Humanized and chimeric monoclonal antibodies that recognize epidermal growth factor receptor (EGF-R); diagnostic and therapeutic use

[0002] This application is a continuation of application Ser. No. 08/560,568, filed Nov. 17, 1995, now U.S. Pat. No. 5,891,996, issued Apr. 6, 1999, which itself claims priority to Cuban patent application 128/94 filed Nov. 8, 1994, the contents of which are incorporated herein by reference.

[0003] The present invention relates to the field of immunology and, in particular, to two new products; chimeric and humanized monoclonal antibodies against epidermal growth factor receptor (“EGF-R”). The chimeric and humanized monoclonal antibodies of the present invention are less immunogenic than original murine monoclonal antibodies and possess improved effector functions.

[0004] The present invention also relates to therapeutic and diagnostic compositions comprising these antibodies.

[0005] The Epidermal Growth Factor (“EGF”) is a 53 amino acid polypeptide with a molecular weight of 6045 D and was isolated and purified, for the first time, from murine submaxillary gland. (Cohen S; J Biol Chem (1962) 237, 1555). Later, a similar molecule was obtained from human urine. (Cohen S and Carpenter G; (1975) PNAS USA 72, 1317). The activity of EGF is primarily performed via its interaction with its membrane receptor, a 170 kDa molecular weight glycoprotein. The receptor's intracellular domain is associated with a tyrosine kinase activity and has structural homology to the oncogene v-erb-B showing relation to the malignant transformation process (Heldin C H et al; (1984) Cell 37, 9-20).

[0006] High levels of EGF-R have been detected in malignant tumors of epithelium origin such as breast, bladder, ovarian, vulva, colonic, lung, brain and esophagus cancers. The role of EGF and its receptor in regulating tumor growth is unknown, but it has been suggested that EGF-R expression in tumor cells provides a mechanism for autocrine growth stimulation which leads to uncontrolled proliferation (Schlessinger J, Schreiber A B, Levi A, Liberman T and Yarden Y; (1983) Crit Rev Biochem 14(2), 93-111).

[0007] It has been reported that EGF produces overgrowth of breast cancer cell lines (Osborne C K et al; (1980) Cancer Research 40, 236 1), as well as modulating the differentiation under some cellular systems (Tonelli C J; Nature (1980) 285, 250-252). These effects on cellular differentiation and proliferation are related to the high expression of EGF-R (Buss J E et al; (1982) PNAS 79, 2574).

[0008] The presence of EGF-R in tumor cells has proven to be an indicator of a poor prognosis in human breast cancer. Approximately 40% of the breast tumors show specific binding sites having high affinity for EGF suggesting this growth factor receptor could broaden the concept of hormone dependency in breast cancer (Perez R, Pascual M R, Macias A, Lage A; (1984) Breast Cancer Research and Treatment 4, 189-193). There is also an inverse correlation between the expression of EGF-R and the presence of estrogen receptor, indicating EGF-R as an indifferentiation marker or an indicator of the potential for proliferation of the malignant cells.

[0009] Other groups have reported that the expression of EGF-R is higher in regional ganglionar metastasis than in primary carcinomas of breast (Sainsbury J R et al; (1985) Lancet 1, 8425, 364-366) and that the expression of the receptor is different in the different histologic subtypes of human breast carcinoma cells, where presence of the receptor constitutes a signal of a poor prognosis (Macias A et al; Anti Cancer Research 6: 849-852).

[0010] The results obtained in different studies have prompted the consideration of the EGF/EGF-R system as a possible target for therapeutic actions.

[0011] We obtained a murine monoclonal antibody (R3), raised against the human placenta as described in European Patent application No. 93202428.4, and found to bind to the external domain of the human EGF-R. The R3 antibody was found to inhibit the binding of EGF at both low and high affinity EGF-R sites.

[0012] Passive immunotherapy using monoclonal antibodies against the EGF-R have been the object of multiple investigations and have demonstrated that the specific recognition of the receptor by the antibody inhibits the EGF binding and has an inhibitory effect on the mitogenic stimulation of malignant cells (Sato J D et al; (1987) Methods in Enzimology 148, 63-81). However, there is evidence that the murine origin of these antibodies produces a human anti-mouse antibody response.

[0013] The development of the hybridoma antibody technique by Kohler and Milstein revolutionized the discipline of immunochemistry and provided a new family of reagents with potential applications in clinical diagnosis and immunotherapy (Kohler G, Milstein C; (1975) Nature 256, 495-497). While it has become routine to produce mouse monoclonal antibodies (mAbs) for use in basic research and clinical diagnosis, it has been difficult to use these mAbs for in vivo immunotherapy because they have reduced half-life in humans, there is poor recognition of mouse antibody effector domains by the human immune system, and the foreign immunoglobulin can elicit an antiglobulin response (HAMA response) that may interfere with therapy.

[0014] The ability to genetically manipulate antibody genes and then express these altered genes by transfection techniques enables us to produce mAbs having more desirable properties than the existing hybridoma antibodies. Thus, genetic engineering can be used to enhance desired effector functions in antibody molecules and to decrease or eliminate undesired effector functions.

[0015] The use of recombinant DNA technology to clone antibody genes has provided an alternative wherein a murine mAb can be converted to a predominantly human form with the same antigen binding properties. In 1984, S L Morrison created mouse-human antibody molecules, of defined antigen-binding specificity, by taking the variable regions genes of mouse antibody producing myeloma cell lines, and joining them to human immunoglobulin constant regions (Morrison S L et al; (1984) PNAS USA 81, 6851-6855).

[0016] Other authors have attempted to build rodent antigen binding sites directly into human antibodies by transplanting only the antigen binding site, rather than the entire variable domain, from a murine antibody (Jones P T et al; (1986) Nature 321, 522-524; Verhoeven M et al; (1988) Science 239, 1534-1536). Some applications of this method have been developed (Rietchmann L et al; (1988) Nature 332, 323-327; Quee C et al; (1989) PNAS USA 86, 10029-10033), other authors have worked with reshaped antibodies, which included some murine residues in human FRs in order to recover the affinity for the original antigen (Tempest PR; (1991) Biotechnology 9, 266-272).

[0017] Orlandi R et al. (Proc Natl Acad Sci USA 86, 3833-3837, 1989) disclose the constant regions of the human gamma-1 heavy chain and the human kappa light chain, and suitable cloning vectors thereof.

[0018] The invention provides a chimeric and a humanized mAb which are directed to the EGF-R comprising, an antigen-binding site of non-human origin, and the constant regions of human origins (chimeric) and the framework regions (“FRs”) of the variable regions and the constant regions of human origins may be, if necessary, modified in a way that the specificity of the binding to EGR-R can be conserved or restored.

[0019] The present invention can be used to characterize the hypervariable regions of the antigen-binding site of an antibody against the EGF-R and provide these CDRs within a humanized and chimeric mAb defined as above.

[0020] These antibodies can play a role as a therapeutic or diagnostic agent in order to combat tumors with high expression of EGF-R.

[0021] The present invention also provides chimeric and humanized antibodies specific for the EGF-R.

[0022] More specifically, the invention provides a chimeric monoclonal antibody comprising variable regions of non-human origin and constant regions of light and heavy chains of human origin, wherein the chimeric monoclonal antibody binds to human EGF-R and inhibits binding of EGF to EGF-R. According to a preferred embodiment of the chimeric monoclonal antibody, the variable regions of the antigen binding sites comprise the amino acid sequences shown in FIGS. 1aand 1b.

[0023] Further, the invention provides a humanized monoclonal antibody comprising antigen binding sites (CDRs) of non-human origin and the FRs of variable region and constant regions of light and heavy chains of human origin, wherein the humanized monoclonal antibody binds to human EGF-R and inhibits binding of EGF to EGF-R. According to a preferred embodiment of the humanized monoclonal antibody, the hypervariable regions of the antigen binding sites comprise the amino acid sequences underlined in FIG. 1 and FIG. 1b. Preferably, the FRs of the variable region which is not related to the antigen binding sites comprise the following amino acid sequences:

[0024] light chain:

[0025] FR1: asp-ie-gln-met-thr-gln-ser-pro-ser-ser-leu-ser-ala-ser-val-gly-asp-arg-val-thr-ile-thr-cys (SEQ ID NO: 1),

[0026] FR2: trp-tyr-gln-gln-thr-pro-gly-lys-ala-pro-lys-leu-leu-ile-tyr (SEQ ID NO: 2),

[0027] FR3: gly-val-pro-ser-arg-phe-ser-gly-ser-gly-ser-gly-thr-asp-phe-thr-phe-thr-ile-ser-ser-leu-gln-pro-glu-asp-ile-ala-thr-tyr-tyr-cys (SEQ ID NO: 3),

[0028] FR4: phe-gly-gln-gly-thr-lys-leu-gln-ile-thr-arg-glu (SEQ ID NO: 4),

[0029] FR1: gln-val-gln-leu-gln-gln-ser-gly-ala-glu-val-lys-lys-pro-gly-ser-ser-val-lys-val-ser-cys-lys-ala-ser-gly-tyr-thr-phe-thr (SEQ ID NO: 5),

[0030] FR2: trp-val-arg-gln-ala-pro-gly-gln-gly-leu-glu-trp-ile-gly (SEQ ID NO: 6),

[0031] FR3: (arg,lys)-(val,ala)-thr-ile-thr-val-asp-glu-ser-(thr,ser)-(thr,asn)-thr-ala-tyr-met-glu-leu-ser-ser-leu-arg-ser-glu-asp-thr-ala-phe-tyr-phe-cys-(ala,thr)-arg (SEQ ID NO: 7),

[0032] FR4: trp-gly-gln-gly-ser-thr-val-thr-val-ser-ser (SEQ ID NO: 8), and wherein the amino acids listed in brackets are alternatives.

[0033] The humanized monoclonal antibody may comprise a derivative of an amino acid sequence modified by amino acid substitution within the variable and constant regions wherein the biological function of specific binding to the antigen is preserved.

[0034] In a preferred embodiment of the humanized and chimeric monoclonal antibodies of the present invention, the constant region of the heavy chain comprises the amino acid sequence of a gamma-1 chain and the constant regions of the light chain comprise the amino acid sequences of a kappa chain of a human immunoglobulin. A purified humanized or chimeric monoclonal antibody which derives from murine mAb R3 is preferred.

[0035] The invention also includes a pharmaceutical composition comprising a chimeric or humanized monoclonal antibody as defined herein. Usually, the composition will also contain a pharmaceutically acceptable carrier.

[0036] The invention also includes in the use of a humanized or chimeric antibody, as defined herein, for the manufacture of a drug directed to tumors, and use of a humanized or chimeric antibody, as defined herein, for diagnostic localization and assessing tumor growth.

[0037] The file of this patent contains at least one drawing executed in color. Copies of this patent with color drawings will be provided by the Patent and Trademark Office upon request and payment of the necessary fee.

[0038] FIG. 1a: Deduced amino acid sequence of VK of murine R3 antibody. CDRs are undelined.

[0039] FIG. 1b: Deduced amino acid sequence of VH of murine R3 antibody. CDRs are underlined. FIG. 2a: Amino acid sequence of murine and humanized VK containing mAb R3 CDRs.

[0040] FIG. 2b: Amino acid sequence of murine and humanized VH containing mAb R3 CDRs.

[0041] FIG. 3: Detection of binding of the chimeric and humanized R3 to EGF-R by RRA. Antigen binding activity was assayed in different concentrations of purified murine R3(+), chimeric R3 (*) and humanized (1) R3 (-.-) and plotted as CPM of bound ¹²⁵1-EGF against log of the concentration of each antibody. (Concentration of IgG was quantitated by ELISA, see detailed description of the invention).

[0042] FIG. 4: Detection of binding of hybrids murine R3 VH/humanized R3 VK and humanized R3 VH/murine R3 VK. The antigen binding activity was assayed in dilutions concentrated supernatants from transfected NSO cells and plotted as cpm (membrane bound radioactivity) against log of the concentration of IgG estimated by ELISA, see experimental protocols.

[0043] FIG. 5: Comparison of the amino acid sequences of reshaped humanized R3 heavy chain variable region. The CDRs are underlined.

[0044] FIG. 6: Molecular Model of the Variable region of murine mAb R3. The binding site is at the top and the molecule has been slightly rotated clockwise around a vertical axis so that VH domain approaches the viewer. The VL framework is light blue and the VH framework is pink. The corresponding CDRs are marked in dark blue and red. Side chains of residue Ser 75, Thr 76 and Thr 93 (with polar hydrogen atoms included) are shown in green. The side chain of the Phe 100f is shown in red.

[0045] FIG. 7: Detection of binding of the different reshaped mAb R3 to EGF-R by RRA. Antigen binding activity was assayed in dilutions concentrated supernatants from transfected NSO cells and plotted as cpm (membrane bound radioactivity) against log of the concentration of IgG estimated by ELISA, see experimental protocols. All versions of reshaped human VH regions were cotransfected with huVKR3 and are represented in the figure.

[0046] FIG. 8: Immunogenicity in monkeys of R3 murine mAb and different recombinant versions. Cercopithecus aethiopsmonkeys were immunized intradermically at 15 days intervals with 2 mg of the different mAbs versions using aluminum hydroxide as adjuvant. The detection of monkey IgG against the different versions of R3 mAb was tested in the sera of the animals using an indirect ELISA.

[0047] cDNA Synthesis and Gene Amplification of Variable Region of R3

[0048] Cytoplasmic RNA was extracted from about 10⁶hybridoma cells of R3 (IgG 2A), this antibody was obtained by us (Fernandez A et al; (1989) IFN y Biotecnologia 6(3), 289-298). The method used to extract RNA was described by Faloro et al: (Faloro J, Treisman R and Kemen R; (1989) Methods in Enzymology 65: 718-749).

[0049] The cDNA synthesis reaction mixture was done as described by Tempest et al. (Tempest P R, Bremner P, Lambert M, Taylor G, Furze J M, Carr F D J and Harris W J; (1991) Biotechnology 9: 266-271). Briefly, a 50 ul reaction mixture containing 5 ug of mRNA, 25 pmol of Vh or VK primer FOR, 250 uM of each dNTP, 10 mM DTT, 50 mM Tris-HCI (pH 8.3), 8 mM MgCl₂, 75 mM KCl and 15 units of RNAse inhibitor, was heated at 70° C., for 10 min and slowly cooled to 37° C. over a period of 30 min. Then, 100 units MLV reverse transcriptase (BRL) were added and the incubation at 37° C. continued for 1 hour.

[0050] The VH and VK cDNAs were amplified using the PCR as described by Orlandi et al (Orlandi R, Gussow D H, Jones P T and Winter G; (1989) Proc Natl Acad Sci USA 86: 3833-3837). For PCR amplification of VH, DNA/primers mixture consisted of 5 ul cDNA, 25 pmoles CG2AFOR and VHIBACK primers. For PCR amplification of VK, DNA/primers mixture consisted of 5 ul cDNA, 25 pmoles CK2FOR and VK10OBACK primers. To these mixtures were added 2.5 mM of each dNTP, 5 ul constituents of 10×buffer thermolase and 1 unit of Thermolase (IBI) in a final volume of 50 ul. Samples were subjected to 25 thermal cycles at 94° C., 30 sec; 50° C., 30 sec; 72° C., 1 min; and a last incubation for 5 min at 72° C. Amplified VH and VK DNA were purified on Prep A Gene purification kit (BioRad).

[0051] Cloning and Sequencing of Amplified cDNA

[0052] The purified VH and VK cDNA were cloned into M13-mp19 vector. Clones were sequenced by the dideoxy method using T7 DNA Pol (Pharmacia). We reamplified the cDNA by PCR using VH1BACK and VH1FOR primers for VH and VK3BACK and VK3FOR primers for VK. The amplified cDNAs were digested with PstI and BstEII for the VH gene or PvuII and BglII for the VK gene. The fragments were cloned into M13-VHPCR1 (digested with PstI and BstEII) or into M13-VKPCR1 (digested with PvuII and BcII). The M13VHPCR-R3 and M13VKPCR-R3 containing V gene inserts were identified directly by sequencing.

[0053] Construction of Chimeric Genes

[0054] The VH gene together with the Ig heavy chain promoter, appropriated splicing sites and signal peptide sequences were excised from M13 vectors by digestion with HindIII and BamHI and cloned into an expression vector (pSVgpt). A human IgG1 constant region (Takahashi N, Veda S, Obatu M, Nikaido T, Nakai S and Honjo T; (1982) Cell 29: 718-749), was then added as a BamHI fragment. The resultant construction was R3VH-pSVgpt. The construction of the R3VK-pSVhyg was essentially the same except that the gpt gene was replaced by the hygromycin resistance gene and a human Kappa chain constant region was added (Hieter P A, Max E E, Seidman J G, Maizel J V Jr and Leder P; (1980) Cell 22: 197-207).

[0055] Chimeric Antibody Expression

[0056] NSO cells were electroporated with 4 ug of R3VH-pSVgpt gamma 1 region and 8 ug R3VK-pSVhyg kappa constant region were linearized by digestion with PvuI. The DNAs were mixed together, ethanol precipitated and dissolved in 25 ul water. Approximately 10⁷NSO cells were grown to semiconfluency, harvested by centrifugation and resuspended in 0.5 ml DMEN together with the digested DNA in an electroporation cuvette. After 5 minutes on ice, the cells were given a pulse of 170 volts and 960 uF (Gene-Pulser, Bio-Rad) and left in ice for a further 30 minutes. The cells were then put into 20 ml DMEN plus 10% fetal calf serum and allowed to recover for 48 hours. At this time the cells were distributed into a 96-well plate and selective medium applied (DMEN, 10% fetal calf serum, 0.8 ug/ml mycophenolic acid, 250 ug/ml xanthine). Transfected clones were visible with the naked eye 14 days later.

[0057] The presence of human antibody in the medium of wells containing transfected clones was measured by ELISA. Microtiter plate wells were coated with goat anti-human IgG (gamma chain specific) antibodies (Sera Lab). After washing with PBST (phosphate buffered saline containing 0.02% Tween 20, pH 7.5), 20 ul of culture medium from the wells containing transfectants was added to each microtiter well for 1 hour at 37° C. The wells were then emptied, washed with PBST and peroxidase-conjugated goat anti-human Kappa, light chain specific (Ser-Lab), were added and incubated at 37° C. for one hour. The wells were then emptied, washed with PBST and substrated buffer containing o-phenylediamine added. Reactions were stopped after a few minutes by the addition of sulphuric acid and absorbance at 492 nm was measured.

[0058] Transplantation of CDRs into Human Frameworks

[0059] The construction of the first version of humanized of R3-huVH(a) and R3-huVK was carried out using a CDR grafting approach similar to that described by Kunkel et al. (Kunkel T A; (1985) Proc Natl Aca Sci USA 82, 488; Kunkel T A; (1987) Methods in Enzymology 155, 166). Briefly: To 0.5 ug of VH or VK single stranded Uracil DNA in M13 VH or VK PCR vectors (coding for the human Eu VH region sequence and human REI VK region sequence) were added ten pmoles of VH or VK phosphorylated oligonucleotides encoding the mouse CDRs sequences. Primers were annealed on the template by heating to 70° C. and slowly cooled to 37° C. After site-directed mutagenesis, the DNA was transformed into competent E. coliTG1 cells. Single strand DNA was prepared from individual plaques and sequenced. If only single or double mutants were obtained, then these were subjected to further rounds of mutagenesis using the appropriate oligonucleotides until the triple CDR mutants were obtained.

[0060] Further versions of reshaped human R3VH were constructed using PCR mutagenesis (Kammann M, Laufs J, Schell J and Gronenbom B; (1989) PNAS USA 86, 4220-4224), using the oligonucleotides described in example 5.

[0061] Cloning and Expression of Humanized R3 Antibody into NSO Cells

[0062] After CDR-grafting, the HindIII-BamHI fragment carrying the R3 humanized VH and R3 humanized VK genes were recloned in expression vectors, yielding the plasmids R3HuVH(1-7)-pSVgpt gamma1 and HuR3VK-pSVhyg kappa constant region. Vectors were linearized with PvuI, and humanized expression was done like chimaeric expression into NSO cells.

[0063] Molecular Modeling of mAb R3 VK and VH

[0064] A model of the variable regions of mouse mAb R3 was built using the molecular modeling program QUANTA/CHARm 4.0 (Molecular Simulations Inc., 1994), running on a 150 MHz Silicon Graphics Indigo Extreme workstation. The VK and VH frameworks were built separately from Fab 26-10 (Jeffrey P D, Strong R K, Sieker L C, Chang C Y, Campbell R L, Petsko G A, Haber E, Margolies M N and Sheriff S; (1993) PNAS USA 90, 10310) and Fab 36-71 (Strong R K, Campbell R L, Rose D R, Petsko G A, Sharon J and Margolies M N; (1993) Biochemistry 30, 3739), respectively, Fab 26-10 and mAb R3 have 92% homology in the VK frameworks and 88% homology in the whole VK region. The VH frameworks of Fab 36-71 and R3 mAb have 85% homology.

[0065] Coordinates were taken from the Brookhaven Protein Data Bank (entries 11GI and 6FAB). The frameworks of Fab 36-71 were fitted to the frameworks of Fab 26-10, matching only those residues that have been found to be often involved in the interface between the light and heavy variable regions (Chotia C, Novotny J, Bruccolery R and Karplus M; (1985) J Mol Biol 186, 651). The VH domain of Fab 26-10 and the VK domain of Fab 36-71 were then deleted leaving the needed hybrid. Side-chain replacements were performed following the maximum overlap procedure (Snow M E and Amzel L M; (1986) Proteins 1, 267), and comparing, where possible, with other crystal structures.

[0066] The hypervariable regions of the R3-Variable Light (VL) domain (L1, L2 and L3) were built retaining the same main-chain conformations as in Fab 26-10, since the corresponding CDRs in both antibodies are highly homologous and belong to the same canonical structural groups (Chotia C, Lesk A M, Tramontano A, Levitt M, Smith-Gill S J, Air G, Sherii S, Padlan E A, Davies D, Tulip W R, Colman P M, Spinelli S, Alzari P M and Poljak R J; (1989) Nature 342, 877). In the VH domain of mAb R3, CDR H1 belongs to canonical structural group 1, as in Fab 36-71, so the main-chain torsion angles of the parent molecule were kept. CDR H2 corresponds to canonical structural group 2 and the main-chain conformation for this loop was taken from the Fv fragment 4D5 (entry 1FVC), which was selected among other highly resolved structures because of the good matching of its H2 loop base with the framework of Fab 36-71. For all the above mentioned loops, comparisons with other CDRs from the Data Bank were made to orient the side chains.

[0067] To model CDR H3, which in mAb R3 was 14 amino acids long, a high temperature molecular dynamics was used for conformational sampling (Bruccoleri R E and Karplus M; (1990) Biopolymers 29, 1847). First, the whole structure without CDR H3 was subjected to an energy minimization keeping residues H-94 and H-103 fixed and using harmonic constraints of 10 Kcal/(mole atom A²) for main chain atoms. Then a loop was constructed with an arbitrary conformation starting from the two previously fixed amino acids. Those residues close to the framework were placed taking into consideration other crystal structures and the top part of the loop was built with an extended conformation avoiding strong steric interactions with the rest of the molecule. For the next modeling steps only CDR H3 and the neighboring side chains within a distance of 5A⁰were permitted to move. An energy minimization was first carried out and then a molecular dynamics at 800K was run for 150 picoseconds. The time step for the run was set to 0.001 picosecond and coordinates were saved every 100 steps. The 120 lowest energy conformations from the dynamics run were extracted and subjected to an energy minimization in which all atoms in the structure were allowed to move. Several low-energy conformations were obtained and the one with the lowest energy was used in the subsequent analyses. Differences between murine and humanized variants of R3 antibody were individually modeled to investigate their possible influence on CDR conformation.

[0068] EGF Receptor Radioligand Competition Assays

[0069] The determination of the apparent inhibition constants (Ki) of the ¹²⁵I-EGF binding to its receptor by the anti EGF-R mAb was performed by homogeneous Radio Receptor Analysis (RRA) with human placenta microsomal fraction. (Macias A, Perez R, Lage A; (1985) Interferon y Biotecnologia 2: 115-127). The affinity constant of the antigen antibody reaction was also estimated by a competitive RRA but using mAb ¹²⁵I-R3 as the radiolabeled probe.

[0070] Materials

[0071] Recombinant Human Epidermal Growth Factor (HEGF) was obtained from the Center of Genetic Engineering and Biotechnology, Havana, Cuba. ¹²⁵I-hEGF was radioiodinated by the chloramine-T method (specific activity 150-200 uCi/ug). Murine hybridoma cell line R3 was obtained by us (European Patent Application No 93202428.4). Rat myeloma NSO is a non-Ig secreting cell line and was grown in Dulbecco's modified Eagles medium (DMEN) containing 10% fetal calf serum. Vectors M13VHPCR1, M13VKPCR1, pSVgpt and pSVhyg have been described in detail (Orlandi R, Gussow D H, Jones P T and Winter G; (1989) Proc Natl Acad Sci USA 86: 3833-3837) and were obtained from Greg Winter, MRC Laboratory of Molecular Biology, Cambridge, UK. Oligonucleotides were synthesized using an Applied Biosystems 381 DNA synthesizer.

[0072] Pharmaceutical Compositions

[0073] With respect to the formulation of suitable compositions for administration to patients in need of treatment, the monoclonal antibodies according to the invention may be admixed or combined with pharmaceutically acceptable carriers known per se, dependent upon the chosen route of administration. There are no particular limitations to the modes of application of the invention, and the choice of suitable administration routes and suitable compositions belong to the routine skills of persons skilled in the art.

[0074] Although other forms of administration are possible as well, a preferred administration form would be a solution for injection, in particular for intravenous or intraarterial injection. Usually, a suitable pharmaceutical composition for injection may comprise a buffer (e.g. acetate, phosphate or citrate buffer), a surfactant (e.g. polysorbate), optionally a stabilizer agent (e.g. human albumine), etc. Persons skilled in the art know, or can easily determine, other possibilities.

[0075] Similarly, persons skilled in the art have the ability to determine the best concentrations and proportions of the various components, administration dose and frequency, etc. For example, a suitable solution for injection will usually contain from about 1 to about 20, preferably 5-10 mg antibody per ml. The dose to be administered could, for example, be from about 0.1 to about 20, preferably 1-5 mg/kg body weight, and administration could be once per day, or three times per week, or whatever the physician thinks best.

[0076] VH and VK were amplified using PCR. The specific oligonucleotides used as primers were:

[0077] For the heavy chain variable region:

[0078] CG2AFOR (5′ GGAAGCTTAGACCGATGGGGCCTGTTGTTTTG 3′) (SEQ ID NO: 9);

[0079] VH1BACK (5′ AGGT(G/C)(A/C)A(A/G)CTGCAG(G/C)AGTC(A/T)GG 3′) (SEQ ID NO: 10).

[0080] For the light chain variable region:

[0081] CK2FOR (5′ GAAGCTTGAAGATGGATACAGTTGGTGCAGC 3′) (SEQ ID NO: 11);

[0082] VK10BACK (5′ TTGAATTCCAGTGATGTTTTGATGACCCA 3′) (SEQ ID NO: 12).

[0083] The purified VH and VK cDNA were cloned into M13 vector. Twelve independent clones were sequenced by the dideoxy method using T7 DNA Pol (Pharmacia). In FIGS. 1aand 1b, you can see the variable region sequences of the murine R3 mAb. The VH sequence is most closely related to kabat subgroup VH IIB and the VK to kabat subgroup II.

[0084] We reamplified the cDNA by PCR using the following oligonucleotides as primers:

[0085] For VH:

[0086] VH1BACK (5′ AGGT(G/C)(A/C)A(A/G)CTGCAG(G/C)AGTC(A/T)GG 3′) (SEQ ID NO: 10);

[0087] VHb 1FOR (5′ TGAGGAGACGGTGACCGTGGTCCCTTGGCCCCAG 3′) (SEQ ID NO: 13).

[0088] For Vk:

[0089] VK3BACK (5′ GACATTCAGCTGACCCA 3′) (SEQ ID NO: 14);

[0090] VK3FOR (5′ GTTAGATCTCCAGTTTGGTGCT 3′) (SEQ ID NO: 15).

[0091] The amplified cDNAs were digested with PstI and BstEII for the VH gene or PvuII and BglII for the VK gene. The fragments were cloned into M13-VHPCR1 (digested with PstI and BstEII) or into M13-VKPCR1 (digested with PvuII and BclI). Details of vectors (Orlandi R et al; Proc Natl Acad Sci USA 86: 3833-3837, 1989). The M13VHPCR-R3 and M13VKPCR-R3 containing V gene inserts were identified directly by sequencing.

[0092] The VH gene together with the Ig heavy chain promoter, appropriated splicing sites and signal peptide sequences were excised from M13 vectors by digestion with HindIII and BamHI and cloned into an expression vector (pSVgpt). A human IgG1 constant region (Takahashi N et al; Cell 29: 718-749, 1982) was then added as a BamHI fragment. The resultant construction was R3VH-pSVgpt. The construction of the R3VK-pSVhyg was essentially the same, except that the gpt gene was replaced by the hygromycin resistance gene and a human Kappa chain constant region was added (Hieter PA et al; Cell 22: 197-207, 1980).

[0093] We compared murine light and heavy chains variable regions of R3 with variable regions sequences of human immunoglobulins.

[0094] The murine R3 heavy chain variable region has 63.4% of homology with human subgroup VHI and the R3 light chain variable region has 63.7% of homology with human subgroup VKI. The human FRs chosen were derived of the human immunoglobulin REI for the light chain and the human immunoglobulin EU for the heavy chain.

[0095] The construction of the first version of the humanized R3 heavy chain and humanized R3 light chain was carried out using CDR grafting approach as described herein.

[0096] Oligonucleotides were designed which consisted of DNA sequences coding for each end by 12 bases of DNA complementary to the DNA sequences coding for the adjacent FRs of human Eu for Vh and human REI for VK. The oligonucleotides designed were:

[0097] For the CDR1 of the light chain:

[0098] 5′ TCTAGATCAGTCTTGTAACATGTATCATTACCTTTGTGGATAAATCTG 3′ (SEQ ID NO: 16).

[0099] We did not use oligonucleotides for the CDR2 because REI human CDR2 is identical to R3 CDR2.

[0100] For the CDR3 of the light chain:

[0101] 5′ GATGACGAAAGTTATCAAGTGTACAAGGGACCTG 3′ (SEQ ID NO: 17).

[0102] For the CDR1 of the heavy chain:

[0103] 5′ ATGTGGAAGTGGTTAATAATATAGATAACCCACTCTGTC 3′ (SEQ ID NO: 18).

[0104] For the CDR2 of the heavy chain:

[0105] 5′ ACTACCTACCCTCCCTATTTGGGGTGGAGACCTCCCTCATTGAAATTACTTTTCA GTTCTGTTCTCACTGTTAA 3′ (SEQ ID NO: 19).

[0106] For the CDR 3 of the heavy chain:

[0107] 5′ AAAACACGTTCTGTCCCGAACACCAAGCTGTCACTGCCTGCCCCGAAACTGAAG 3′ (SEQ ID NO: 20).

[0108] NSO cells were electroporated with 4 ug of murine or humanized R3 VH-CMMAR gamma 1 region and 8 ug of murine or humanized R3VK-CMMARhyg kappa constant region were linearized by digestion with PvuI. The DNAs were mixed together, ethanol precipitated and dissolved in 25 ul water. Approximately 10⁷NSO cells were grown to semiconfluency, harvested by centrifugation and resuspended in 0.5 ml DMEN together with the digested DNA in an electroporation cuvette. After 5 minutes on ice, the cells were given a pulse of 170 volts and 960 uF (Gene-Pulser, Bio-Rad) and left in ice for a further 30 minutes. The cells were put into 20 ml DMEN plus 10% fetal calf serum and allowed to recover for 48 hours. At this time the cells were distributed into 96-well plate and selective medium applied (DMEN, 10% fetal calf serum, 0.8 ug/ml mycophenolic acid, 250 ug/ml xanthine). Transfected clones were visible with the naked eye 14 days later.

[0109] The presence of human antibody in the medium of wells containing transfected clones was measured by ELISA. Microtiter plate wells were coated with goat anti-human IgG (gamma chain specific) antibodies (Sera Lab). After washing with PBST (phosphate buffered saline containing 0.02% Tween 20, pH 7.5), 20 ul of culture medium from the wells containing transfectants was added to each microtiter well for 1 hour at 37° C. The wells were then emptied, washed with PBST and peroxidase-conjugated goat anti-human Kappa, light chain specific (Sera-Lab), were added and incubated at 37° C. for one hour. The wells were then emptied, washed with PBST and substrated buffer containing o-phenylenediamine added. Reactions were stopped after a few minutes by the addition of sulphuric acid and absorbance at 492 nm was measured.

[0110] In FIG. 3 the chimeric and murine R3 antibody bind to the EGF-R with the same affinity (tested in RRA), and the first version of humanized R3 antibody did not bind to the EGF-R.

[0111] The introduction of some murine residues in the human FRs was necessary because the first version of humanized R3 antibodies that we constructed did not bind to the antigen.

[0112] We constructed hybrids between murine R3VH/humanized R3VK and humanized R3VH/murine R3VK. FIG. 4 shows that the hybrid murine R3VH/humanized R3VK binds to the EGF-R with the same affinity as the original murine R3.

[0113] Humanized R3huVk retained all the Vernier's residues of murine R3VK. This fact may explain why the humanizing procedure did not affect the binding capability of the kappa chain.

[0114] Nevertheless humanized R3VH/murine R3VK did not work. This result suggested to us that we had to include murine residues into human FRs of the heavy chain, for the recovery of the binding to the antigen of this antibody.

[0115] Then, other versions of the reshaped humanized heavy chains were constructed by PCR mutagenesis. In these new versions we tried to keep the Vernier's zone intact, recommending changes in the position 66, 67, 75, 76 and 93 (Kabat numbering) in the FRs regions.

[0116] The oligonucleotides designed to get the reshaped R3VH-K66A67 were:

[0117] Top strand:

[0118] 5′ GAAAAGTTCAAGACAAAAGCGACAATTACGGTAGAC 3′ (SEQ ID NO: 21).

[0119] Lower strand:

[0120] 5′ GTCTACCGTAATTGTCGCTTTTGTCTTGAACTTGAACTTTTC 3′ (SEQ ID NO: 22).

[0121] The oligonucleotides designed to get the reshaped R3VH-S75T76 were:

[0122] Top strand:

[0123] 5′ GTAGACGAGAGCAGCACCACGGCGTACATG 3′ (SEQ ID NO: 23).

[0124] Lower strand:

[0125] 5° CATGTACGCCGTGGTGCTGCTCTCGTCTAC 3′ (SEQ ID NO: 24).

[0126] The oligonucleotides designed to get the reshaped R3VH-T93 were:

[0127] Top strand:

[0128] 5′ TTCTATTTTTGTACAAGACAGGGCTTG 3′ (SEQ ID NO: 25).

[0129] Lower strand:

[0130] 5° CAAGCCCTGTCTTGTACAAAAATAGAA 3′ (SEQ ID NO: 26).

[0131] A model of the variable regions of mouse mAb R3 was built using the molecular modeling program QUANTA/CHARm 4.0 (Molecular Simulations Inc, 1994), running on a 150 MHz Silicon Graphics Indigo Extreme workstation.

[0132] Differences between murine and humanized variants of R3 antibody were individually modeled to investigate their possible influence on CDR conformation.

[0133] Heavy chain variable region residues at positions 66, 67, 75, 76 and 93, (Kabat numbering) which had been changed in humanized R3huVH (FIG. 6), are close to the hypervariable loops and therefore may influence the CDR conformations. Both our experimental results and the modeling studies suggest that only Thr76 and Thr93 (Kabat numbering) are critical for binding affinity.

[0134] A molecular model of the murine R3 heavy chain variable regions was constructed to analyze the possible effects of these mutations. Residue 93 was located just below CDR H3, close to Phe100f(Kabat numbering) (FIG. 6), and replacing Thr with a smaller amino acid like Ala may provoke some rearrangements of the neighboring side chains and modify the overall conformation of the H3 loop.

[0135] THR 76 is close to CDR H2 (FIG. 2) and the introduction of the larger Asn residue at this position could lead to hydrogen-bonding interactions with the backbone of CDR H1. Moreover, residue 76 is accessible from the top of the variable region and could be directly involved in the binding to EGF-R. The substitution Ser75-Thr (Kabat numbering) alone did not seem to have any influence, but taken together with mutation at position 76, could be important. The changes Lys66-Arg and Ala67-Val (Kabat numbering) did not appear to affect the structure, but since they had been found to have some influence in the functional binding of reshaped mAb 425 (Kettleborough C, Saldanha J, Heath V J, Morrison C J and Bending M M; (1991) Protein Engineering 4, 773-783), we decided to make substitutions at these positions as well.

[0136] Humanized R3huVK retained all the Vernier's residues of murine R3VK. This fact may explain why the humanizing procedure did not affect the binding capability of the kappa chain.

[0137] Reshaped R3 hu VH antibodies containing either S75/T76 or T93 (Kabat numbering) recovered partially the binding capacity, whereas those constructs containing both of them retained full binding activity (FIG. 7).

[0138] The determination of the affinity constant of the ¹²⁵-EGF binding to its receptor by murine R3 chimeric and humanized antibodies was performed by an homogeneous Radio Receptor Analysis (RRA) with human placenta microsomal fraction (Macias A et al; Interferon y Biotecnologia 2: 115-127, 1985).

[0139] These different versions of chimeric and humanized antibodies were assayed, using this technique, for its ability to bind to EGF-R (FIG. 7). The different versions of reshaped human VH regions result in a wide range of levels of antigen binding (FIG. 7). Versions 6 and 7 have the same affinity as original murine antibody, these versions with highest levels of binding were followed of versions 3, 4 and 5 and then followed by version 2.

[0140] Based on these results, it is possible to comment on the relative contributions individual residues in the FRs make to antigen binding. The 75 and 76 changes are, together with 93 (all Kabat numbering) very important to the binding, while the introduction of changes at positions 66 and 67 (Kabat numbering) fail to produce significant antigen binding.

[0141] Three treatment groups with two Cercopithecus aethiopsmonkeys in each group were immunized as follows: 1. Murine R3 monoclonal antibody (2 mg) with 5 mg of aluminum hydroxide as adjuvant; 2. Chimeric R3 antibody (2 mg) with 5 mg of aluminum hydroxide as adjuvant; and 3. Humanized (version 6) R3 antibody (2 mg) with 5 mg of aluminum hydroxide as adjuvant. All the groups were immunized intradermically on weeks 1, 3, 5, and 7. Starting on week one, blood was collected from all the groups weekly.

[0142] The serum obtained and the titer of antibodies against EGF-R was determined by an ELISA technique.

[0143] Costar plates (Inc, high binding) were coated with murine R3 monoclonal antibody at a concentration of 10 ug/ml in bicarbonate buffer (pH 9.6) and incubated overnight. After, the plates were washed with PBST and blocked with the same buffer containing 1% BSA for one hour at room temperature.

[0144] The washing step was repeated and 50 ul/well of the different serum dilutions were added. After incubating for 2 hours at 37° C., the plates were washed again and incubated for 1 hour at 37° C. with alkaline phosphated conjugated goat anti-human total or anti-human IgG Fc region specific antiserum (Sigma, Inc). After washing with PBST, the wells were incubated with 50 ul of substrate buffer (1 mg/ml of p-nitrophenylphosphate diluted in diethanolamine buffer (pH 9.8) and absorbance at 405 nm was read with an ELISA reader (Organon Teknika, Inc).

[0145] A high IgG response to murine R3 antibody was obtained when this antibody was used as immunogen. A lower but still measurable IgG response (1/10000) to the murine R3 antibody was obtained when monkeys were immunized with the chimeric antibody, contrary to the results obtained with the humanized (version 6) (FIG. 8). With the humanized antibody no response was measurable after 3 immunizations.

[0146] The cell line hr-R3 expressing the antibody was deposited pursuant to the Budapest Treaty with European Collection of Animal Cell Cultures, Centre for Applied Microbiology and Research Microbiological, Porton Down, Salisbury Wiltshire SP4, OJG, United Kingdom on Nov. 10, 1995, and obtained deposit number 9511 10101.