[CANCER RESEARCH 55, 63-70, January 1, 1995]

Development of a Humanized Disulfide-stabilized Anti-pl85HER2 Fv-ß-Lactamase

Fusion Protein for Activation of a Cephalosporin Doxorubicin Prodrug

Maria L. Rodrigues, Leonard G. Presta, Claire E. Kotts, Cindy Wirth, Joyce Mordenti, Gary Osaka, Wai Lee T.Wong, Andrew Nuijens, Brent Blackburn, and Paul Carter1

Departments of Cell Genetics ¡M.L. R., P. CJ, Protein Engineering ¡LP.I. Bio-Analytical Technology ¡C.E. K., C. W.. W. L T. W.. A. N.I, Experimental Therapeutics¡J.M., G. O.I, and Bioorganic Chemistry [B. B.J, Genenlech Inc. South San Francisco, CA 94080-4990

ABSTRACT

The humanized anti-pl85HER2 antibody, humAb4D5-8, has completedPhase II clinical trials for pl85HER2-overexpressing breast cancer. Here,

this antibody is used as a building block to engineer a disulfide-linked Fv(dsFv) ß-lactamase fusion protein for use in antibody-dependent enzyme-mediated prodrug therapy using cephalosporin-based prodrugs. Three Fv

variants were designed with an interchain disulfide bond buried at theVL/VH interface and secreted from Escherichia coli. One variant, dsFv3(VL L46C VH D101C), has similar affinity for antigen (Kd = 0.7 nM) as thewild-type Fv and was used to construct a fusion protein in which ß-lactamase, RTEM-1, is joined to the carboxy terminus of V,,. The dsFv3-ß-lactamase fusion protein secreted from E. coli efficiently activates a ceph-alothin doxorubicin prodrug (PRODOX, kca/km = 1.5 x 10s s'1 NT1).

PRODOX is approximately 20-fold less toxic than free doxorubicinagainst breast tumor cell lines SK-BR-3 and MO 7, which expresspl8SHER2 at elevated and normal levels, respectively. Prebinding the

dsFv3-ß-lactamase fusion protein specifically enhances the toxicity level ofPRODOX to that of doxorubicin against SK-BR-3 but not MCF7 cells.The fusion protein retains both antigen-binding plus kinetic activity in

murine serum and is cleared rapidly as judged by pharmacokinetic analysis in nude mice (initial and terminal half-lives of 0.23 and 1.27 h,respectively). Development and characterization of the dsFv3-ß-lactamase

fusion protein is an important step toward targeted prodrug therapy ofpl85HER2-overexpressing tumors.

INTRODUCTION

The HER2 proto-oncogene (also known as c-erbB2, neu, and HER-21neu) is amplified and/or overexpressed in 20-30% of primary

human breast and ovarian cancers and is a strong prognosticator ofdecreased overall survival and time to relapse (1, 2). Hence thenumerous antibody-based strategies that have been developed aspotential therapeutics for cancers which overexpress the pl85HER2

product of the HER2 gene (reviewed in Ref. 3).The humanized anti-pl85HER2 antibody, humAb4D5-8 (4), has

recently completed Phase II clinical testing for the treatment ofmetastatic breast cancers overexpressing pl85HER2. HumAb4D5-8

has overcome, at least in part, some of the problems which havebroadly stymied the successful use of rodent antibodies for cancerimmunotherapy. For example, the humanized anti-pl85HER2 supports

antibody-dependent cellular toxicity with human effector cells,

whereas the murine parent antibody does not (4, 5). In addition, thelack of antibody response against the humanized antibody has permitted multiple dosing of patients.2 Anticipated potential problems

that remain include the likely small fraction of antibody accumulatingat the target site, binding of antibody to circulating shed antigen, andpotential tumor cell evasion from antibody therapy by down modulation of the target antigen (reviewed in Ref. 6). This has prompted

Received 7/5/94; accepted 10/25/94.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 with18 U.S.C. Section 1734 solely to indicate this fact.

' To whom requests for reprints should be addressed, at Genentech Ine, Department of

Cell Genetics, 460 Point San Bruno Boulevard, South San Francisco. CA 94080-4990.2 S. Baughman, personal communication.

our efforts to use this humanized antibody as a building block todesign potentially more potent immunotherapeutics including a humanized bispecific F(ab')2 fragment for retargeting of cytotoxic T

cells (7, 8), stealth immunoliposomes for targeted drug delivery (9),and dsFv3 enzyme fusion proteins for prodrug activation described

here.ADEPT (6, 10, 11) is a two-step approach in which an antibody-

enzyme fusion protein or conjugate is first allowed to localize to thetumor target. Prodrug is administered after waiting for serum clearance of unbound fusion protein. The prodrug is enzymatically activated by the fusion protein localized to the tumor target. The principalpotential advantage of ADEPT over naked antibodies is amplificationof the cytotoxic activity including against neighboring tumor cells.This bystander effect is anticipated to diminish the risk of tumor cellsescaping therapy via down modulation of the target antigen. In addition the cytotoxic effects of the drug beyond the tumor target shouldbe diminished.

Cephalosporins have proved to be highly versatile triggers in theconstruction of enzyme-activatable prodrugs. For example, the synthesis of cephalosporin-based prodrugs of 5-fluorouracil, 6-mercap-topurine, 6-thioguanine (12), vinca alkaloids (13-15), nitrogen mus

tard drugs (16, 17), a carboplatin analogue (18), and DOX (19, 20).This wide spectrum of antineoplastic agents is possible because substrate specificity of the ß-lactamase-activating enzyme is based on the

cephalosporin part of the prodrug rather than the drug itself. TheADEPT strategy has proved much more efficacious than free drug ina preclinical nude mouse xenograft model for a cephalosporin prodrugof the vinca alkaloid 4-desacetylvinblastine-3-carboxhydrazide (21).

In addition, in vivo antitumor efficacy has recently been demonstratedusing a glucoronide prodrug of DOX in combination with a humanglucoronidase/humanized anti-CEA antibody fusion protein (22).

Here we report the development of a fusion protein comprisinghumAb4D5-8 dsFv fused to the ß-lactamase,RTEM-1. The utility ofthe dsFv ß-lactamasefusion protein to target the prodrug PRODOX(20) to tumor cells overexpressing pl85HER2 is assessed in vitro. The

pharmacokinetics of the fusion protein in nude mice are also investigated with a view to subsequent in vivo testing of the antitumorefficacy of PRODOX.

MATERIALS AND METHODS

Prodox and Dox. PRODOX was synthesized as described by Jungheim etal. (20) with the following modifications. A im-butyl ester of cephalothin was

made by adding rerr-butyl trichloroacetimidate (71.9 mmol) to a solution of

cephalothin sodium salt (23.9 mmol) acidified with anhydrous HCI (24 mmol)in CH2C12 (600 ml). The reaction mixture was stirred overnight at RT and thesolvents were evaporated. The residue was then dissolved in CH2Cl2:EtOAc:

1 The abbreviations used are: ADEPT, antibody dependent enzyme-mediated prodrug

therapy; CDR, complementarity determining region; dsFv, scFv, and wtFv, disulfide-linked, single-chain, and wild-type Fv fragments, respectively; cv, column volumes;DOX, doxorubicin hydrochloride; ECD, extracellular domain; FR, framework region;MTT, 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide; PADAC, 7-(thie-nyl-2-acetamido)-3-[2-(4-N,N-dimethylaminophenylazo)pyridinium-methyl-3-cephem-4

carboxylic acid; PMSF, phenylmethylsulfonyl fluoride; PRODOX, cephalothin prodrug ofdoxorubicin; RT, room temperature.

63

Research. on August 12, 2019. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

TOWARD ADEPT FOR pl85HER2 OVEREXPRESSING TUMORS

hexane (1:1:3) and purified on a silica gel column using the same solventsystem. The ester (0.23 mmol) was deprotected by treatment with trifluoro-

acetic acid (20%, v/v) and triethylsilane (0.23 mmol) in CH2C12 (42 ml) for 5h at RT. The solvents were evaporated and the cephem sulfoxide free acid wasthen added to a solution of DOX (0.1 mmol; Fluka, Buchs, Switzerland) intetrahydrofuran (42 ml) and aqueous NaHCO3 (4 ml, 0.1N) and stirred for 3 hat RT. The product was extracted with 10% (v/v) butanol in ethyl acetate andpurified on a C]K column under isocratic conditions (15%, v/v acetonitrile).The PRODOX was characterized by high resolution mass spectroscopy and 'Hand I3C nuclear magnetic resonance spectroscopy. 'H data agreed closely with

those reported by Jungheim et al. (20). Stock solutions of 1 mg/ml DOX and1 mg/ml PRODOX were prepared in PBS for kinetic and cytotoxicity assaysand either used immediately or stored at -70°C prior to use.

Molecular Modeling of Disulfide Bonds in humAb4D5-8 Fv. Potentialinterchain disulfide bonding pairs were identified in the X-ray crystallographicstructure of humAb4D5-8 Fv (23). First pairs of residues, one each in VL and

VH, were identified where the corresponding Ca separation was 5.0 to 7.3 A(24). The upper limit of Ca separation (—6.8A) for disulfide-bonding residues

was increased by 0.5 A in order to include pairs in which slight backbonemovement would juxtapose corresponding Ca within 6.8 A. Cysteine replacements were modeled for each residue pair where the side chains were pointedtoward one another using the Insight II program (Biosym Technologies, SanDiego, CA). The cysteine side chains were rotated about their Ca—Cßbond

to determine whether any set of conformations could provide for an acceptabledistance (<3 A) between the two Sy atoms and acceptable geometry (24). Thecysteine and adjacent residues were manipulated in order to bring the S-yatoms

to the optimal bonding distance of 2.0 A for residue pairs satisfying thesecriteria (Fig. 1).

Construction of humAb4D5-8 wtFv and dsFv Expression Plasmids.The humAb4D5-8 Fv expression plasmid pAK20 (Fig. 2A) was constructedfrom plasmid pAK19 (25). Briefly, the humAb4D5-8 Fab' expression unit

from pAK19 was subcloned into pUCl 19 (26) and the gene segments encodingCL and CH1 were then simultaneously deleted using the oligonucleotidesMLR 1 5 ' -CAAGGTGGAGATCAAACGAACTTAAGCTGATCCTCTACG-CCGG-3' and MLR2 S'-CCCTGGTCACCGTCTCCTCGTGACCACCGCA-TGCAAGCTTG-3', respectively, using a high efficiency mutagenesis proce

dure (27). After nucleotide sequence verification the Fv expression unit wassubcloned back into pAK19 to create pAK20. HumAb4D5-8 dsFv variants

phoA stlli terH-

ß-lactamase

C ß-lactamase

phoA stll i ter

SCORI Hindin

Fig. 2. Schematic representation of synthetic opéronsfor expression of Fv variants (A),dsFv-ß-lactamase fusion proteins (B), and ß-lactamase (C). In each case, expression is

under the transcriptional control of the E. coli alkaline phosphatase promoter (phoA),which is inducible by phosphate starvation. The heat-stable enterotoxin II (stlf) signalsequence is used to direct secretion to the periplasmic space of E. coli. The transcriptionalunits are flanked on the 3' side by the At,, transcriptional terminator (ter) (24).

were constructed by simultaneous site-directed mutagenesis of V[ and VH after

subcloning the wtFv expression unit from pAK20 into phagemid pmy95 (28).Briefly, dsFvl (VL P44C VH W103C), dsFv2 (VL F98C VH L45C), and dsFv3(VL L46C VH D101C) were constructed by site-directed mutagenesis usingcorresponding pairs of the oligonucleotides: MLR3 VL P44C 5'-CAGAAAC-CAGGAAAgGCctgtAAACTACTGATTTAC-3'; MLR4 VH W103C 5'-GA-CGGCTTCTATGCcATGGACTACtgtGGTCAAGGAACC-3'; MLR5 VLF98C 5'-ACTACTCCTCCCACGTgtGGACAGGGcACCAAGGTGGAG-3';MLR6 VH L45C 5'-TGGGTGCGTCAGGCaCCGGGTAAGGGCtgtGAAT-GGGTTGCA-3'; MLR7 VL L46C 5'-GGAAAAGCTCCGAAAtgtCTGATT-TACTCG-3'; and MLR8 VH D101C S'-GCCGTCTATTATTGcTCgAGAT-GGGGAGGGGACGGCTTCTATGCTATGtgtTACTGGGGTCAA-3'; where

targeted mutations are indicated by lower case, and amino acid changes aredenoted by the amino acid residue and number followed by the replacement,and utilize the numbering scheme of Kabat et al. (29). After sequence

Fig. 1. Siructural model for humAb4D5-8 dsFvbased on the X-ray crystallographic structure ofhumAb4D5-8 wtFv (22). The VH FR (dark blue)and CDR (light blue) residues and V, FR (red) andCDR residues (salmon) are shown. Modeled in arethree disulfide bonds: VL P44C V„W103C (yellow), VL F98C VH L45C (gold), and V, L46C VHD101C (green), which were separately installedinto the wtFv to create dsFvl, dsFv2, and dsFv3,respectively.

64

Research. on August 12, 2019. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

TOWARD ADEPT FOR pl85HER2 OVEREXPRESSING TUMORS

verification the dsFvl, dsFv2, and dsFvS variants were subcloned into pAK20to create pAK20.1, pAK20.2, and pAK20.3, respectively.

Expression and Purification of humAb4D5-8 Fv Variants.HumAb4D5-8 wtFv was secreted from E. coli strain 35E6 [iVvGRderivative of

25F2, (25)] containing the plasmid pAK20 grown for approximately 32 h at30°C in an aerated 10-liter fermentor as described previously (25). The

humAb4D5-8 dsFv variants were secreted from E. coli strain 33B6 [W3110fo/iA ¿>hoAAE15deoC KanR ilvGR degP A(argF-/ac)169] containing corre

sponding expression plasmids under similar fermentation conditions.The wtFv and dsFv fragments were released from 200 g of corresponding

fermentation pastes by thawing in the presence of 300 ml 20% (w/v) sucrose,50 mM Tris-HCl (pH 8.0) containing 10 min EDTA-1 mM PMSF (Sigma) over

approximately 60 min with gentle stirring. After centrifugation (30,000 X g for10 min at 4°C)to remove cell debris, the resultant supernatant was passed over

60 ml DEAE fast flow sepharose (Pharmacia Biotech, Piscataway, NJ) andloaded onto a 50 ml protein A-controlled pore glass column (ProSepA;

Bioprocessing Ltd., Durham, UK). The column was washed extensively withPBS, and Fv variants were then eluted with 100 mM acetic acid, pH 2.8. Theeluted material was adjusted to pH ~5 with 1.0 M Tris-HCl, pH 8.0, and then

loaded on to a mono S column (HR 10/10, Pharmacia). Fv variants were elutedwith a gradient of 0-100 mM NaCl in 10 mM MES, pH 5.5, over 20 cv. Purified

Fv variants were buffer exchanged into PBS by gel filtration (Econo columns;BioRad, Hercules, CA) and their concentration estimated from the absorbanceat 280 nm and the extinction coefficient of wtFv determined by amino acidcomposition analysis (e2go = 5.0 X IO4 cm"1 M"1).

Construction of Expression Plasmids for ß-Lactamase and dsFv-ß-Lactamase Fusion Proteins. DNA encoding the fusion protein dsFv3-ß-lactamase was constructed by in-frame joining of the 5' end of the geneencoding RTEM-1 ß-lactamasewith the 3' end of the gene encoding the VHdomain (Fig. 2B). Briefly, sites for Apal and Nhel were installed at the 5' and

3' ends, respectively, of ß-lactamasegene in pmy95 by site-directed mutagen-esis using the oligonucleotides MLR9 5'-GTAACCCACTCgggcccCCAACT-GATC-3' and MLR10 5'-AAACTTGGTCgctagcTTACCAATGC-3'. The 3'

end of VH of dsFv3 was then modified by site-directed mutagenesis using theoligonucleotide MLR11 S'-GTCACCGTCTCCTCGcacccagaaacgctggtgaaa-gtaaaagatgctgaagatcagttgggggcccCCACCGCATGCGAC-3'. After sequence

verification, the mutagenized ß-lactamase and dsFv3 fragments were sub-

cloned into a pmy95 derivative in which the ampicillin resistance gene was

replaced by the chloramphenicol resistant gene from pACYC184 (30).A control fusion protein, dsFvS.l-ß-lactamase, was constructed by install

ing the VH mutations W95A and YlOOaA into the dsFv3-ß-lactamase using theoligonucleotide MLR125 5:-GCCGTCTATTATtgtTCGAGAgcaGGAGGGG-ACGGCTTCgcaGCTATGTGTTAC-3'. This pair of mutations has beenshown to decrease the binding affinity of humAb4D5-5 Fab for pl85HER2 BCD

by > 1,000-fold (31).A unique Nhel site was inserted into the ß-lactamase gene of pUC119 by

mutagenesis using the oligonucleotide MLR13 5'-AAACTTGGTCgctagcT-TACCAATGC-3', facilitating precise fusion of the ß-lactamasegene to the 3'

end of the stll signal sequence in pmy95.Expression plasmids pRZl, pRZ2, and pRZ3 were constructed by subclon-

ing dsFv3-ß-lactamase, dsFv3.1-ß-lactamase, and ß-lactamase-encoding vari

ants, respectively, into a pAK19 derivative in which the ampicillin resistantgene is replaced with the chloramphenicol resistant gene from pACYC184 andthe tetracycline resistance gene is insertionally inactivated by subcloning.

Expression and Purification of dsFv-ß-Lactamase Fusion Proteinsand ß-Lactamase. HumAb4D5-8 dsFv-ß-lactamase fusion proteins andß-lactamasewere secreted from E. coli strain 27C7 [W3110 ronAp/ioAAElSdeoC KanR degf ompT ptr3 A(argF-/ac)169] containing corresponding ex

pression plasmids grown under the same conditions as the dsFv variants in the

fermentor.The dsFv-ß-lactamase fusion proteins were released from 200 g fermenta

tion pastes by thawing in the presence of 300 ml 20 mM MES, pH 5.5,containing 1 mM PMSF for approximately 30 min with gentle stirring. Fifty figof hen egg white lysozyme (Canadian Lysozyme, Inc., Abbortsford, Canada)was then added (25) and further stirred for 60 min. Cell debris was removedby centrifugation (30,000 X g for 10 min at 4°C).Polyethyleneimine was then

added to the supernatant to a final concentration of 0.2% (v/v) and precipitateremoved by centrifugation (30,000 X g for 10 min at 4°C).The supernatant

was loaded onto a 50-ml Bakerbond ABx column (J. T. Baker, Inc., Philips-

burg, NJ), equilibrated and extensively washed with 20 mM MES, pH 5.5, andthen eluted with a linear gradient of 0-150 mM (NH4)2SO4 in 20 mM MES, pH5.5, over 20 cv. The eluted fractions containing ß-lactamase activity were

identified by following the change in absorbance at 570 nm on hydrolysis ofthe chromogenic substrate, PADAC (Calbiochem, San Diego, CA). Pooledfractions were adjusted to pH 7.5 with 1.0 M Tris-HCl, pH 8.0, and NaCl wasadded to a final concentration of 0.5 M. The sample was loaded onto a 5-ml

phenylboronic acid affinity column and eluted with 0.5 M borate, pH 7,containing 0.5 M NaCl as described by Cartwright and Waley (32). The elutedmaterial was further purified by using a protein A column and buffer exchanged into PBS (see above). The concentration of purified dsFv-ß-lactamase

fusion protein was estimated from the absorbance at 280 nm and theextinction coefficient determined by amino acid composition analysis(e28() = 1.2 X IO5 cnT1 M'1).

ß-lactamase was purified from a corresponding fermentation paste as de

scribed for the fusion proteins with the following modifications. The supernatant after removal of the PEI precipitate was incubated with 50 ml DEAEfast flow sepharose resin overnight at 4°C.The resin was washed with 10 mM

Tris-HCl, pH 8, containing 1 mM EDTA and eluted with a gradient of0-0.2 M NaCl in 10 mM Tris-HCl, pH 7.5, over 8 cv. ß-lactamase-

containing fractions were purified using a phenylboronic acid affinitycolumn and buffer exchanged into PBS (see above). The concentration ofpurified ß-lactamase was estimated from the absorbance at 280 nm[e280 = 2.94 X IO4 cm'1 M'1 (33)].

Mass Spectrometry. The molecular masses of Fv variants and dsFv-ß-

lactamase fusion proteins were estimated by electrospray mass spectrometry asdescribed previously (34).

Kinetic Procedures. Kinetic parameters for purified ß-lactamase anddsFv3-ß-lactamase with the substrates cephalothin (Sigma) and PRODOX

were determined by the method of initial rates using the extinction coefficients4 Ae260 = 7.7 X IO3 cm'1 M'1 (35) and Ae260 = 1.3 X IO4 cm"1 NT1,

respectively. Kinetic assays were performed in PBS at 25 ±0.2°Cusing a

0.2-cm path length cuvette and a Kontron Uvikon 860 spectrophotometer.

Initial rates of hydrolysis of cephalothin (25 /XMto 1.24 mM) or PRODOX(47-460 /AM)in the presence of either 3.6 nM ß-lactamase or 4.8 nM dsFv3-

ß-lactamasewere determined by monitoring the reduction in absorbance at 260

nm over approximately 1.5 min. kM and kcal were estimated using a nonlinearleast squares fit of the initial rate data to the Michaelis-Menten equation using

the program Kaleidagraph version 3.0 (Synergy Software, Reading, PA).Kd Determination. Binding of Fv and dsFv ß-lactamase variants to

pl85HER2ECD (36) was followed by surface plasmon resonance using the

BIAcore system (Pharmacia Biosensor AB, Uppsala, Sweden) as describedpreviously (37).

In Vitro Cytotoxicity. SK-BR-3 and MCF7 breast carcinoma cells (American Type Culture Collection, Rockville, MD) were cultured in DMEM:Ham's

nutrient F-12 (50:50) supplemented with 2 mM glutamine, 100 units/ml penicillin, 100 fig/ml streptomycin (GIBCO-BRL, Grand Island, NY), and 10%(w/v) bovine fetal serum (Hyclone, Logan, UT) (cell media) at 37°C,5% CO2.

Cells were seeded at 12,000 cells/well (SK-BR-3) or 8,000-10,000 cells/well(MCF7) in 96-well tissue culture plates (Falcon; Becton-Dickinson, Franklin

Lakes, NJ) and allowed to attach for ^14 h. Cell medium was aspirated and

replaced with fresh medium (100 /jl/well) in the absence or presence ofdsFv3-ß-lactamase (10 /xg/ml), dsFv3.1-ß-lactamase (10 jiig/ml), or ß-lactamase (5.4 ng/ml). After incubation at 37°C,5% CO2 for 2 h, the plates werewashed three times with cell medium (37°C).Test media consisting of 0-17

/LIMDOX or 0-17 IJLMPRODOX were added (200 u,l/well) and the platesincubated at 37°C,5% CO2, for 2 h. Plates were washed twice with cellmedium (37°C)and further incubated for a total assay length of 3 days. The

assay was terminated by staining with 0.5% (w/v) crystal violet in methanol,and absorbance was read at 540 nm (SLT 340 ATTC plate reader; SLT LabInstruments, Salzburg, Austria). Very similar results were obtained by themore direct but more laborious methods of alamar blue assay, MTT staining ordirect cell counting (not shown).

Pharmacokinetics of dsFv3-ß-Lactamase. Forty-Two female NIH IIIbeige nude mice (16-21 g; Charles River, Wilmington, MA) received a singletail vein injection (3 mg/kg) of dsFv3-ß-lactamase (0.54 mg/ml) in PBS.

Groups of three mice were sacrificed at scheduled times ranging from 1 min to10 h postinjection. Blood (~0.5 ml) was collected following jugular exsan-

guination or cardiac puncture, serum obtained using Microtainer Serum

65

Research. on August 12, 2019. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

TOWARD ADEPT FOR pl85HER2 OVEREXPRESS1NG TUMORS

Separators (Becton-Dickinson, San Lorenzo, Puerto Rico), and stored frozen at-70°C. Serum concentrations of the dsFv3-ß-lactamase fusion protein were

then determined by ELISA (see below). The time course of serum concentrations of dsFv3-ß-lactamase was plotted, and a biexponential function wasfitted to the data using a nonlinear least squares regression program, PCNON-

LIN v4.2 (Statistical Consultants, Lexington, KY).ELISA Procedure for dsFv3-ß-Lactamase. Ninety-six-well microtiter

plates (Maxisorb; Nunc, Kamstrup, Denmark) were coated with pl85HER2

BCD, blocked with BSA, and washed with 0.05% (v/v) Tween 20 in PBS(wash buffer) as described previously (37). A dsFv3-ß-lactamase standard andsamples were diluted in wash buffer containing 0.5% (w/v) BSA-0.01% (w/v)

thimerosal and dispensed onto the coated wells (100 fil/well). Plates weresealed and incubated at room temperature for 2 h with gentle agitation. Plateswere washed six times with wash buffer before adding substrate solutioncontaining 160 /j,g/ml of PADAC in PBS (100 /¿I/well). After overnightincubation at room temperature, the resulting absorbance at 595 nm wasdetermined on a Vmax plate reader (Molecular Devices, Menlo Park, ÇA).Astandard curve was generated by plotting absorbance versus the log of dsFv3-ß-lactamase concentration (2,000-180 pg/ml), using a 4-parameter nonlinearregression curve-fitting program (developed at Genentech Inc.). Sample con

centrations were estimated by interpolation of their absorbance on the standard

RESULTS

Design of humAb4D5-8 dsFv Variants. Four criteria were used to

identify pairs of residues for engineering a disulfide bond between VLand VH domains in the X-ray crystallographic structure ofhumAb4D5-8 Fv (23): (a) The Ça separation should be similar tothose found in natural disulfide bonds (5.0-6.8 A) (24); (b) bothresidues should be fully buried at the VH-VL interface to minimize the

risk of disulfide bond cleavage in serum; (c) the residues shouldpermit disulfide bonding with favorable geometry and likely little orno strain (24); and (d) ideally both residues should be in the FR tominimize the risk of impairing antigen binding.

Of 20 residue pairs satisfying the distance criteria, only 2fulfilled the other 3 criteria and were chosen for construction:dsFvl VL P44C VH W103C and dsFv2 VL F98C VH L45C. Anadditional variant, dsFv3 VL L46C VH D101C, was constructedwhich satisfies these criteria except that the VH partner is a CDRresidue (Fig. 1).

Expression, Purification and Characterization of huruAb-H)5-8

Fv Variants. Secretion of Fv variants from E. coli was accomplishedby coexpression of corresponding VL and VH variants from a synthetic dicistronic operon (Fig. 2A) very similar to that described forhumAb4D5-8 Fab' (25). Fv variants were purified from correspond

ing fermentation pastes by affinity purification using Staphylococcalprotein A followed by mono S ion exchange chromatography.

The Fv variants were analyzed by SDS-PAGE (Fig. 3). Under

nonreducing conditions wtFv gave two major bands with the expectedelectrophoretic mobility for VL (Mr -12,000) and VH (Mr -13,000)

domains respectively. In contrast, the three dsFv variants gave singlebands with mobilities close to that expected (Mr —25,000). The

differences in electrophoretic mobility of the dsFv variants likelyreflect the impact of the disulfide bonds on unfolding and SDSbinding rather than mass differences. The molecular masses of dsFvand wtFv were determined precisely by electrospray mass spectrom-

etry and found to agree closely with those expected (Table 1).On reduction, all Fv variants gave two major bands with the

expected electrophoretic mobility of VL and VH. DsFvl and dsFv3 butnot dsFv2 were readily reduced with 100 mM DTT (Fig. 3B). Fullreduction of dsFv2, however, could be achieved with the strongerreducing reagent, tributyl phosphine (not shown).

-DTT1 234567

kD92-66-45-

31 -

21 -

14-

+DTT1 234567

kD92-66- —•45-

31 -

21 -

14- ^.

Fig. 3. SDS-PAGE analysis of purified Fv and dsFv-ß-lactamase variants together withß-lactamase. Proteins (2-5 /¿g/lane)were electrophoresed on a 16% gel (Novex, San

Diego, CA) under nonreducing (top) or reducing conditions (bottom) and stained withCoomassie blue R-250. Lane 1, wtFv; Lane 2, dsFvl; Lane 3, dsFv2; Lane 4, dsFv3; Lane5, dsFv3-/3-lactamase; Lane 6, dsFv3.1-ß-lactamase; Lane 7, ß-lactamase. kD, molecularweight in thousands.

The dsFv variants had similar (dsFv3) or 3-fold reduced (dsFvl anddsFv2) binding affinity for pl85HER2 BCD as compared to the wtFv

(Table 1). This lead to the choice of dsFv3 for construction of fusionproteins with ß-lactamase. In addition, the interchain disulfide in

dsFv3 is formed in 95% of molecules compared to 53% for dsFvl and55% for dsFv2 as judged by scanning densitometry of a Coomassie-stained gel of the protein A-purified pool (data not shown).

Expression, Purification and Characterization of dsFv-ß-Lac-taniase Variants and ß-Lactamase. Secretion of the targeting fusion protein, dsFv3-ß-lactamase, and the non-pl85HER2-binding

fusion protein, dsFv3.1-ß-lactamase, from E. coli was accomplishedby coexpression of corresponding VL and VH-ß-lactamase variants

Table 1 Characterization of humAb4D5-8 Fv and dsFv-ß-lactamase variants

4 M. L. Rodrigues, unpublished data.

humAb4D5-8

variantwtdsFvldsFv2dsFv3dsFv3-ß-lactamasedsFv3.

1-ß-lactamaseInterchain

disulfideNoneVLC44VLC98VLC46V(

C46VLC46VH

C103VHC45VHC101VHC101VH

C101Kd

(DM)0.61.82.00.71.0>1,000MassExpected13,16311,95425,03525,05825,090ND"NDObserved13,16311,95225,03525,05925,08912222NDND

" ND, not determined.

66

Research. on August 12, 2019. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

TOWARD ADEPT FOR plK5HER: OVEREXPRESSING TUMORS

Table 2 Aclivin nfdsFv3 ß-luaamase anil ß-laclamase RTEM-1 with Cephalothin"and PRODOX1'

1.5

Enzyme Substrate kca, s" V*",s"1 M'1

ß-lactamasc RTEM-1 cephalothin 269 ±27 346 ±6PRODOX 54 ±5 199 ±45

dsFv3-ß-lactamase cephalothin 239 ±24 289 ±14PRODOX 25 ±3 165 ±60

7.8 X 10'2.7 X l O5

8.3 X IO51.5 X IO5

C02®

from a synthetic discistronic operon (Fig. 2ß),whereas ß-lactamase

itself was expressed from a monocistronic operon (Fig. 2C). Proteinpurification from E. coli fermentation pastes was facilitated by affinitychromatography using Staphylococcal protein A (dsFv-ß-lacta-mase variants) and phenyl boronate (dsFv-ß-lactamase variantsand ß-lactamase).

The dsFv-ß-lactamase variants and ß-lactamasewere analyzed bySDS-PAGE (Fig. 3). Under nonreducing conditions, the dsFv-ß-lactamase variants gave one major band with the expected electro-phoretic mobility (A/r —54,000). Under reducing conditions the dsFv-

ß-lactamase variants gave two major bands with the expectedelectrophoretic mobilities of VL (Mr ~ 12,000) and V,, ß-lactamase(Mr —42,000).The electrophoretic mobility of free ß-lactamase is

similar under reducing and nonreducing conditions, despite the anticipated reduction of the intrachain C27-C123 disulfide bond.

Activities of ß-Lactamase and dsFv3-ß-Lactamase withPRODOX and Cephalothin. The dsFv3-ß-lactamasefusion proteinand ß-lactamasehave similar kinetic activities against the PRODOX

prodrug and the parent cephalosporin, cephalothin (Table 2). FordsFv3-ß-lactamase and ß-lactamasethe catalytic efficiency (kca,/A:m)is lower against PRODOX than cephalothin by approximately 3-foldand 5-fold, respectively.

Cytotoxicity of PRODOX and DOX against SK-BR-3 and

MCF7 Cells. The cytotoxicity of free PRODOX and DOX wereinvestigated against the breast cancer cell lines SK-BR-3 and MCF7

following exposure of the cells for 2 h to PRODOX or DOX. Thenumber of cells present after a total of 72 h incubation was thenestimated by crystal violet staining. PRODOX is approximately 20-fold less toxic than free DOX against both SK-BR-3 (Fig. 4-4) and alsoMCF7 cells (Fig. 4ß)which express pl85HER2 at 900 ng/mg and 7

ng/mg total protein, respectively (5). A similar difference in toxicitybetween prodrug and drug was observed for all of the other cell linestested to date (not shown): MDA-MB-468 (breast tumor); Wl-38(lung epithelium); and SK-OV-3 (ovarian tumor), which expresspl85"hli2 at undetectable; low, and intermediate (540 ng/mg total

protein) levels, respectively (5).The cytotoxicity of targeted PRODOX was investigated against

SK-BR-3 and MCF7 cells. In this case the cells were preincubatedwith dsFv3-ß-lactamase, dsFvS.l-ß-lactamase, or ß-lactamasefor 2 h

and then washed prior to exposure with PRODOX, followed byfurther incubation and crystal violet staining as described above. For

1 •

0.5-

Eeoin

O

SK-BR-3

oo>oI 1.5oM

10 -9 10 -8 10 -7 10 -6 10 -51CT

0.5

MCF7 'O—O—O

1CT8 10 -7 10 -6 10 -5 10 -4 10 -3

[Cytotoxic Agent] MFig. 4. In vitro eytoloxic effect of DOX and targeted PRODOX against (A) SK-BR-3

and (ß)MCF7 cells. The cells were preincubated with dsFv3-ß-Iactamase (•),dsFv3.1ß-lactamase (•).or ß-lactamase (D) for 2 h and then washed prior to exposure with

PRODOX for 2 h, or alternatively exposed directly to PRODOX (A) or DOX (O) alunefor 2 h. In each case the assay was terminated after a total of 72 h incubation prior tocrystal violet staining as described in "Materials and Methods." Points, mean; bars, SD.

SK-BR-3, PRODOX is equally toxic to the free DOX for cellsprebound to the dsFv3-ß-lactamase fusion protein (Fig. 4A). In contrast to SK-BR-3, PRODOX toxicity is not enhanced by prebindingMCF7 cells with the dsFv3-ß-lactamase fusion protein (Fig. 4B).

Toxicity of PRODOX after pretreating the SK-BR-3 cells witheither ß-lactamase or the non-pl85HKR:!-binding fusion protein

dsFvS.l-ß-lactamase is intermediate between that of DOX or PRO

DOX alone. This intermediate level toxicity likely reflects a smallamount of nonspecific protein binding exacerbated by the high catalytic efficiency of ß-lactamaseas observed previously by others (17).

Thus, precautions to circumvent nonspecific binding by preblockingthe plates with medium containing 10% (v/v) bovine PCS were onlypartially successful. Nonspecific binding is predominantly to plasticrather than to cells as judged by incubating preblocked plates in theabsence or presence of cells with either dsFv3.1-ß-lactamase or ß-lactamase followed by plate washing and detection of residual ß-lacta

mase activity with the chromogenic substrate, PDAC.Pharmacokinetics of dsFv3-ß-Lactamase Fusion Protein in

Nude Mice. The time course of the dsFv3-ß-lactamasefusion protein

in the serum of nude mice was determined from serial sacrifice ofmultiple animals (three per time point) followed by antigen-bindingELISA using the chromogenic ß-lactamasesubstrate, PADAC, as a

reporter (Fig. 5). From these data, the following pharmacokineticparameters estimates were obtained: serum clearance, 133 ml h"'kg~'; initial volume of distribution, 44.9 ml kg"1; steady-state volumeof distribution, 47.4 ml kg~'; initial half-life, 0.23 h; and terminal

half-life, 1.27 h.

67

Research. on August 12, 2019. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

TOWARD ADEPT FOR p]85HER! OVEREXPRESSINO TUMORS

100

CO.0.01-

0.00014 6

Time (h)10

Fig. 5. Serum concentration of dsFv3-ß-lactamase in female NIH III beige nude micefollowing a single tail vein injection (3 mg/kg). Shown are the experimental data (•)estimated by antigen binding ELISA together with a nonlinear least squares fit ( ):

C, = 66.9e-'"»' + 0.190e-°546'

where C, is the serum concentration ((ig/ml) at time. / (h), after injection.

DISCUSSION

We have developed a fusion protein comprising a dsFv fragment ofthe humanized anti-pl85HfcR2 antibody. humAb4D5-8, fused via VHto the ß-lactamase,RTEM-1, for use in ADEPT of pl85HER2 over-

expressing tumors. The first design step was to stabilize the wtFvagainst chain dissociation (Kd, —10nM).5 Single chain constructions

have been widely used to stabilize Fv fragments including scFv-ß-

lactamase fusion proteins (38, 39); however, interchain disulfidebonds involving either CDR (40) or FR (40-43) residues have proved

more effective than single chains as means of stabilizing rodent Fvfragments and have similar or improved antigen binding. Additional advantages of dsFv over scFv include enhanced serumstability, reduced tendency to aggregate, and greater ease of intra-cellular production [although not secretion (40)] in E. coli (41-43).

Therefore, we chose disulfide engineering as a strategy to stabilizehumAb4D5-8 wtFv.

Three humAb4D5-8 dsFv were designed by molecular modelingusing the X-ray crystallographic structure of the wtFv (23) choosing

novel sites for disulfide formation: dsFvl, VL P44C VH W103C;dsFv2, VL F98C VH L45C; and dsFv3, FL L46C VH D101C. Functional dsFv was obtained in each case by secretion from E. coli. dsFv3was chosen for fusion protein construction because it binds pl85HER2

BCD with similar affinity to the wtFv and the corresponding interchain disulfide is very readily formed. Thus, in the case of dsFv3 itwas possible to utilize one CDR residue (VH D101) for cysteinereplacement without compromising antigen binding.

Unlike previous dsFv (40-43) we chose residues that are fully

buried at the VL/VH interface in an attempt to protect the resultantdisulfide from cleavage in serum. The six selected residues (five FR,one CDR: VH D101) are conserved between humAb4D5-8 and other

humanized antibodies that we have built using the same human FR:CD3 (7, 8), IgE (44). and CD18 (45). This bodes well for constructionof dsFv from these different humanized antibodies.

The dsFv3-ß-lactamase fusion protein efficiently activates PRO-DOX and enhances its toxicity against the pl85HER2-overexpressing

breast tumor cells, SK-BR-3, by 20-fold to a level indistinguishable

from DOX. In contrast, a related cephalosporin prodrug of the vincaalkaloid 4-desacetylvinblastine-3-carboxhydrazide is only 5-fold less

toxic than the corresponding free drug (13) against tumor cells in vitrobut is highly efficacious in a tumor xenograft model in vivo (21). Thisencourages us to test the antitumor efficacy of the dsFv3-ß-lactamaseand PRODOX in tumor-bearing nude mice.

Fully functional dsFv3-ß-lactamase fusion protein is rapidlycleared from murine serum as judged by antigen-binding ELISA usinga chromogenic substrate for ß-lactamase. This should permit PRO

DOX administration a few hours following fusion protein withoutsignificant enzymatic activation of PRODOX by residual circulatingfusion protein. In addition to the demonstrated rapid clearance, thesmall size of dsFv3-ß-lactamase (Mr 54,000) is anticipated to promote

tumor penetration leading to high tumor to nontumor ratios, albeit atthe expense of low tumor localization rates. Future in vivo studies willtest these predictions. Biodistribution studies of mono and divalentversions of an anti-pi85HER2 scFv in tumor xenografts in seid mice

have shown significant tumor localization ( 1.0% injected dose/at 24 h)and favorable tumor: nontumor ratios (46).

Tumors overexpressing pl85HER2 are well suited to ADEPT.

Firstly, synergistic or at least additive antitumor effects have beenseen in nude mice xenograft modes on combining antibodies directedagainst pl85HRR2 antibodies and the chemotherapeutics cisplatin (47,48) and taxol (49). In addition pl85HER2 overexpression may identify

patients that are most responsive to high dose adjuvant chemotherapy(50). Regression of breast tumor xenografts has recently been demonstrated by ADEPT directed against pl85HER2 (51). In this system

ADEPT proved both more efficacious and less toxic than conventional chemotherapy.

The tumor targeting specificity of dsFv3-ß-lactamase requires in

vestigation since HER2 is expressed on the surface of a variety ofepithelial cells, albeit at much lower levels than found in many breastand ovarian cancers (52). Here the dsFv3-ß-lactamase fusion protein

is demonstrated in vitro to target efficiently PRODOX against thebreast tumor cell line SK-BR-3 but not MCF7, which expresspl85HER2 at elevated and normal levels, respectively. In addition, no

specific targeting was observed for the normal lung epithelium, WI-38, which expresses pl85HKR2 at similar levels to MCF7.6 Fortunately

there is no detectable expression of HER2 on heart tissue, which is asite of chronic and dose-limiting toxicity for DOX, nor on bone

marrow, which is a site of acute toxicity (53).Repeated clinical use of fusion proteins for ADEPT may be con

founded by an immune response mounted against the xenogeneiccomponents. This limitation has long been recognized and the use ofhumanized antitumor and catalytic antibodies has been proposed as apossible solution (54). The advent of large human antibody phagelibraries (55), together with advances in the development of catalyticantibodies (56) and efficient routes for constructing bispecific antibodies (57), significantly enhance the feasibility of such an approach.Human enzymes offer an alternative to human catalytic antibodies,albeit with potential drawbacks of unwanted activation of prodrug byendogenous enzyme, interference from endogenous substrates or inhibitors, and potential immunogenicity if intracellular enzymes areselected. Nevertheless, a humanized anti-CEA antibody fused to a

human enzyme, glucuronidase, in conjunction with a glucuronideprodrug of DOX was found to be efficacious in tumor-bearing nude

mice (22), which augurs well for this approach.

s R. Kelley, personal communication. 5 C. Wirth. unpublished data.

68

Research. on August 12, 2019. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

TOWARD ADEPT FOR pl85HE»- OVEREXPRESSING TUMORS

The development of a dsFv-ß-lactamase fusion protein described

here is an important step toward ADEPT of human tumors overexpressing the HER2 proto-oncogene.

ACKNOWLEDGMENTS

We thank our Genentech colleagues, Mark Yasser, Parkash Jhurani, andPeter Ng for oligonucleotide synthesis; Brad Snedecor for E. coli fermentations; Allan Padua for amino acid composition analysis; Allison Nixon, AnneWalters, and Christine Fratino for animal husbandry and expert technicalassistance with pharmacokinetic experiments; Sharon Baughman and RobertKelley for sharing unpublished data on the immunogenicity of the anti-pl85HFR2 antibody and stability of wtFv, respectively; Jim Bourell for mass

spectrometry data; John W. Park and Mark Troll for helpful discussions;Wayne Anstine for preparing figures 3—5;and Paul Godowski for unstinted

support.

REFERENCES

Slamon, D. J., Clark, G. M., Wong, S. G., Levin, W. J., Ullrich, A., and McGuire.W. L. Human breast cancer: correlation of relapse and survival with amplification ofthe HER-2/flpH oncogene. Science (Washington DC), 235: 177-182, 1987.Slamon, D. J., Godolphin, W., Jones, L. A., Holt, J. A.. Wong, S. G., Keith, D. E.,Levin, W. J., Stuart, S. G., Udove, J., Ullrich, A., and Press, M. F. Studies of IheHER-2/n«i proto-oncogene in human breast and ovarian cancer. Science (Washington DC), 244: 707-712, 1989.

Carter, P., Rodrigues, M. L., Lewis, G. D., Figari, I., and Shalaby, M. R. Towards animmunotherapy for pl85HER2 overexpressing tumors, in: Ceriani, R. L. (ed.), Antigenand Antibody Engineering in Breast Cancer Diagnosis, pp. 83-94. New York: Plenum

Publishing Corp., 1994.Carter, P., Presta, L., Gorman, C. M., Ridgway, J. B. B., Henner, D., Wong, W. L. T.,Rowland, A. M., Kotts, C., Carver, M. E., and Shepard, H. M. Humanization of ananti-pl85"ER2 antibody for human cancer therapy. Proc. Nati. Acad. Sci. USA, 89:4285-4289, 1992.

Lewis, G. D., Figari, I., Fendly, B., Wong, W. L., Carter, P., Gorman, C., andShepard, H. M. Differential responses of human tumor cells lines to anti-pi85HER2

monoclonal antibodies. Cancer Immunol. Immunother., 37: 255—263,1993.Sedalacek, H-H., Seemann, G., Hoffmann, D., Czech, J., Lorenz, P., Kolar, C., and

Bosslet, K. (eds.). Antibodies as carriers of cytotoxicity. Contributions to Oncology,Vol. 43. Munich, Germany: Karger, 1992.Shalaby, M. R., Shepard, H. M.. Presta, L.. Rodrigues, M., Beverley, P. C. L.,Feldmann. M., and Carter, P. Development of humanized bispecific antibodiesreactive with cytotoxic lymphocytes and tumor cells overexpressing the HER2protooncogene. J. Exp. Med., 175: 217-225, 1992.Rodrigues, M. L., Shalaby, M. R„Werther, W„Presta, L., and Carter, P. Engineeringa humanized bispecific F(ab'), fragment for improved binding to T cells. Int. J.

Cancer, 7(Suppl.): 45-50, 1992.

Park, J. W.. Hong, K.. Carter. P., Asgari, H., Guo, L. Y., Shalaby, R., Wirth, C., Kotts,C., Keller. G. A., Wood, W. I., Papahajopoulos, D., and Benz, C. C. Development ofanti-pl85HER2 immunoliposomes for cancer therapy. Proc. Nati. Acad. Sci. USA, in

press, 1994.Bagshawe, K. D. Antibody directed enzyme prodrug therapy (ADEPT). In: J. G.Fortner and J. E. Rhoads (eds.). Accomplishments in Cancer Research, pp. 154-170.

London: Chapman and Hall Medical, 1991.Senter, P. D., Wallace, P. M., Svensson, H. P., Vrudhula, V. M., Kerr, D. E.,Hellström, I., and Hellström, K. E. Generation of cytotoxic agents by targetedenzymes. Bioconjugate Chem., 4: 3-9, 1993.Eaton, M. A. W., Alexander, R. P., and Pratt, A. J. Immunoconjugates and prodrugsand their use in association for drug delivery. WO 90/11782, 1990.Shepherd, T. A., Jungheim, L. N.. Meyer, D. L.. and Starling, J. J. A novel targeteddelivery system utilizing a ccphalosporin-oncolytic prodrug activated by an antibodyß-lactamaseconjugate for the treatment of cancer. Bioorg. & Med. Chem. Lett., /:21-26, 1991.Jungheim, L. N., Shepard. T. A., and Myer, D. L. Synthesis of acylhydrazido-substituted cephems. Design of cephalosporin-vinca alkaloid prodrugs: substrates foran antibody-targeted enzyme. J. Org. Chem., 57: 2334-2340, 1992.

Meyer, D. L., Jungheim, L. N., Mikolajczyk, S. D., Shepherd, T. A., Starling, J. J.,and Ahlem, C. N. Preparation and characterization of a ß-Iactamase-Fab' conjugate

for the site-specific activation of oncolytic agents. Bioconjugate Chem., 3: 42-48,

1992.Alexander, R. P., Beeley, N. R. A., O'Driscoll, M.. O'Neill, F. P., Millican, T. A.,

20. Junghcim. L. N.. Shepard. T. A., and King, J. K. Synthesis of a cephalosporin-doxorubicin antitumor prodrug: a substrate for an antibody-targeted enzyme. Hetero-cycles (Tokyo), 35: 329-348, 1993.

21. Meyer, D. L., Jungheim, L. N., Law. K. L., Mikolajczyk, S. D., Sheperd, T. A.,Mackensen, D. G., Briggs, S. L., and Starling, J. J. Site-specific prodrug activation by antibody-ß-lactamase conjugates: regression and long-term growth inhibition of human colon carcinoma xenograft models. Cancer Res., 5.?: 3956-3963,

1993.22. Bosslet, K.. Czech. J.. and Hoffmann. D. Tumor-selective prodrug activation by

fusion protein-mediated catalysis. Cancer Res., 54: 2151-2159, 1994.23. Eigenbrot, C., Randal. M.. Presta. L., Carter, P., and Kossiakoff, A. X-ray structures

of the antigen-binding domains from three variants of humanized anti-pi85HI R2

antibody 4D5 and comparison with molecular modelling. J. Mol. Biol., 229:969-995, 1993.

24. Srinivasan, N.. Sowdhamini. R., Ramakrishnan, C., and Balaram, P. Conformationsof disulfide bridges in proteins. Ini. J. Pepi. Protein Res., 36: 147-155, 1990.

25. Carter, P., Kelley, R. F.. Rodrigues, M. L., Snedecor, B., Covarrubias, M., Velligan,M. D., Wong, W. L. T., Rowland, A. M., Kotts, C. E., Carver, M. E.. Yang, M.,Bourell. J. H., Shepard, H. M., and Henner, D. High level Eschericliia coli expressionand production of a bivalent humanized antibody fragment. Bio-Technology, 10:163-167, 1992.

26. Vieira, J., and Messing, J. Production of single-stranded plasmid DNA. MethodsEnzymol., 153: 3-11, 1987.

27. Carter, P. Mutagenesis facilitated by the removal or introduction of unique restrictionsites. In: M. J. McPherson (ed.), Mutagenesis: A Practical Approach, pp. 1-25.Oxford: IRL Press, 1991.

28. Garrard, L. J., Yang, M., O'Connell, M. P., Kelley, R. F., and Henner, D. J. Fab

assembly and enrichment in a monovalent phage display system. Bio-Technology, 9:1373-1377, 1991.

29. Kabat, E. A., Wu, T. T., Perry, H. M., Gottesman, K. S., and Foeller, C. Sequencesof Proteins of Immunological Interest. Ed. 5. Bethesda, MD: NIH, 1991.

30. Chang, A. C. Y., and Cohen, S. N. Construction and characterization of amplifiablcmulticopy DNA cloning vehicles derived from the pl5A cryptic miniplasmid.J. Bacteriol., 134: 1141-1156, 1978.

31. Kelley, R. F.. and O'Connell, M. P. Thermodynamics of an antibody functional

epitope. Biochemistry, 32: 6828-6835, 1993.32. Cartwright, S. J., and Waley, S. G. Purification of ß-lactamases by affinity chroma-

tography on phenylboronic acid-agarose. Biochem. J., 22/: 505-512, 1984.

33. Zafaralla, G., Manavathu. E. K.. Lerner, S. A., and Mobashery. S. Elucidation of therole of arginine-244 in the turnover processes of class A ß-lactamases. Biochemistry,31: 3847-3852, 1992.

34. Bourell, J. H., Clauser, K. P., Kelley, R., Carter, P., and Stulls, J. T. Eleclrosprayionization mass spectrometry of engineered antibody fragments. Anal. Chem., 66:2088-2095, 1994.

35. Dalbadie-McFarland, G., Neitzel, J. J., and Richards, J. H. Active-site mutants ofß-lactamase: use of an inactive double mutant to study requirements for catalysis.Biochemistry, 25: 332-338, 1986.

36. Fendly, B. M., Kotts, C., Vetterlein, D., Lewis, G. D., Wingel, M.. Carver, M. E.,Watson, S. R., Sarup, J., Saks, S.. Ullrich, A., and Shepard, H. M. The extracellulardomain of HER2/nei/ is a potential immunogen for active specific immunotherapy ofbreast cancer. J. Biol. Response Modif., 9: 449-455, 1990.

37. Rodrigues, M. L., Snedecor. B., Chen, C., Wong, W. L. T., Garg, S., Blank, G. S.,Maneval, D., and Carter, P. Engineering Fab' fragments for efficient F(ab')2 forma

tion in Escherichia coli and for improved in rivo stability. J. Immunol., 757:6954-6961, 1993.

38. Seehaus. T.. Breitling. F.. Dübel,S.. Klewinghaus. I., and Little, M. A vector for theremoval of deletion mutants form antibody libraries. Gene (Amsl.), 114: 235-237,

1992.39. Goshorn, S. C., Svensson, H. P., Kerr, D. E., Somerville, J. E., Senlcr, P. D., and Fell,

H. P. Genetic construction, expression, and characterization of a single chain anti-carcinoma antibody fused to ß-lactamase.Cancer Res., 53: 2123-2127, 1993.

40. Glockshuber, R.. Malia. M.. Pfitzinger, I., and Pliickthun, A. A comparison ofstrategies to stabilize immunoglobulin Fv-fragments. Biochemistry, 29: 1362-1367,1990.

41. Brinkmann. U., Reiter, Y., Jung, S-H., Lee, B., and Pastan. I. A recombinantimmunotoxin containing a disulfide-stabilized Fv fragment. Proc. Nati. Acad. Sci.USA, 90: 7538-7542, 1993.

42. Reiter, Y., Brinkmann, U., Webber, K. O., Jung, S-H., and Pastan, I. Engineeringinterchain disulfide bonds into conserved framework regions of Fv fragments: improved biochemical characteristics of recombinant immunotoxins containing disul-fide-linked Fv. Protein Eng., 7: 697-704, 1994.

43. Reiter. Y., Brinkmann, U., Kreitman, R. J, Jung, S-H., Lee, B., and Pastan, I.Stabilization of the Fv fragments in recombinant immunotoxins by disulfide bondsengineered into conserved framework regions. Biochemistry, 33: 5451-5459,

1994.44. Presta, L. G., Lahr. S. J., Shields, R. L., Porter, J. P., Gorman, C. M., Fendly, B. M.,

Pratt, A. J., and Willenbrock, F. W. Cephalosporin nitrogen mustard carbamateprodrugs for "ADEPT." Tetrahedron Lett., 32: 3269-3272, 1991.

Svensson, H. P., Kadow, J. F., Vrudhula, V. M., Wallace, P. M., and Senter, P. D.Monoclonal antibody-ß-lactamase conjugates for the activation of a cephalosporinmustard prodrug. Bioconjugate Chem., 3: 176-181, 1992.Hanessian, S., and Wang, J. Design and synthesis of a cephalosporin-carboplatinumprodrug activatable by a ß-lactamase.Can. J. Chem., 71: 896-906, 1993.

Hudyma, T. W., Bush, K., Colson, K. L., Firestone, R. A., and King, H. D. Synthesisand release of doxorubicin from a cephalosporin based prodrug by a ß-lactamase-immunoconjugate. Bioorg. & Med. Chem. Lett., 3: 323-328, 1993.

and Jardieu, P. M. Humanization of an antibody directed against IgE. J. Immunol.,151: 2623-2632, 1993.

45. Eigenbrot, C., Gonzales, T., Mayeda, J., Carter, P., Werther, W., Hotaling, T., Fox, J.,and Kessler, J. X-ray structures of fragments from binding and nonbinding versionsof a humanized anti-CD18 antibody: structural indications of the key role of V,,residues 59 to 65. Proteins Struct. Funct. Genet., 18: 49-62, 1994.

46. Adams. G. P., McCartney, J. E.. Tai, M.-S.. Oppermann, H.. Huston. J. S.. Stafford,

W. F., Ill, Bookman, M. A.. Fand, I., Houston, L. L., and Weiner, L. M. Highlyspecific in viva tumor targeting by monovalent and divalent forms of 741F8 anti-c-erbB-2 single-chain Fv. Cancer Res., 53: 4026-4034. 1993.

Research. on August 12, 2019. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

TOWARD ADEPT FOR OVEREXPRESSING TUMORS

47. Hancock, M. C, Langton, B. C., Chan, T., Toy, P., Monahan, J. J., Mischak, R. P.,and Shawver, L. K. A monoclonal antibody against the c-erbB-2 protein enhances thecytotoxicity of ctf-diamminedichloroplatinum against human breast and ovariantumor cell lines. Cancer Res., 5^: 4575-4580, 1991.

48. Shepard, H. M., Lewis, G. D., Sarup, J. C., Fendly, B. M., Maneval, D., Mordenti, J.,Figari, I., Kotts, C. E., Palladino, M. A., Jr., Ullrich, A., and Slamon, D. Monoclonalantibody therapy of human cancer: taking the HEK2 protooncogene to the clinic.J. Clin. Immunol., //.•117-127, 1991.

49. Baselga, J., Norton, L., Copian, K., Shalaby, R., and Mendelsohn, J. Antitumoractivity of paclitaxel in combination with anti-growth factor receptor monoclonal

antibodies in breast cancer xenografts. Proc. Am. Assoc. Cancer Res., 85: 380, 1994.50. Muss, H. B., Thor, A. D., Berry, D. A., Kute, T., Liu, E. T., Koerner, F., Cirrincione,

C. T., Budman, D. R., Wood, W. C., Barcos, M. B., and Henderson, I. C. c-erfeB-2expression and response to adjuvant therapy in women with node-positive early breastcancer. N. Engl. J. Med., 330: 1260-1266, 1994.

51. Eccles, S. A., Court, W. J., Box, G. A., Dean, C. J., Melton, R. G., and Springer, C. J.Regression of established breast carcinoma xenografts with antibody-directed enzymeprodrug therapy against c-erbB2 pl85. Cancer Res., 54: 5171-5177, 1994.

52. Press, M. F., Cordon-Cardo, C., and Slamon, D. J. Expression of the HER-2/neu

proto-oncogene in normal human adult and fetal tissues. Oncogene, 5: 953—962,

1990.53. Erlichman, C. Pharmacology of anticancer drugs. In: I. F. Tannock, and R. P. Hill

(eds.). The Basic Science of Oncology, Ed. 2, pp. 330. New York: McGraw-Hill,

1992.54. Bagshawe, K. D., Sharma, S. K., Springer, C. J., Antoniw, P., Rogers, G. T., Burke,

P. J., Melton, R., and Sherwood, R. Antibody-enzyme conjugates can generate

cytotoxic drugs from inactive precursors at tumor sites. Antibody Immunoconj.Radiopharm., 4: 915-922, 1991.

55. Griffiths, A. D., Williams, S. C., Hartley, O., Tomlinson, I. M., Waterhouse, P.,Crosby, W. L., Kontermann, R., Jones, P. T., Low, N. M., Allison, T. J., Prospero,T. D., Hoogenboom, H. R., Nissim, A., Cox, J. P. L., Harrison, J. L., Zaceólo, M.,Gherardi, E., and Winter, G. Isolation of high affinity human antibodies directly fromlarge synthetic repertoires. EMBO J., 13: 3245-3260, 1994.

56. Stewart, J. D., and Benkovic, S. J. Recent developments in catalytic antibodies. Int.Rev. Immunol., 10: 229-240, 1993.

57. Holliger, P., and Winter, G. Engineering bispecific antibodies. Curr. Opin.Biotechnol., 4: 446-449, 1993.

70

Research. on August 12, 2019. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from

1995;55:63-70. Cancer Res   Maria L. Rodrigues, Leonard G. Presta, Claire E. Kotts, et al.   Cephalosporin Doxorubicin Prodrug

-Lactamase Fusion Protein for Activation of aβ Fv-HER2Development of a Humanized Disulfide-stabilized Anti-p185

  Updated version

  http://cancerres.aacrjournals.org/content/55/1/63

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/55/1/63To request permission to re-use all or part of this article, use this link

Research. on August 12, 2019. © 1995 American Association for Cancercancerres.aacrjournals.org Downloaded from