`
`
`
`
`
`
`PHIGENIX
`PHIGENIX
`Exhibit 1013
`Exhibit 101 3
`
`

`

`Proc. Natl. Acad. Sci. USA
`Vol. 89. pp. 5867—5871, July 1992
`Medical Sciences
`
`Recombinant anti-erbB2 immunotoxins containing
`Pseudomonas exotoxin
`
`(chemotherapy/monoclonal antibodies/growth factor receptors)
`
`JANENDRA K. BATRA*T, PHILIP G. KASPRZYKi, ROBERT E. B1RD1, IRA PASTAN*, AND C. RICHTER K1NG¢§
`
`‘Iaboratory of Molecular Biology, Division of Cancer Biology, Diagnosis and Centers, National Cancer Institute, National Institutes of Health, 9000 Rockville
`Pike, 37/4E16, Bethesda, MD 20892; and *Molecular Oncology, Inc., 19 Firstfield Road, Gaithersburg, MD 20878
`
`Contributed by Ira Pastan, March 31 , 1992
`
`Immunotoxins were made using five different
`ABSTRACT
`murine monoclonal antibodies to the human erbBZ gene prod-
`uct and LysPE40, a 40-kDa recombinant form of Pseudomonas
`exotoxin (PE) lacking its cell-binding domain. All five conju-
`gates were specifically cytotoxic to cancer cell lines overex-
`pressing erbBZ protein. The most active conjugate was e23-
`LysPE40, generated by chemical crosslinking of anti-erbBZ
`monoclonal antibody e23 to LysPE40. In addition, a recombi-
`nant immunotoxin, e23(Fv)PE40, was constructed that consists
`of the light-chain variable domain of e23 connected through a
`peptide linker to its heavy-chain variable domain, which in turn
`is fused to PE40. The recombinant protein was made in
`Escherichia coli, purified to near homogeneity, and shown to
`selectively kill cells expressing the erbBZ protooncogene. To
`improve the cytotoxic activity of e23(Fv)PE40, PE40 was
`replaced with a variant, PE38KDEL, in which the carboxyl end
`of PE is changed from Arg-Glu-Asp-Leu-Lys to Lys-Asp-Glu-
`Leu and amino acids 365—380 of PE are deleted. The
`e23(Fv)PE38KDEL protein inhibits the growth of tumors
`formed by the human gastric cancer cell line N87 in immuno-
`deficient mice.
`
`Clinical trials are under way in which monoclonal antibodies
`are used to carry cytotoxic substances to tumor cells. Im-
`munotoxins made by coupling monoclonal antibodies (mAbs)
`to toxins have been found to kill cancer cells in vitro and also
`to have antitumor activities in mice bearing human tumor
`xenografts (1—4). Amplification and overexpression of the
`erbBZ gene has been shown to occur in many human cancers,
`including ~30% of lung, breast, ovary, and stomach adeno-
`carcinomas (5—11). In breast carcinoma, a correlation has
`been observed between gene amplification and overexpres-
`sion of erbB2 protein and the aggressiveness of the malig-
`nancy (7, 8).
`In cases of gene amplification,
`there is a
`resulting 50- to 100-fold increase in erbBZ mRNA compared
`with normal cell levels (11). The overexpression of erbB2 has
`been directly linked to the malignant conversion of cancer
`cells (12, 13). This causative role for the erbB2 protein makes
`it an excellent target for immunotoxin therapy because can-
`cer cells are unlikely to be able to escape treatment by loss
`of the antigen.
`Pseudomonas exotoxin A (PE) and its recombinant forms
`have been used to make immunotoxins either by conven-
`tional chemical coupling methods or by recombinant DNA
`methods (3, 4, 14—17). PE is made of three structural do-
`mains. The N-terminal domain (I) is responsible for the
`binding of toxin to its receptor on the cells, the middle domain
`(domain II) has a role in the translocation of toxin across the
`membrane, and the C-terminal domain (III) has the ADP-
`ribosylation activity (18). Recently, molecularly defined im-
`munotoxins have been engineered by fusing domains II and
`
`
`The publication costs of this article were defrayed in part by page charge
`payment. This article must therefore be hereby marked “advertisement"
`in accordance with 18 U.S.C. §1734 solely to indicate this fact.
`
`III of PE to the light- and heavy-chain variable regions of
`specific mAbs. In the resulting single-chain immunotoxin the
`cell-binding domain of PE is replaced with an antigen com-
`bining site in the form of a single-chain antibody, or SC(Fv).
`Here, we report the construction of several immunotoxins
`using various anti-erbBZ antibodies that have been coupled to
`native LysPE40, which is a recombinant form of PE devoid
`of its cell-binding domain. The resulting immunotoxins were
`specifically cytotoxic to cells expressing erbB2 with varying
`degrees of activity. We selected one mAb, e23, to make a
`single-chain chimeric immunotoxin termed e23(Fv)PE40. In
`e23(Fv)PE40, the variable domain of the light chain of mAb
`e23 is attached through a peptide linker to the variable
`domain of the heavy chain, which in turn is fused to domains
`11 and III of PE. The immunotoxin was found to be specif-
`ically cytotoxic to cells expressing erbB2 and was more
`active than the chemical conjugate. To increase the activity
`further, other derivatives of this single-chain immunotoxin
`were made in which the toxin part of the molecule was
`altered. One of these molecules, e23(Fv)PE38KDEL, inhib-
`ited the human gastric cancer cell line N87 growing as a tumor
`in immunodeficient mice.
`
`MATERIALS AND METHODS
`
`Antibodies and Cell Lines. e1, e23, e21, e68, and e94 are
`mouse mAbs against erbBZ gene product (19). The erbB2-
`expressing cell lines used were BT474 (breast carcinoma),
`N87 (gastric carcinoma), and SK-OV-3 (ovarian carcinoma).
`A431 and KB are human epidermoid carcinomas that express
`low levels of erbBZ.
`Construction of Chemical Conjugates. LysPE40 or PE was
`coupled chemically to mAbs by a thioether linkage using
`2-iminothiolane and succinimidyl-4-(N—maleimidomethyl)cy-
`clohexane-l-carboxylate to derivatize the antibody and the
`toxin, respectively (3). Conjugates were purified as described
`(3).
`Generation of a SC(Fv) from mAb e23. Poly(A)+ RNA was
`extracted from hybridoma cells by oligo(dT) affinity chro-
`matography (Invitrogen, San Diego). cDNA was prepared
`using random hexanucleotide primer (Boehringer Mann-
`heim). The immunoglobulin light- and heavy-chain clones
`were isolated using PCR with the following primers: light
`chain, 5’-CAC-GTC-GAC-ATT-CAG-CTG-ACC-CAC-
`TCT-CCA-3’ and 5’-GAT-GGA-TCC-AGT-TGG-TGC-
`AGC-ATC-3’; heavy chain, 5’-C-GGA-ATT-TCA-GGT-
`TCT-GCA-GIA-GTC-WGG-S’ and 5'-AGC-GGA-TCC-
`AGG-GGC-CAG-TGG-ATA-GAC-3' (I, deoxyinosine; W, A
`
`
`Abbreviations: mAb, monoclonal antibody; PE, Pseudomonas ex-
`otoxin; SC(Fv), single-chain antibody comprising heavy- and light-
`chain variable regions.
`TPresent address: National Institute of Immunology, National Insti-
`tute of Immunology Campus, J.N.U. Complex, Shahid Jeet Singh
`Marg, New Delhi, 110067, India.
`_
`§To whom reprint requests should be addressed.
`
`5867
`
`PHIGENIX
`
`Exhibit 1013-01
`
`

`

`5868
`
`Medical Sciences: Batra et al.
`
`Proc. Natl. Acad. Sci. USA 89 (1992)
`
`or T). The products of the PCR reaction were cloned into
`plasmid pUCl8. Linkage into a SC(Fv) was by PCR involving
`the individual light- and heavy-chain cDNA clones and four
`oligonucleotides. The light- and heavy-chain regions were
`joined by a synthetic linker with amino acid sequence
`GSTSGSGKSSEGKG specified by overlapping oligonucle-
`otides as described. The intact SC(Fv) coding region was
`inserted in frame with an ompA leader sequence under
`direction of the A PL promoter. Induction of protein, and lysis
`of bacteria, and refolding of protein were as previously
`described (20). The SC(Fv) was purified as a single peak from
`CM chromatography and judged to be >70% pure by SDS/
`PAGE.
`Plasmids. Plasmid pJB23-40 encodes e23(Fv)PE40 and con-
`tains cDNA for light—chain variable region of anti-erbB2 an-
`tibody e23 connected through a 42-base-pair linker to its
`heavy-chain variable region, which in turn is fused to the DNA
`for PE40 (see Fig. 3). The fusion gene is under the control of
`a bacteriophage T7 late promoter and the plasmid also contains
`phage f1 origin and a terminator. Plasmid pJB23-38 encodes
`the protein termed e23(Fv)PE38. In pJB23-38, codons for
`amino acids 365-380 in domain 11 of PE40 have been deleted.
`The Fv portion is the same as in pJBZ3-40. In pJB23-38K, the
`codons for amino acids REDLK at the 3’ end of pJBZ3-38 have
`been replaced with those coding for amino acids KDEL (21).
`The protein encoded by this plasmid is termed e23(Fv)-
`PE38KDEL.
`In pJB23-40K,
`the codons for amino acids
`REDLK at the 3’ end of pJB23-40 have been replaced with
`those coding for amino acids KDEL (22). The protein encoded
`by this plasmid is termed e23(Fv)PE40KDEL.
`Expression and Purification of Proteins. Expression and
`purification of various fusion proteins in Escherichia coli
`were performed as described (21).
`Characterization of Immunotoxins. Cytotoxic activities of
`the chimeric toxins were determined by assaying the inhibi-
`tion of protein synthesis in various target and nontarget cells
`(3). Cells were plated in 24-well plates and 24 hr later washed
`once with medium before addition of the toxins. Results are
`described as percent of control where no toxin was added.
`For competition experiments, an excess of antibody (20
`pg/ml) was added prior to the addition of the toxin.
`In binding studies 125I-labeled e23 Fab was added as a
`tracer with various concentrations of the competitor (16).
`Protein was assayed by the Bradford method using Pierce
`Coomassie blue “plus” reagent. SDS/PAGE was done by
`the method of Laemmli (23).
`For in vivo studies N87 tumor cells (5 X 106 per mouse)
`were subcutaneously injected into the flanks of BNX (beige,
`nude, xid) mice on day 0. Starting on day 10, six intravenous
`tail vein treatments (2 p.g each) were carried out over 3 days.
`Tumor growth and animal weights were monitored twice a
`week. Tumor grth is reported as an average relative tumor
`volume calculated as (w2 x 0/2 (mm’), where w is the width
`and I is the length of the tumor, measured with calipers.
`
`RESULTS
`
`Previous studies have shown that individual mAbs to a single
`antigen can result in immunotoxins of widely variable activity
`(1). To determine which available anti-erbBZ mAb resulted in
`an immunotoxin of highest activity, we first used chemical
`crosslinking to produce anti-erbB2 immunotoxins in which
`the mAbs were coupled to LysPE40 (3, 4, 21). The antibodies
`used are designated e1, e21, e23, e68, and e94 (19). The
`activity of the immunotoxins was assessed by measuring their
`ability to inhibit protein synthesis in target and nontarget
`cells. BT474, N87, and SK-OV-3 are cell lines that overex-
`press erbB2, whereas A431 and KB do not. All the conjugates
`were active on BT474 cells, with e23-LysPE40 being the most
`active. The same general pattern of activity was observed on
`
`Table 1. Activity of anti-erbBZ—PE40 conjugates on various
`human cell lines
`
`Toxin
`
`BT474
`
`N87
`
`IDso, ns/ml
`SK-OV-3
`
`A431
`
`KB
`
`2000
`700
`180
`37
`18
`e23-LysPE40
`>2000
`>2000
`200
`38
`47
`e21-LysPE40
`ND
`ND
`500
`64
`36
`e1-LysPE40
`2000
`2000
`>1000
`130
`180
`e68-LysPE40
`>2000
`>2000
`600
`100
`42
`e94-LysPE40
`>2000
`650
`>2000
`ND
`160
`LysPE40
`IDso values are the concentrations of protein required to inhibit
`protein synthesis by 50%. ND, not done.
`
`BT474 and N87 cells, but SK-OV-3 cells were less sensitive
`to these conjugates (Table 1). All of these immunotoxins had
`little or no toxicity for A431 and KB cells, indicating their
`specificity for erbBZ. Specificity was also demonstrated by
`showing that excess unconjugated antibodies prevented the
`inhibition of protein synthesis by the respective immunotox—
`ins (data not shown). Since e23 produced the most active
`conjugate, it was used for further studies.
`To compare the binding activity of e23-LysPE40 with that
`of the native antibody, competition binding analyses were
`performed on mouse NIH 3T3 cells engineered to express
`high levels of human erbBZ protein (12). In these studies we
`determined the abilities of the two immunotoxins to compete
`for the binding of 125I-labeled e23 to transfected NIH 3T3 cells
`overexpressing human erbB2 protein. e23-LysPE40 and
`e23-PE were found to compete for binding to the erbB2
`antigen very efficiently, with binding affinity slightly lower
`than that of the native antibody (data not shown).
`ATGGACCTGCAGCTGACCCAGTCTCCAGCAATCCTGTCTGCATCTCCAGG
`MetAspLeuGlnLeuThrGlnSerProAlaIleLeuSerAlaSerProGly
`GGAGAAGGTCACAATGACTTGCAGGGCCACCCCAAGTGTAAGTTACATGC
`GluLysValThrMetThrCysArgAlaThrProSerValSerTeretHis
`ACTGGTATCAGCAGAAGCCAGGATCCTCCCCCAAACCTTGGATTTATACC
`TrpTyrGlnGlnLysProGlySerSerProLysProTrpIleTerhr
`ACATCCAACCTGGCTTCTGGAGTCCCTGCTCGCTTCAGTGGCGGTGGGTC
`ThrSerAsnLeuAlaSerGlyValProAlaArgPheSerGlyGlyGlySer
`TGGGACCTCTTACTCTCTCACAGTCAGCAGAGTGGAGGCTGAAGATGCTG
`GlyThrSerTyrSerLeuThrValSerArgValGluAlaGluAspAlaAla
`CCACTTATTACTGCCAGCAGTGGAGTCGTAGCCCACCCACGTTCGGAGGG
`ThrTererysGlnGlnTrpSerArgSerProProThrPheGlyGly
`GGGTCCAAGCTGGAAATAAAAGGTTCTACCTCTGGTTCTGGTAAATCTTC
`GlySerLysLeuGluIleLysGlySorihx80201y80201yLyl30:80:
`TGAAGGTAAAGGTGTGCAGCTGCAGGAGTCAGGACCTGAGGTGGTGAAGC
`GluGlyLyIGlyValGlnLeuG1nGluSerGlyProGluVa1ValLysPro
`CTGGAGGTTCAATGAAGATATCCTGCAAGACTTCTGGTTACTCATTCACT
`GlyG1ySerMetLysIleSerCysLysThrSerGlyTyrSerPheThr
`GGCCACACCATGAACTGGGTGAAGCAGAGCCATGGAAAGAACCTTGAGTG
`GlyHisThrMetAsnTrpValLysGlnSerHisGlyLysAsnLeuGluTrp
`GATTGGACTTATTAATCCTTACAATGGTGATACTAACTACAACCAGAAGT
`IleGlyLeuIleAsnProTyrAsnGlyAspThrAsnTyrAsnGlnLysPhe
`TCAAGGGCAAGGCCACATTTACTGTAGACAAGTCGTCCAGCACAGCCTAC
`LysGlyLysAlaThrPheThrValAspLysSerSerSerThrAlaTyr
`ATGGAGCTCCTCAGTCTGACATCTGAGGACTCTGCAGTCTATTACTGTGC
`MetGluLeuLeuSerLeuThrSerGluAspSerAlaValTererysAla
`AAGGAGGGTTACGGACTGGTACTTCGATGTCTGGGGCGCAGGGACCACGG
`ArgArgValThrAspTrpTerheAspValTrpGlyAlaGlyThrThrVal
`TCACCGTCTCC
`ThrValSer
`
`Fro. 1. Nucleotide and amino acid sequence of the SC(Fv) for
`e23. The linker joining the light-chain and heavy-chain variable
`regions is shown in bold.
`
`PHIGENIX
`
`Exhibit 1013-02
`
`

`

`Medical Sciences: Batra et al.
`
`Proc. Natl. Acad. Sci. USA 89 (1992)
`
`5869
`
`p.1823-4o
`
`---II-IE-l]- REDLK
`
`pJB23-38
`
`---II--1I-
`
`REDLK
`
`9132mm
`
`----l--II- KDEL
`
`p.1823-40K
`
`----l-l]3-l1- KDEL
`
`FIG. 3. Various e23(Fv)PE40 derivatives. VH, variable region of
`heavy chain; VL, variable region of light chain; L, linker; 11, domain
`11 of PE; lb, domain lb of PE; 111, domain 11] of PE. Carboxyl-
`terminal amino acid sequences are shown in single-letter code.
`
`conjugate, e23-LysPE40 (Table 2). On all four target cells,
`e23(Fv)PE40 was active and the IDso values on a molar basis
`were 2- to 3-fold lower than those of e23-LysPE40. Both
`molecules had very little activity on KB cells, showing their
`specificity for erbB2-expressing cells.
`Derivatives of e23(Fv)PE40. The carboxyl terminus of PE
`ends in the amino acids REDLK. Replacing REDLK with
`KDEL results in molecules that have 3- to 10-fold higher
`cytotoxic activities (21). Also, deleting part of domain 1b of
`PE—i.e., amino acids 365—380, thereby deleting a disulfide
`bond—does not result in any loss of activity of several fusion
`proteins including TGFa-PE40, anti-Tac(Fv)-PE40, and
`B3(Fv)PE40 (17, 24). Since mixed disulfide bonds can form
`during the renaturation of recombinant proteins, we thought
`it would be helpful to delete this region. To explore the
`possibility of making a much more active derivative of
`e23(Fv)PE40, we made three new constructions:
`(i)
`e23(Fv)PE40KDEL, where the carboxyl terminus REDLK is
`replaced by KDEL in e23(Fv)PE40; (ii) e23(Fv)PE38,
`in
`which amino acids 365—380 have been deleted from
`e23(Fv)PE40 but the carboxyl terminus is still REDLK; and
`(iii) e23(Fv)PE38KDEL, where REDLK is replaced by
`KDEL in e23(Fv)PE38. These derivatives are diagramed in
`Fig. 3. The chimeric proteins were also purified to >70%
`purity and tested for cytotoxic activity on target cells. As
`shown in Fig. 4, all of the new derivatives inhibited the
`protein synthesis of BT474 cells in a dose-dependent manner,
`with e23(Fv)PE38KDEL being the most active. Table 3
`summarizes the IDso values of the e23(Fv)P40 derivatives on
`various cell
`lines. On all three target cell lines, e23(Fv)-
`PE38KDEL, was found to be the most active. e23(Fv)-
`PE38KDEL was 6- to 10-fold more active than e23(Fv)PE40
`(Table 3). None of the proteins had any cytotoxicity on KB
`cells, a cell line that does not overexpress erbB2. In the
`presence of excess e23, the cytotoxic activity of all deriva-
`tives was abolished (data not shown). The binding activity of
`e23(Fv)PE38KDEL was monitored in a competition bind-
`
`120
`
`100
`
`3
`.‘5
`C
`8
`°\°
`.9“
`g
`J:..
`E(n
`.E
`g 20
`a.
`0
`
`so
`
`60
`
`40
`
`+ 923(Fv)40
`—o— 923(Fv)38K
`+ 923(Fv)40K
`—D— 023(Fv)38
`
`
`
`.01
`
`.1
`
`1
`
`10
`
`—
`100
`
`1000
`
`ng/ml
`
`FIG. 4. Comparative cytotoxic activities of various e23(Fv)PE40
`derivatives on BT474 cells. 0, e23(Fv)PE40; o, e23(Fv)PE38KDEL;
`I, e23(Fv)PE40KDEL; El, e23(Fv)PE38.
`
`PHIGENIX
`
`Exhibit 1013-03
`
`
`+ 923(Fv)PE40
`
`—o— 923(FV)PE40+Ab
`+923LysPE40
`
`
`
`.1
`
`1
`
`10
`
`100
`
`1000
`
`ng/ml
`
`100
`
`80
`
`60
`
`73
`E-
`o\°
`.1;-
`(I)
`o
`.1:
`E 40
`I)
`.S
`9 20
`9CL
`
`>(
`
`0
`
`FIG. 2. Cytotoxicity of e23(Fv)PE40 on BT474 cells as shown by
`inhibition of protein synthesis. Results are shown as percentage of
`control cells to which no toxin was added. 0, e23(Fv)PE40 alone; 0,
`e23(Fv)PE40 plus native e23 (20 ug/ml); A, e23-LysPE40.
`
`Construction of Single-Chain Immunotoxin with Anti-erbBZ
`Antibody e23. Single-chain immunotoxins made by the fusion
`of the antigen-binding region (Fv) and PE40 can retain the
`binding affinity of the native antibody and are often more
`active than the respective chemical conjugates (15, 16). For
`this reason, we selected antibody e23 for construction of a
`first-generation recombinant immunotoxin. First, an intact
`SC(Fv) for e23(Fv) coding region was generated essentially
`as previously described. The sequence of the single-chain
`protein is shown in (Fig. 1). To verify the binding activity of
`the purified e23(Fv) protein we conducted competition bind-
`ing using 12Sl-labeled e23 Fab (see Fig. 5). The overall
`structure of our first recombinant immunotoxin is the amino-
`terminal e23 SC(Fv) domain joined to the translocation (II)
`and ADP-ribosylating (III) domains of PE. The assembled
`gene is under control of a bacteriophage T7 promoter. The
`resulting plasmid, pJB23-40, expresses the variable region of
`the light chain of e23, a 14-amino acid linker peptide, the
`variable region of the heavy chain of e23, and amino acids
`253—613 of PE. The chimeric protein was expressed in E. coli
`and purified. The resulting protein was >70% pure as judged
`by SDS/PAGE (data not shown).
`Cytotoxicity of e23(Fv)PE40. e23(Fv)PE40 was tested on
`BT474 breast cancer cells and was found to inhibit protein
`synthesis in a dose-dependent manner with an ID50 of 1.5
`ng/ml (Fig. 2, Table 2). The cytotoxic activity was blocked by
`competition with excess native e23, demonstrating the spec-
`ificity of e23(Fv)PE40 for erbBZ-containing cells (Fig. 2).
`Another anti-erbBZ monoclonal antibody, e21, which binds to
`a different site, had no efl‘ect on the toxicity of e23(Fv)PE40
`(data not shown). In the same experiment, e23-LysPE40, the
`chemical conjugate, had an mm of 12 ng/ml (Fig. 2, Table 2).
`The activity of e23(Fv)PE40 was assayed on several cell lines
`expressing erbBZ and compared with that of the chemical
`
`Table 2. Comparison of activity of e23(Fv)PE40 and chemical
`conjugates on various human cell lines
`leo ng/ml (pM)
`
`Cells
`BT474
`N87
`SK-OV-3
`SK-Br—3
`A431
`
`e23(Fv)PE40
`1.5
`(23)
`3.5
`(54)
`22.0 (338)
`32.0 (492)
`170.0 (2615)
`
`e23-LysPE40
`12
`(63)
`24
`(126)
`180
`(947)
`180
`(947)
`>500 (>2631)
`
`Relative
`activity“
`0.37
`0.43
`0.36
`0.52
`NC
`
`*Ratio of IDso (pM) of e23(Fv)PE40 to IDso of e23-LysPE40 on the
`same cell line. NC, not calculated.
`
`

`

`5870
`
`Medical Sciences: Batra et al.
`
`Proc. Natl. Acad. Sci. USA 89 (1992)
`
`Table 3. Activity of e23(Fv)PE40 and derivatives on various
`human cell lines
`
`Protein
`
`BT474
`
`11350 (us/ml)
`N87
`SK-OV-3
`
`e23(Fv)PE40
`e23(Fv)PE40KDEL
`e23(Fv)PE38
`e23(Fv)PE38KDEL
`
`3
`1.6
`3.6
`0.18
`
`8
`3.8
`3.7
`1.2
`
`80
`22
`62
`5
`
`KB
`
`>500
`>500
`>500
`>500
`
`ing assay. As shown in Fig. 5, e23(Fv)PE38KDEL was able
`to compete with homologous e23 Fab for binding, but a
`higher concentration was required than for e23 Fv. This
`result is consistent with either a lower overall affinity of
`e23(Fv)PE38KDEL or the purified protein being a mixture of
`active and inactive species. Current purification methods for
`e23(Fv)PE38KDEL do not allow us to separate forms on the
`basis of binding activity. To verify the binding activity of the
`e23 Fv, we conducted a similar competition binding assay
`and found that e23 Fv binds with slightly lower affinity than
`intact antibody and monomeric Fab produced from e23 (Fig.
`5).
`Growth Inhibition of Human Tumors in 8 Nude Mouse
`Model. The selective toxicity of the e23(Fv)PE38KDEL to
`cells overexpressing erbBZ encouraged us to attempt to treat
`human tumor cells growing in nude mice. The human gastric
`cancer cell line N87 has been shown to overexpress erbB2
`protein at high levels as a result of gene amplification, and
`N87 cells grow well as a subcutaneous tumor in immuno-
`compromised mice (19). Injections of 5 x 106 cells on day 0
`were followed by six intravenous treatments over 3 days,
`starting on day 10. Immunotoxin treatment inhibited growth
`of established tumors (Fig. 6). No animal deaths were ob-
`served at doses of 2 pg. Equivalent amounts of either e23
`Fab, e23 SC(Fv) (data not shown), or LysPE38KDEL had no
`effect on tumor growth. Nonspecific toxicity was assayed by
`monitoring the animal weight; no weight loss was observed at
`doses of 2 pg.
`
`DISCUSSION
`
`We examined the activity of immunoconjugates of anti-erbB2
`mAbs. All
`immunotoxins were selectively toxic to cells
`overexpressing erbB2. Most active were conjugates of mAb
`e23, which was thus selected for constructing a recombinant
`
`
`
`0.1
`
`1
`
`10
`
`100
`
`1000
`
`Competitor, nM
`
`1 500
`
`.4 OOO
`
`l'l'lm3
`
`
`Meantumorvolume,
`
`500
`
`0
`
`1 0
`
`20
`
`30
`
`Days after therapy started
`
`FIG. 6. Therapy of tumors formed in mice by human gastric
`cancer cell line N87. Tumors were established by injection of 5 X 106
`N87 cells subcutaneously on the backs of BNX mice. Therapy was
`initiated 7 days following injection of cells. Treatments were twice
`daily injections in the tail vein with 2 p.g of e23(Fv)PE38KDEL (O),
`LysPES8KDEL (0), or e23 Fab (A) or with phosphate-buffered saline
`(Cl). Measurements were conducted externally with calipers.
`
`immunotoxin. This recombinant immunotoxin contains a
`SC(Fv) linked to PE40. As observed previously (17), the
`recombinant immunotoxin was 2- to 3-fold more active than
`the analogous chemical conjugate, and, we achieved an
`additional 6- to 10-fold increase in cytotoxic activity by
`changing the carboxyl terminus of PE from REDLK to
`KDEL and by deleting 15 amino acids from domain 11 of PE.
`This latter modification is likely to aid in formation of
`properly folded molecules. Our results indicate that
`e23(Fv)PE38KDEL is a potent cytotoxic molecule capable of
`binding specifically to the erbBZ protein. Improved refolding
`of e23(Fv)PE38KDEL or increasing its binding affinity could
`significantly improve the effectiveness of this molecule.
`The potential clinical application of an immunotoxin such
`as e23(Fv)PE38KDEL is based on the overexpression of the
`185-kDa erbB2 glycoprotein (gplsS erbBZ) in about 30% of
`adenocarcinomas of the breast, stomach, lung, and ovary.
`Since gpl$5 erbBZ is expressed on normal cells, therapeutic
`efficacy will probably require administration of doses in a
`range sufficient to kill cells that overexpress erbBZ, but with
`limited toxicity to normal cells. Direct evidence for the
`potential of e23(Fv)PE38KDEL is provided by our results
`showing inhibition of tumor growth in nude mice. The N87
`cells used in these experiments overexpress gplSS erbBZ at
`high levels. At sublethal doses of e23(Fv)PE38KDEL, tumor
`growth was significantly reduced. It should be noted that a
`potentially important form of toxicity was not apparent in this
`experiment, as we do not expect the e23(Fv)PE38KDEL to
`bind to murine gp185 erbBZ (neu) gene product. Such toxicity
`will need to be addressed in animal toxicity experiments
`where the e23(Fv)PE38KDEL binds to endogenous erbBZ
`proteins. Our results do suggest that e23(Fv)PE38KDEL may
`have application in treatment of certain highly malignant
`tumors, such as adenocarcinoma of the stomach,
`lung,
`breast, and ovary.
`
`We thank M. Gallo, E. Lovelace, and T. Clayton for their valuable
`assistance.
`
`FIG. 5. Binding of SC(Fv) and immunotoxin to erbB2. The ability
`of purified e23 Fv (o) and e23(Fv)PE38KDEL (o) to inhibit the
`binding of Ila-labeled e23 Fab was measured using cells overex-
`pressing erbBZ (N87) as the binding target. Also shown are e23 Fab
`(El) and intact antibody e23 (A).
`
`1. Pastan, I. & FitzGerald, D. J. P. (1991) Science 254, 1173—
`1177.
`2. Vitetta, E. S., Fulton, R. J., May, R. D.. Till, M. & Uhr, J. W.
`(1987) Science 238, 1098—1104.
`3. Batra, J. K., Jinno, Y., Chaudhary, V. K., Kondo, T., Will-
`
`PHIGENIX
`
`Exhibit 1013-04
`
`

`

`Medical Sciences: Batra et al.
`
`Proc. Natl. Acad. Sci. USA 89 (1992)
`
`5871
`
`ingham, M. C., FitzGerald, D. J. & Pastan, I. (1989) Proc.
`Natl. Acad. Sci. USA 86, 8545—8549.
`Pai, L. H., Batra, J. K., FitzGerald, D. J., Willingham, M. C.
`& Pastan, I. (1991) Proc. Natl. Acad. Sci. USA 88, 3358—3362.
`King, C. R., Kraus, M. H. & Aaronson, S. A. (1985) Science
`229, 974—976.
`Yokota, J., Yamamoto, T., Toyoshima, K., Sugimura, T.,
`Yamamoto, T., Terada, M., Battifora, H. & Cline, M. J. (1986)
`Lancet ii, 765—767.
`Slamon, D. J., Clark, G. M., Wong, S. G., Levin, W. J.,
`Ulln'ch, A. & McGuire, W. L. (1987) Science 237, 177—182.
`Slamon, D. J., Godolphi, W., Jones, L. A., Holt, J. A., Wong,
`S. 0., Keith, D. E., Levin, W. J., Stuart, S. G., Udove, J.,
`Ullrich, A. & Press, M. F. (1989) Science 244, 707-712.
`Gusterson, B. A., Machin, L. 0., Gullick, W. J., Gibbs,
`N. M., Powllo, T. J., Elliott, C., Ashley, 8., Monaghan, P. &
`Harrison, S. (1988) Br. J. Cancer 58, 453—457.
`Zhou, D., Battifora, H., Yokota, J ., Yamamoto, T. & Cline,
`M. J. (1987) Cancer Res. 47, 6123-6125.
`Kraus, M. H., Popescu, N. C., Amsbaugh, C. & King, C. R.
`(1987) EMBO J. 6, 605—610.
`DiFiore, P. P., Pierce, J. H., Kraus, M. H., Segatto, 0., King,
`C. R. & Aaronson, S. A. (1986) Science 237, 178—182.
`Hudziak, R. M., Schlessinger, J. & Ullrich, A. (1987) Proc.
`Natl. Acad. Sci. USA 84, 7159—7163.
`
`10.
`
`11.
`
`12.
`
`13.
`
`14.
`
`15.
`
`16.
`
`17.
`
`18.
`
`19.
`
`20.
`
`21.
`
`22.
`
`23.
`24.
`
`Chaudhary, V. K., Queen, C., Junghans, R. P., Waldmann,
`T. A., FitzGerald, D. J. & Pastan, I. (1989) Nature (London)
`339, 394-397.
`Chaudhary, V. K., Batra,J. K.,Gallo, M., Willingham, M. C.,
`FitzGerald, D. J. & Pastan, I. (1990) Proc. Natl. Acad. Sci.
`USA 87, 1066—1070.
`Batra, J. K., FitzGerald, D. J., Chaudhary, V. K. & Pastan, I.
`(1991) Mol. Cell. Biol. 11, 2200—2205.
`Brinkmann, U., Pai, L. H., FitzGerald, D. J., Willingham,
`M. C. & Pastan, I. (1991) Proc. Natl. Acad. Sci. USA 88,
`8616—8620.
`Hwang, J ., FitzGerald, D. J. P., Adhya, S. & Pastan, I. (1987)
`Cell 48, 129—136.
`Kasprzyk, P., Song, S. V., DiFiore, P. P. & King, C. R. (1992)
`Cancer Res. 52, 2771—2776.
`Pantaliano, M. W., Bird, R. E., Johnson, 8., Ansel, E. D.,
`Dodd, S. W., Wood, J. F. & Hardman, K. D. (1991) Biochem-
`istry 30, 117-125.
`Seetharam, S., Chaudhary, V. K., FitzGerald, D. J. & Pastan,
`I. (1991) J. Biol. Chem. 266, 17376—17381.
`Kondo, T., FitzGerald, D., Chaudhary, V. K., Adhya, S. &
`Pastan, l. (1988) J. Biol. Chem. 263, 9470—9475.
`Laemmli, U. K. (1970) Nature (London) 227, 680—685.
`Siegall, C. B., Chaudhary, V. K., FitzGerald, D. J. & Pastan,
`I. (1989) J. Biol. Chem. 264, 14256—14261.
`
`PHIGENIX
`
`Exhibit 1013-05
`
`