In addition, the MR1(scFv) could be used for immunohistological applications such as the MR1(scFv)-fusion protein generation with peroxidase, b-galactosidase, or alkaline phosphatase (40) for one-step detection of EGFRvIII. However, upon i.v. administration, only relatively low tumor accumulation of radiolabeled specific MR1(scFv) was observed, most likely due to the generally rapid clearance of scFv fragments from circulation. These results may hinder the utility of MR1(scFv) for tumor imaging. Results from several groups have suggested that larger fragments, such as diabody (41, 42) or minibody (43), may improve the circulation times so as to facilitate imaging and may be better suited to in vivo targeting applications. Quality control measurements were performed on each batch of radioiodinated MR1(scFv) to insure purity and immunoreactivity of the radiolabeled material. Both the Iodogen and SIPC methods demonstrated a radiolabeled monomer with the molecular weight corresponding to the scFv. The shoulder associated with the SIPC-labeled MR1(scFv) is likely an effect of charge modification of a fraction of the preparation, because the SDS-PAGE analysis followed by autoradiography confirmed a single labeled band. In addition, similar profiles were obtained when using SIPC to radioiodinate an scFv specific for the TAC


Fig. 6 Tumor:normal tissue uptake ratios by 125I-labeled MR1(scFv) (E) and 131I-labeled anti-Tac(scFv) (). Ratios obtained for liver, blood, stomach, kidney, spleen, and lungs are displayed for the times indicated.

antigen (data not shown). The immunoreactive fractions of the radioiodinated preparations was comparable to that of the antiEGFRvIII MAb L8A4 (30). There was a significant increase in immunoreactivity when SIPC was used for radiolabeling compared to Iodogen. This observation may be due to differences in the amino acid residues modified by each procedure. The Iodogen method modifies tyrosine residues, whereas SIPC labeling involves coupling through epsilon amino groups of lysines. The percentage of tyrosines in the CDR regions of MR1(scFv) is 50% (6 of 12), whereas only 15% (2 of 13) of the lysine residues are located in the CDR regions. Therefore, a greater probability exists that, when using Iodogen, a residue potentially involved in binding is modified by the labeling. This modification could then result in an inactive molecule, resulting in a lower immunoreactive fraction for the preparation. The Scatchard analysis-determined affinity (KA) of MR1(scFv) for EGFRvIII-expressing cells is between 1 3 108 21 M and 2 3 108 M21, ;5–10 times lower than what has been observed for the anti-EGFRvIII MAbs (26, 30). In general, such a “medium” affinity of scFv is sufficient for tumor imaging when the cancer cells express very large numbers of antigen, such as cancer-associated carbohydrates (44), but for mediumdensity antigens, antibodies with higher affinity are desirable (45). Increased affinity can be achieved by phage mutagenesis techniques (4), and this may improve the diagnostic and therapeutic potential of the anti-EGFRvIII antibody scFv. The current construct of scFv exhibits a lower binding affinity and lacks bivalency compared to the related anti-EGFRvIII MAbs. Further improvements of recombinant MR1(scFv) fragments may include the generation of bivalent “diabody” (41) or trivalent “triabody” (46) and the “humanization” of the mouse antibody to reduce or diminish antimouse antibody responses. The number of EGFRvIII receptors per cell determined using the Scatchard defined bmax was in good agreement with our previously published values for both cell lines (26, 38, 47). The Scatchard analysis-determined affinity values were slightly higher than the

determinations obtained using surface plasmon resonance. One possible explanation for these differences may be that the receptor is less reactive when immobilized onto the biosensor chip. Slight modifications in structure could occur during membrane solubilization and amine coupling to the chip, which would result in less efficient binding. Perhaps the most significant limitation of i.v. injected scFv is that the percentage of the injected dose taken up by tumor is lower than that of intact IgG, and furthermore, the clearance of radioactivity from tumor is considerably more rapid than that with whole MAb (42, 48, 49). In this study, similar behavior was observed with 125I-labeled MR1(scFv) after i.v. administration, with a maximum of 4% ID/g observed after 0.5 h in EGFRvIII-expressing U87MG.DEGFR xenografts. Nonetheless, tumor levels of MR1(scFv) were twice as high as those for coadministered anti-Tac scFv control, demonstrating that MR1(scFv) tumor accumulation was specific. Similar ratios of specific:control scFv have been reported for monovalent anti-cerbB-2 scFv (42). Although the specificity of MR1(scFv) was similar to that reported previously for other scFv (42), its tumor retention, particularly at later time points, was considerably lower than that observed with other scFv molecules (21, 42, 50). One possibility is that an insufficient number of receptors were available in tumor for MR1(scFv). This is not expected because this xenograft has been shown to express an average of 1.8 3 105 EGFRvIII per cell, allowing excellent targeting (up to 25% ID/g) of SIPC-labeled intact murine L8A4 IgG (47). A more likely explanation is that the rapid internalization of MR1(scFv; data not shown) accelerates the degradation of this protein within the lysosomes. Catabolic studies are planned to investigate this possibility. A number of approaches are being investigated in an attempt to improve the clinical utility of scFv for targeting radionuclides to tumors. These include the use of continuous infusions (48), disulfide stabilization (51, 52), and more stable radioiodination methodologies (53). The rapidity of


scFv clearance from the blood pool probably contributes to the relatively low levels of scFv that can be delivered to tumor. One approach for increasing interaction of labeled molecules with tumor is to use intratumoral administration routes. In this study, we have investigated whether intratumoral injection of radioiodinated MR1(scFv) could be used to increase the magnitude of its tumor uptake. This injection route is of particular relevance for brain tumors, where convection-enhanced delivery by direct brain infusion over a period of several days has been shown to saturate an entire hemisphere in animals (54, 55). Moreover, Laske et al. (12) have shown that convection-enhanced infusion delivery of a transferrin-diphtheria toxin conjugate (Tf-CRM107) is well tolerated and produced radiographic responses in recurrent malignant glioma patients. The smaller and more specific MR1 molecule may be even more efficacious than the transferrin-diphtheria toxin conjugate for convection-enhanced infusion delivery treatment of intracranial tumors. The results obtained with MR1(scFv) given by intratumoral infusion are highly encouraging. Tumor levels were .20 times higher than those achieved following i.v. injection and remained .15% ID/g, even after 24 h. Specific/control scFv tumor uptake ratios of more than 20:1 at 24 h demonstrate that retention of MR1(scFv) in this xenograft is specific. It is worth noting that the degree of specificity observed after intratumoral administration of intact MAbs is much lower, with specific: control MAb tumor uptake ratios of only ;2:1–3:1 being observed at 24 h (33, 56, 57). Furthermore, the tumor:normal tissue ratios observed for 125I-labeled MR1(scFv) are extremely favorable and even higher than those obtained after intratumoral administration of intact MAbs (56, 57). As an initial indicator of therapeutic potential, the AUC (in units of %ID/g 3 h) for tumor and blood uptake curves was calculated. In the i.v. experiment, the AUCs for tumor and blood were 8.6 and 4.8, respectively, yielding an AUC tumor:blood ratio of only 1.9. On the other hand, after intratumoral administration, the AUC for tumor and blood were 861.8 and 4.8, yielding an AUC tumor:blood ratio of 180. Clearly, intratumoral administration should improve the therapeutic potential of MR1(scFv). If ongoing autoradiographic studies confirm that MR1(scFv) is homogeneously distributed within tumor xenografts, this molecule might be a valuable carrier for use in tandem with the a emitter 211At for the treatment of EGFRvIII-expressing tumors via intratumoral administration routes.

ACKNOWLEDGMENTS We thank Nicholas Xanthakos, Scott Szafranski, and Susan Slade for technical assistance and Dr. Carol J. Wikstrand for reading the manuscript.

REFERENCES 1. Ehrlich, P. Collected Studies in Immunity. New York: J. Wiley & Sons, 1905. 2. Huston, J. S., George, A. J., Adams, G. P., Stafford, W. F., Jamar, F., Tai, M. S., McCartney, J. E., Oppermann, H., Heelan, B. T., Peters, A. M., Houston, L. L., Bookman, M. A., Wolf, E. J., and Weiner, L. M. Single-chain Fv radioimmunotargeting. Q. J. Nucl. Med., 40: 320 –333, 1996.

3. Reisfeld, R. A., and Gillies, S. D. Recombinant antibody fusion proteins for cancer immunotherapy. Curr. Top. Microbiol. Immunol., 213: 27–53, 1996. 4. Winter, G., Griffiths, A. D., Hawkins, R. E., and Hoogenboom, H. R. Making antibodies by phage display technology. Annu. Rev. Immunol., 12: 433– 455, 1994. 5. Kaminski, M. S., Zasadny, K. R., Francis, I. R., Milik, A. W., Ross, C. W., Moon, S. D., Crawford, S. M., Burgess, J. M., Petry, N. A., Butchko, G. M., Glenn, S. D., and Wahl, R. L. Radioimmunotherapy of B-cell lymphoma with [131I]anti-B1 (anti-CD20) antibody. N. Engl. J. Med., 329: 459 – 465, 1993. 6. Kaminski, M. S., Zasadny, K. R., Francis, I. R., Fenner, M. C., Ross, C. W., Milik, A. W., Estes, J., Tuck, M., Regan, D., Fisher, S., Glenn, S. D., and Wahl, R. L. Iodine-131-anti-B1 radioimmunotherapy for B-cell lymphoma. J. Clin. Oncol., 14: 1974 –1981, 1996. 7. Press, O. W., Eary, J. F., Appelbaum, F. R., Martin, P. J., Badger, C. C., Nelp, W. B., Glenn, S., Butchko, G., Fisher, D., Porter, B., Matthews, D. C., Fisher, L. D., and Bernstein, I. D. Radiolabeledantibody therapy of B-cell lymphoma with autologous bone marrow support. N. Engl. J. Med., 329: 1219 –1224, 1993. 8. Bigner, D. D., Brown, M. T., Friedman, A. H., Coleman, R. E., Akabani, G., Friedman, H. S., Thorstad, W. L., McLendon, R. E., Bigner, S. H., Zhao, X. G., Pegram, C. N., Wikstrand, C. J., Herndon, J. N., Vick, N. A., Paleologos, N., Cokgor, I., Provenzale, J. M., and Zalutsky, M. R. Iodine-131-labeled antitenascin monoclonal antibody 81C6 treatment of patients with recurrent malignant gliomas: Phase I trial results. J. Clin. Oncol., 16: 2202–2212, 1998. 9. Brown, M. T., Coleman, E., Friedman, A. H., Friedman, H. S., McLendon, R. E., Reiman, R., Felsberg, G. J., Tien, R. D., Bigner, S. H., Zalutsky, M. R., Zhao, X. G., Wikstrand, C. J., Pegram, C. N., Herndon, J. N., Vick, N. A., Paleologos, N., Fredericks, R. K., Schold, S., Jr., and Bigner, D. D. Intrathecal 131I-labeled antitenascin monoclonal antibody 81C6 treatment of patients with leptomeningeal neoplasms or primary brain tumor resection cavities with subarachnoid communication: Phase I trial results. Clin. Cancer Res., 2: 963–972, 1996. 10. Pastan, I. H., Archer, G. E., McLendon, R. E., Friedman, H. S., Fuchs, H. E., Wang, Q-C., Pai, L. H., Herndon, J., and Bigner, D. D. Intrathecal administration of a single-chain immunotoxin, LMB-7 [B3(Fv)-PE38], produces cures of carcinomatous meningitis in a rat model. Proc. Natl. Acad. Sci. USA, 92: 2765–2769, 1995. 11. Morrison, P. F., Laske, D. W., Bobo, H., Oldfield, E. H., and Dedrick, R. L. High-flow microinfusion: tissue penetration and pharmacodynamics. Am. J. Physiol., 266: R292–R305, 1994. 12. Laske, D. W., Youle, R. J., and Oldfield, E. H. Tumor regression with regional distribution of the targeted toxin TF-CRM107 in patients with malignant brain tumors. Nat. Med., 3: 1362–1368, 1997. 13. Raag, R., and Whitlow, M. Single-chain Fvs. FASEB J., 9: 73– 80, 1995. 14. Yokota, T., Milenic, D. E., Withlow, M., and Schlom, J. Rapid tumor penetration of a single-chain Fv and comparison with other immunoglobulin forms. Cancer Res., 52: 3402–3408, 1992. 15. Huston, J. S., Levinson, D., Mudgett-Hunter, M., Tai, M. S., Novotny, J., Margolies, M. N., Ridge, R. J., Bruccoleri, E. H., Crea, R., and Oppermann, H. Protein engineering of antibody binding sites: recovery of specific activity in an anti-digoxigenin single-chain Fv analogue produced in Escherichia coli. Proc. Natl. Acad. Sci. USA, 85: 5879 –5883, 1988. 16. Brinkmann, U., Pai, L. H., FitzGerald, D. J., Willingham, M., and Pastan, I. B3(Fv)-PE38KDEL, a single-chain immunotoxin that causes complete regression of a human carcinoma in mice. Proc. Natl. Acad. Sci. USA, 88: 8616 – 8620, 1991. 17. Chester, K. A., Begent, R. H., Robson, L., Keep, P., Pedley, R. B., Boden, J. A., Boxer, G., Green, A., Winter, G., Cochet, O., and Hawkins, R. E. Phage libraries for generation of clinically useful antibodies. Lancet, 343: 455– 456, 1994. 18. Skerra, A. Bacterial expression of immunoglobulin fragments. Curr. Opin. Immunol., 5: 256 –262, 1993.


19. Takkinen, K., Laukkanen, M. L., Sizmann, D., Alfthan, K., Immonen, T., Vanne, L., Kaartinen, M., Knowles, J. K., and Teeri, T. T. An active single-chain antibody containing a cellulase linker domain is secreted by Escherichia coli. Prot. Eng., 4: 837– 841, 1991. 20. Jain, R., and Baxter, L. Mechanisms of heterogeneous distribution of monoclonal antibodies and other macromolecules in tumors: significance of elevated interstitial pressure. Cancer Res., 48: 7022–7033, 1988. 21. Milenic, D. E., Yokota, T., Filpula, D. R., Finkelman, M. A. J., Dodd, S. W., Wood, J. F., Withlow, M., Snoy, P., and Schlom, J. Construction, binding properties, metabolism, and tumor targeting of a single chain Fv derived from the pancarcinoma monoclonal antibody CC49. Cancer Res., 51: 6363– 6371, 1991. 22. Bigner, S. H., Humphrey, P. A., Wong, A. J., Vogelstein, B., Mark, J., Friedman, H. S., and Bigner, D. D. Characterization of the epidermal growth factor receptor in human glioma cell lines and xenografts. Cancer Res., 50: 8017– 8022, 1990. 23. Yamazaki, H., Ohba, Y., Tamaoki, N., and Shibuya, M. A deletion mutation within the ligand binding domain is responsible for activation of epidermal growth factor receptor gene in human brain tumors. Jpn. J. Cancer Res., 81: 773–779, 1990. 24. Garcia de Palazzo, I. E., Adams, G., P., Sundareshan, P., Wong, A. J., Testa, J. R., Bigner, D. D., and Weiner, L. M. Expression of mutated epidermal growth factor receptor by non-small cell lung carcinomas. Cancer Res., 53: 3217–3220, 1993. 25. Moscatello, D. K., Holgado-Madruga, M., Godwin, A. K., Ramirez, G., Gunn, G., Zoltick, P. W., Biegel, J. A., Hayes, R. L., and Wong, A. J. Frequent expression of a mutant epidermal growth factor receptor in multiple human tumors. Cancer Res., 55: 5536 –5539, 1995. 26. Wikstrand, C. J., Hale, L. P., Batra, S. K., Hill, M. L., Humphrey, P. A., Kurpad, S. N., McLendon, R. E., Moscatello, D., Pegram, C. N., Reist, C. J., Traweek, S. T., Wong, A. J., Zalutsky, M. R., and Bigner, D. D. Monoclonal antibodies against EGFRvIII are tumor specific and react with breast and lung carcinomas and malignant gliomas. Cancer Res., 55: 3140 –3148, 1995. 27. Humphrey, P. A., Wong, A. J., Vogelstein, B., Zalutsky, M. R., Fuller, G. N., Archer, G. E., Friedman, H. S., Kwatra, M. M., Bigner, S. H., and Bigner, D. D. Anti-synthetic peptide antibody reacting at the fusion junction of deletion-mutant epidermal growth factor receptors in human glioblastoma. Proc. Natl. Acad. Sci. USA, 87: 4207– 4211, 1990. 28. Batra, S. K., Castelino-Prabhu, S., Wikstrand, C. J., Zhu, X., Humphrey, P. A., Friedman, H. S., and Bigner, D. D. Epidermal growth factor ligand-independent, unregulated, cell-transforming potential of a naturally occurring human mutant EGFRvIII gene. Cell Growth Differ., 6: 1251–1259, 1995. 29. Nishikawa, R., Ji, X-D., Harmon, R. C., Lazar, C. S., Gill, G. N., Cavanee, W. K., and Huang, H-J. S. A mutant epidermal growth factor receptor common in human glioma confers enhanced tumorigenicity. Proc. Natl. Acad. Sci. USA, 91: 7727–7731, 1994. 30. Reist, C. J., Garg, P. K., Alston, K. L., Bigner, D. D., and Zalutsky, M. R. Radioiodination of internalizing monoclonal antibodies using N-succinimidyl 5-iodo-3-pyridinecarboxylate. Cancer Res., 56: 4970 – 4977, 1996. 31. Reist, C. J., Archer, G. E., Wikstrand, C. J., Bigner, D. D., and Zalutsky, M. R. Improved targeting of an anti-epidermal growth factor receptor variant III monoclonal antibody in tumor xenografts after labeling using N-succinimidyl 5-iodo-3-pyridinecarboxylate. Cancer Res., 57: 1510 –1515, 1997. 32. Lorimer, I. A., Keppler-Hafkemeyer, A., Beers, R. A., Pegram, C. N., Bigner, D. D., and Pastan, I. Recombinant immunotoxins specific for a mutant epidermal growth factor receptor: targeting with a single chain antibody variable domain isolated by phage display. Proc. Natl. Acad. Sci. USA, 93: 14815–14820, 1996. 33. Larsen, R. H., and Bruland, O. S. Intratumour injection of immunoglobulins labelled with the a-particle emitter 211At: analyses of tumour retention, microdistribution and growth delay. Br. J. Cancer, 77: 1115–1122, 1998.

34. Otsuji, E., Yamaguchi, T., Tsuruta, H., Yata, Y., Nishi, H., Okamoto, K., Taniguchi, K., Kato, M., Kotani, T., Kitamura, K., and Takahashi, T. The effect of intravenous and intra-tumoural chemotherapy using a monoclonal antibody-drug conjugate in a xenograft model of pancreatic cancer. Eur. J. Surg. Oncol., 21: 61– 65, 1995. 35. Borchardt, P. E., Quadri, S. M., Freedman, R. S., and Vriesendorp, H. M. Intralesional radiolabeled human monoclonal IgM in human tumor xenografts. Radiother. Oncol., 44: 283–293, 1997. 36. Bird, R. E., Hardman, K. D., Jacobson, J. W., Johnson, S., Kaufman, B. M., Lee, S-M., Lee, T., Pope, S. H., Riordan, G. S., and Withlow, M. Single-chain antigen binding proteins. Science (Washington DC), 242: 423– 426, 1988. 37. Lindmo, T., Boven, E., Cuttitta, F., Fedorko, J., and Bunn, P., Jr. Determination of the immunoreactive fraction of radiolabeled monoclonal antibodies by linear extrapolation to binding at infinite antigen excess. J. Immunol. Methods, 72: 77– 89, 1984. 38. Chu, C. T., Everiss, K. D., Wikstrand, C. J., Batra, S. K., Kung, H. J., and Bigner, D. D. Receptor dimerization is not a factor in the signalling activity of a transforming variant epidermal growth factor receptor (EGFRvIII). Biochem. J., 324: 855– 861, 1997. 39. Begent, R. H., Verhaar, M. J., Chester, K. A., Casey, J. L., Green, A. J., Napier, M. P., Hope-Stone, L. D., Cushen, N., Keep, P. A., Johnson, C. J., Hawkins, R. E., Hilson, A. J., and Robson, L. Clinical evidence of efficient tumor targeting based on single-chain Fv antibody selected from a combinatorial library. Nat. Med., 2: 979 –984, 1996. 40. Wels, W., Harwerth, I. M., Zwickl, M., Hardman, N., Groner, B., and Hynes, N. E. Construction, bacterial expression and characterization of a bifunctional single-chain antibody-phosphatase fusion protein targeted to the human erbB-2 receptor. Biotechnology, 10: 1128 –1132, 1992. 41. Holliger, P., Prospero, T., and Winter, G. “Diabodies”: small bivalent and bispecific antibody fragments. Proc. Natl. Acad. Sci. USA, 90: 6444 – 6448, 1993. 42. Adams, G. P., McCartney, J. E., Tai, M. S., Oppermann, H., Huston, J. S., Stafford, W. D., Bookman, M. A., Fand, I., Houston, L. L., and Weiner, L. M. Highly specific in vivo tumor targeting by monovalent and divalent forms of 741F8 anti-c-erbB-2 single-chain Fv. Cancer Res., 53: 4026 – 4034, 1993. 43. Hu, S., Shively, L., Raubitschek, A., Sherman, M., Williams, L. E., Wong, J. Y., Shively, J. E., and Wu, A. M. Minibody. A novel engineered anti-carcinoembryonic antigen antibody fragment (single-chain Fv-CH3) which exhibits rapid, high-level targeting of xenografts. Cancer Res., 56: 3055–3061, 1996. 44. Pastan, I., Lovelace, E. T., Gallo, M. G., Rutherford, A. V., Magnani, J. L., and Willingham, M. C. Characterization of monoclonal antibodies B1 and B3 that react with mucinous adenocarcinomas. Cancer Res., 51: 3781–3787, 1991. 45. Brinkmann, U., and Pastan, I. Immunotoxins against cancer. Biochim. Biophys. Acta, 1198: 27– 45, 1994. 46. Kortt, A. A., Lah, M., Oddie, G. W., Gruen, C. L., Burns, J. E., Pearce, L. A., Atwell, J. L., McCoy, A. J., Howlett, G. J., Metzger, D. W., Webster, R. G., and Hudson, P. J. Single-chain Fv fragments of anti-neuraminidase antibody NC10 containing five- and ten-residue linkers form dimers and with zero-residue linker a trimer. Prot. Eng., 10: 423– 433, 1997. 47. Reist, C. J., Batra, S. K., Pegram, C. N., Bigner, D. D., and Zalutsky, M. R. In vitro and in vivo behavior of radiolabeled chimeric anti-EGFRvIII monoclonal antibody: comparison with its murine parent. Nucl. Med. Biol., 24: 637– 647, 1997. 48. Weiner, L. M., Houston, L. L., Huston, J. S., McCartney, J. E., Tai, M. S., Apell, G., Stafford, W. F., III, Bookman, M. A., Gallo, J. M., and Adams, G. P. Improving the tumor-selective delivery of single-chain Fv molecules. Tumor Target., 1: 51– 60, 1995. 49. Nieroda, C. A., Milenic, D. E., Carrasquillo, J. A., Scholm, J., and Greiner, J. W. Improved tumor radioimmunodetection using a singlechain Fv and g-interferon: potential clinical applications for radioim-


munoguided surgery and g scanning. Cancer Res., 55: 2858 –2865, 1995. 50. Wu, A. M., Chen, W., Raubitschek, A., Williams, L. E., Neumaier, M., Fischer, R., Hu, S. Z., Odom-Maryon, T., Wong, J. Y., and Shively, J. E. Tumor localization of anti-CEA single-chain Fvs: improved targeting by non-covalent dimers. Immunotechnology, 2: 21–36, 1996. 51. Webber, K. O., Kreitman, R. J., and Pastan, I. Rapid and specific uptake of anti-Tac disulfide-stabilized Fv by interleukin-2 receptorbearing tumors. Cancer Res., 55: 318 –323, 1995. 52. Choi, C. W., Lang, L., Lee, J. T., Webber, K. O., Yoo, T. M., Chang, H. K., Le, N., Jagoda, E., Paik, C. H., Pastan, I., Eckelman, W. C., and Carrasquillo, J. A. Biodistribution of 18F- and 125I-labeled anti-Tac disulfide-stabilized Fv fragments in nude mice with interleukin 2a receptor-positive tumor xenografts. Cancer Res., 55: 5323–5329, 1995. 53. Adams, G. P., McCartney, J. E., Wolf, E. J., Eisenberg, J., Huston, J. S., Bookman, M. A., Moldofsky, P., Stafford, W. R., Houston, L. L., and Weiner, L. M. Enhanced tumor specificity of 741F8 –1 (sFv9)2, an

anti-c-erbB-2 single-chain Fv dimer, mediated by stable radioiodine conjugation. J. Nucl. Med., 36: 2276 –2281, 1995. 54. Bobo, R. H., Laske, D. W., Akbasak, A., Morrison, P. F., Dedrick, R. L., and Oldfield, E. H. Convection-enhanced delivery of macromolecules in the brain. Proc. Natl. Acad. Sci. USA, 91: 2076 – 2080, 1994. 55. Lieberman, D. M., Laske, D. W., Morrison, P. F., Bankiewicz, K. S., and Oldfield, E. H. Convection-enhanced distribution of large molecules in gray matter during interstitial drug infusion. J. Neurosurg., 82: 1021–1029, 1995. 56. Ning, S., Trisler, K., Brown, D. M., Yu, N. Y., Kanekal, S., Lundsten, M. J., and Knox, S. J. Intratumoral radioimmunotherapy of a human colon cancer xenograft using a sustained-release gel. Radiother. Oncol., 39: 179 –189, 1996. 57. Rowlinson-Busza, G., Bamias, A., Krausz, T., and Epenetos, A. A. Uptake and distribution of specific and control monoclonal antibodies in subcutaneous xenografts following intratumor injection. Cancer Res., 51: 3251–3256, 1991.