ANTI-cMET ANTIBODY

Abstract

The invention relates to a novel antibody capable of binding specifically to the human c-Met receptor and/or capable of specifically inhibiting the tyrosine kinase activity of said receptor, with an improved antagonistic activity, said antibody comprising a modified hinge region.

The invention also relates to a composition comprising such an antibody antagonist to c-Met and its use as a medicament for treating cancer.

Claims

1. A method for inhibiting the growth and/or the proliferation of tumor cells, the method comprising the administration to a subject in need thereof of a monoclonal antibody comprising: a heavy chain comprising complementarity determining regions (CDR) CDR-H1, CDR-H2, and CDR-H3 comprising amino acid sequences SEQ ID Nos. 1, 2, and 3, respectively; a light chain comprising CDR-L1, CDR-L2, and CDR-L3 comprising amino acid sequences SEQ ID Nos. 5, 6, and 7; and a modified hinge region comprising the amino acid sequence SEQ ID No. 26.

2. The method of claim 1, wherein the antibody is a chimeric antibody.

3. The method of claim 1, wherein the antibody is a human antibody.

4. The method of claim 1, wherein the antibody is a humanized antibody.

5. The method of claim 1, wherein the antibody comprises: a heavy chain variable domain comprising the amino acid sequence SEQ ID No. 4; and a light chain variable domain comprising amino acid sequence SEQ ID No. 10.

6. The method of claim 5, wherein the antibody comprises a human light chain constant region and a human heavy chain constant region.

7. The method of claim 6, wherein the human light chain constant region is of the IgG1 kappa isotype.

8. The method of claim 1, wherein the antibody comprises a human light chain constant region and a human heavy chain constant region.

9. The method of claim 8, wherein the human light chain constant region is of the IgG1 kappa isotype.

10. The method of claim 6, wherein the antibody is chemically coupled to a mitotic inhibitor.

11. The method of claim 8, wherein the antibody is chemically coupled to a mitotic inhibitor.

12. A method for treating cancer, the method comprising administering to a subject in need thereof a monoclonal antibody comprising: a heavy chain comprising complementarity determining regions (CDR) CDR-H1, CDR-H2, and CDR-H3 comprising amino acid sequences SEQ ID Nos. 1, 2, and 3, respectively; a light chain comprising CDR-L1, CDR-L2, and CDR-L3, comprising amino acid sequences SEQ ID Nos. 5, 6, and 7; and a modified hinge region comprising the amino acid sequence SEQ ID No. 26.

13. The method of claim 12, wherein the antibody is a chimeric antibody.

14. The method of claim 12, wherein the antibody is a human antibody.

15. The method of claim 12, wherein the antibody is a humanized antibody.

16. The method of claim 12, wherein the antibody comprises: a heavy chain variable domain comprising the amino acid sequence SEQ ID No. 4; and a light chain variable domain comprising amino acid sequence SEQ ID No. 10.

17. The method of claim 16, wherein the antibody comprises a human light chain constant region and a human heavy chain constant region.

18. The method of claim 17, wherein the human light chain constant region is of the IgG1 kappa isotype.

19. The method of claim 12, wherein the antibody comprises a human light chain constant region and a human heavy chain constant region.

20. The method of claim 19, wherein the human light chain constant region is of the IgG1 kappa isotype.

21. The method of claim 17, wherein the antibody is chemically coupled to a mitotic inhibitor.

22. The method of claim 19, wherein the antibody is chemically coupled to a mitotic inhibitor.

23. The method according to claim 12, wherein the cancer is chosen from prostate cancer, osteosarcoma, lung cancer, breast cancer, endometrial cancer, glyoblastoma, and colon cancer.

24. The method according to claim 12, wherein the cancer is a c-Met-activation-related cancer chosen from c-Met-activation-related cancers that are HGF-dependent, HGF-independent, or both.

25. A method for inhibiting the growth and/or the proliferation of tumor cells, the method comprising administering to a subject in need thereof a composition comprising a pharmaceutically acceptable vehicle and a monoclonal antibody comprising: a heavy chain comprising complementarity determining regions (CDR) CDR-H1, CDR-H2, and CDR-H3 comprising amino acid sequences SEQ ID Nos. 1, 2, and 3, respectively; a light chain comprising CDR-L1, CDR-L2, and CDR-L3, comprising amino acid sequences SEQ ID Nos. 5, 6, and 7; and a modified hinge region comprising the amino acid sequence SEQ ID No. 26.

26. The method of claim 25, wherein the antibody is a chimeric antibody.

27. The method of claim 25, wherein the antibody is a human antibody.

28. The method of claim 25, wherein the antibody is a humanized antibody.

29. The method of claim 25, wherein the antibody comprises: a heavy chain variable domain comprising the amino acid sequence SEQ ID No. 4; and a light chain variable domain comprising amino acid sequence SEQ ID No. 10.

30. The method of claim 29, wherein the antibody comprises a human light chain constant region and a human heavy chain constant region.

31. The method of claim 30, wherein the human light chain constant region is of the IgG1 kappa isotype.

32. The method of claim 25, wherein the antibody comprises a human light chain constant region and a human heavy chain constant region.

33. The method of claim 32, wherein the human light chain constant region is of the IgG1 kappa isotype.

34. The method of claim 30, wherein the antibody is chemically coupled to a mitotic inhibitor.

35. The method of claim 32, wherein the antibody is chemically coupled to a mitotic inhibitor.

36. A method for treating cancer, the method comprising administering to a subject in need thereof a composition comprising a pharmaceutically acceptable vehicle and a monoclonal antibody comprising: a heavy chain comprising complementarity determining regions (CDR) CDR-H1, CDR-H2, and CDR-H3 comprising amino acid sequences SEQ ID Nos. 1, 2, and 3, respectively; a light chain comprising CDR-L1, CDR-L2, and CDR-L3, comprising amino acid sequences SEQ ID Nos. 5, 6, and 7; and a modified hinge region comprising the amino acid sequence SEQ ID No. 26.

37. The method of claim 36, wherein the antibody is a chimeric antibody.

38. The method of claim 36, wherein the antibody is a human antibody.

39. The method of claim 36, wherein the antibody is a humanized antibody.

40. The method of claim 36, wherein the antibody comprises: a heavy chain variable domain comprising the amino acid sequence SEQ ID No. 4; and a light chain variable domain comprising amino acid sequence SEQ ID No. 10.

41. The method of claim 40, wherein the antibody comprises a human light chain constant region and a human heavy chain constant region.

42. The method of claim 41, wherein the human light chain constant region is of the IgG1 kappa isotype.

43. The method of claim 36, wherein the antibody comprises a human light chain constant region and a human heavy chain constant region.

44. The method of claim 43, wherein the human light chain constant region is of the IgG1 kappa isotype.

45. The method of claim 41, wherein the antibody is chemically coupled to a mitotic inhibitor.

46. The method of claim 43, wherein the antibody is chemically coupled to a mitotic inhibitor.

47. The method according to claim 25, wherein the cancer is chosen from prostate cancer, osteosarcoma, lung cancer, breast cancer, endometrial cancer, glyoblastoma, and colon cancer.

48. The method according to claim 25, wherein the cancer is a c-Met-activation-related cancer chosen from c-Met-activation-related cancers that are HGF-dependent, HGF-independent, or both.

49. The method according to claim 12, further comprising administering an anti-tumoral antibody in a simultaneous, separate, or sequential fashion.

50. The method according to claim 12, further comprising administering cytotoxic/cytostatic agent in a simultaneous, separate, or sequential fashion.

51. The method according to claim 2546, further comprising administering an anti-tumoral antibody in a simultaneous, separate, or sequential fashion.

52. The method according to claim 25, further comprising administering cytotoxic/cytostatic agent in a simultaneous, separate, or sequential fashion.

53. The method according to claim 49, wherein at least one of said antibodies is conjugated with a cell toxin and/or a radioelement.

54. The method according to claim 50, wherein said cytotoxic/cytostatic agent is coupled chemically to the antibody.

55. The method according to claim 51, wherein at least one of said antibodies is conjugated with a cell toxin and/or a radioelement.

56. The method according to claim 52, wherein at least one of said antibodies is conjugated with a cell toxin and/or a radioelement.

Description

[0254] Other characteristics and advantages of the invention appear in the continuation of the description with the examples and the figures wherein:

[0255] FIG. 1: Effect of irrelevant IgG1 Mabs from mouse and human origin and PBS on c-Met receptor phosphorylation on A549 cells.

[0256] FIGS. 2A and 2B: Effect of murine and humanized 224G11 Mabs produced as a human IgG1/kappa isotype on c-Met receptor phosphorylation on A549 cells.

[0257] FIG. 2A: agonist effect calculated as percentage versus maximal stimulation of c-Met phosphorylation by HGF [100 ng/ml].

[0258] FIG. 2B: antagonist effect calculated as percentage of inhibition of the maximal stimulation of c-Met phosphorylation by HGF [100 ng/ml].

[0259] FIGS. 3A and 3B: Comparison between murine 224G11 Mab and chimeric 224G11 Mabs containing various engineered hinge regions, on c-Met receptor phosphorylation on A549 cells.

[0260] FIG. 3A: agonist effect calculated as percentage versus maximal stimulation of c-Met phosphorylation by HGF [100 ng/ml].

[0261] FIG. 3B: antagonist effect calculated as percentage of inhibition of the maximal stimulation of c-Met phosphorylation by HGF [100 ng/ml].

[0262] FIGS. 4A and 4B: Comparison between murine 224G11 Mab and chimeric and humanized 224G11 Mabs produced as a human IgG2/kappa isotype, on c-Met receptor phosphorylation on A549 cells.

[0263] FIG. 4A: agonist effect calculated as percentage versus maximal stimulation of c-Met phosphorylation by HGF [100 ng/ml].

[0264] FIG. 4B: antagonist effect calculated as percentage of inhibition of the maximal stimulation of c-Met phosphorylation by HGF [100 ng/ml].

[0265] FIGS. 5A and 5B: Comparison between murine 224G11 Mab and chimeric and humanized 224G11 Mabs produced as an engineered hinge mutant TH7IgG1/kappa, on c-Met receptor phosphorylation on A549 cells.

[0266] FIG. 5A: agonist effect calculated as percentage versus maximal stimulation of c-Met phosphorylation by HGF [100 ng/ml].

[0267] FIG. 5B: antagonist effect calculated as percentage of inhibition of the maximal stimulation of c-Met phosphorylation by HGF [100 ng/ml].

[0268] FIGS. 6A and 6B, FIGS. 7A and 7B, FIGS. 8A and 8B, FIGS. 9A and 9B, FIGS. 10A and 10B: BRET models with Figures A: c-Met dimerization model; and Figures B: c-Met activation model.

[0269] FIG. 11: c-Met recognition by chimeric and humanized 224G11 forms.

[0270] FIG. 12: Effect of murine and chimeric antibodies on HGF-induced proliferation of NCI-H441 cells in vitro. NCI-H441 cells were plated in serum-free medium. 24 hours after plating m224G11 and [224G11]chim were added either in absence or in presence of HGF. Black arrows indicate the wells plated with cells alone either in absence or in presence of HGF. A murine IgG1 (mIgG1) was introduced as an isotype control.

[0271] FIG. 13: In vivo comparison of murine and IgG1 chimeric 224G11 Mabs on the NCI-H441 xenograft model.

[0272] FIGS. 14A and 14B: Effect of the murine 224G11 Mab and of various chimeric and humanized versions of this antibody on HGF-induced proliferation of NCI-H441 cells in vitro. NCI-H441 cells were plated in serum-free medium. Twenty four hours after plating antibody to be tested were added either in absence or in presence of HGF. In panel (FIG. 14A), the murine m224G11, chimeric IgG1 [224G11]chim, humanized IgG1 [224G11] [Hz1], [224G11] [Hz2], [224G11] [Hz3] versions were shown. In panel (FIG. 14B), the murine m224G11 and various chimeric IgG1 forms ([224G11] chim, [224G11] [MH chim], [224G11] [MUP9H chim], [224G11] [MMCH chim], [224G11] [TH7 chim]) were presented. Black arrows indicate the wells plated with cells alone either in absence or in presence of HGF. A murine IgG1 was introduced as a negative control for agonist activity. The m5D5 was used as a dose-dependent full agonist control.

[0273] FIG. 15: Effect of the murine 224G11 Mab and of various chimeric and humanized versions of this antibody on HGF-induced proliferation of NCI-H441 cells in vitro. NCI-H441 cells were plated in serum-free medium. Twenty four hours after plating antibody to be tested were added either in absence or in presence of HGF. The murine m224G11, [224G11] chim, [224G11] [TH7 chim]) IgG1 chimeric forms and [224G11] [TH7 Hz1], [224G11] [TH7 Hz3]) were presented. Black arrows indicate the wells plated with cells alone either in absence or in presence of HGF. A murine IgG1 was introduced as a negative control for agonist activity. The m5D5 was used as a dose-dependent full agonist control.

[0274] FIG. 16: In vivo comparison of murine, chimeric and humanized 224G11 Mabs on the NCI-H441 xenograft model.

[0275] FIG. 17A: agonist effect calculated as percentage versus maximal stimulation of c-Met phosphorylation by HGF [100 ng/ml].

[0276] FIG. 17B: antagonist effect calculated as percentage of inhibition of the maximal stimulation of c-Met phosphorylation by HGF [100 ng/ml].

[0277] FIG. 18: BRET models with c-Met activation model.

[0278] FIG. 19: Effect of m224G11 and h224G11 on c-Met degradation on A549 cells. A) Mean of 4 independent experiments+/−s.e.m. B) Western blot image representative of the 4 independent experiments performed.

[0279] FIG. 20: Effect of m224G11 and h224G11 on c-Met degradation on NCI-H441 cells. A) Mean of 4 independent experiments+/−s.e.m. B) Western blot image representative of the 4 independent experiments performed.

[0280] FIG. 21: Set up of an ELISA to evaluate c-Met shedding.

[0281] FIG. 22: In vitro evaluation of c-Met shedding on NCI-H441 cells treated for 5 days with m224G11. mIgG1 is an irrelevant antibody used as an isotype control.

[0282] FIG. 23: In vitro evaluation of c-Met shedding on amplified Hs746T, MKN45 and EBC-1 cell lines treated for 5 days with m224G11. mIgG1 is an irrelevant antibody used as an isotype control. PMA is a shedding inducer used as a positive control.

[0283] FIG. 24: In vitro evaluation of c-Met shedding on NCI-H441 and amplified Hs746T, MKN45 and EBC-1 cell lines treated for 5 days with m224G11. mIgG1 is an irrelevant antibody used as an isotype control. PMA is a shedding inducer used as a positive control.

[0284] FIG. 25: Study of intrinsic phosphorylation of h224G11 on Hs746T cell line.

[0285] FIG. 26: Study of intrinsic phosphorylation of h224G11 on NCI-H441 cell line. A) phospho-ELISA and B) Western analysis.

[0286] FIG. 27: Study of intrinsic phosphorylation of h224G11 on Hs578T cell line. A) phospho-ELISA and B) Western analysis.

[0287] FIG. 28: Study of intrinsic phosphorylation of h224G11 on NCI-H125 cell line. A) phospho-ELISA and B) Western analysis.

[0288] FIG. 29: Study of intrinsic phosphorylation of h224G11 on T98G cell line. A) phospho-ELISA and B) Western analysis.

[0289] FIG. 30: Study of intrinsic phosphorylation of h224G11 on MDA-MB-231 cell line. A) phospho-ELISA and B) Western analysis.

[0290] FIG. 31: Study of intrinsic phosphorylation of h224G11 on PC3 cell line. A) phospho-ELISA and B) Western analysis.

[0291] FIG. 32: Study of intrinsic phosphorylation of h224G11 on HUVEC cells.

[0292] FIG. 33: In vivo comparison of the wild type murine 224G11 antibody with a chimeric hinge-engineered 224G11[C2D5-7] Mabs on the NCI-H441 xenograft model.

[0293] FIG. 34: ADCC induction by h224G11 on both Hs746T and NCI-H441 cells. .sup.51Cr-labeled Hs746T (A) or NCI-H441 (B) cells loaded (bold squares) or not (empty squares) with h224G11 were mixed with different ratio of human NK cells and incubated for 4 hr. Cells were harvested and cpm of .sup.51Cr released by lysis was counted. The results are plotted as percentage of lysis against the effector/target cell ratio. NL for non loaded cells.

[0294] FIG. 35: h224G11 staining in tumor xenograft which expressed various level of c-Met (A: Hs746T amplified cell line for c-Met, B: NCI-H441 high level of c-Met expression and C: MCF-7 low level of c-Met).

EXAMPLE 1: GENERATION OF ANTIBODIES AGAINST C-MET

[0295] To generate anti-c-Met antibodies 8 weeks old BALB/c mice were immunized either 3 to 5 times subcutaneously with a CHO transfected cell line that express c-Met on its plasma membrane (20×10.sup.6 cells/dose/mouse) or 2 to 3 times with a c-Met extracellular domain fusion protein (10-15 μg/dose/mouse) (R&D Systems, Catalog # 358MT) or fragments of this recombinant protein mixed with complete Freund adjuvant for the first immunization and incomplete Freund adjuvant for the following ones. Mixed protocols in which mice received both CHO-cMet cells and recombinant proteins were also performed. Three days before cell fusion, mice were boosted i.p. or i.v. with the recombinant protein or fragments. Then spleens of mice were collected and fused to SP2/0-Ag14 myeloma cells (ATCC) and subjected to HAT selection. Four fusions were performed. In general, for the preparation of monoclonal antibodies or their functional fragments, especially of murine origin, it is possible to refer to techniques which are described in particular in the manual “Antibodies” (Harlow and Lane, Antibodies: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor N.Y., pp. 726, 1988) or to the technique of preparation of hybridomas described by Kohler and Milstein (Nature, 256:495-497, 1975).

[0296] Obtained hybridomas were initially screened by ELISA on the c-Met recombinant protein and then by FACS analysis on A549 NSCLC, BxPC3 pancreatic, and U87-MG glioblastoma cell lines to be sure that the produced antibodies will be able to also recognize the native receptor on tumor cells. Positive reactors on these 2 tests were amplified, cloned and a set of hybridomas was recovered, purified and screened for its ability to inhibit in vitro cell proliferation in the BxPC3 model.

[0297] For that purpose 50 000 BxPC3 cells were plated in 96 well plates in RPMI medium, 2 mM L. Glutamine, without SVF. 24 hours after plating, antibodies to be tested were added at a final concentration ranging from 0.0097 to 40 μg/ml 60 min before addition of 100 ng/ml of hHGF. After 3 days, cells were pulsed with 0.5 μCi of [.sup.3H]thymidine for 16 hours. The magnitude of [.sup.3H]thymidine incorporated into trichloroacetic acid-insoluble DNA was quantified by liquid scintillation counting. Results were expressed as raw data to really evaluate the intrinsic agonistic effect of each Mab.

[0298] Then antibodies inhibiting at least 50% cell proliferation were evaluated for their activity on c-Met dimerization and activation BRET analysis on transfected cells. c-Met receptor activity was quantified by measuring the Gab1 signalling molecule recruitment on activated c-Met. For that purpose, CHO stable cell lines expressing C-Met-Rluc or C-Met-Rluc and C-Met-K1100A-YFP for c-Met dimerization or C-Met-Rluc and a mutated form of Gab1 [Maroun et al., Mol. Cell. Biol., 1999, 19:1784-1799] fused to YFP for c-Met activation were generated. Cells were distributed in white 96 well microplates in DMEM-F12/FBS 5% culture medium one or two days before BRET experiments. Cells were first cultured at 37° C. with CO.sub.2 5% in order to allow cell attachment to the plate. Cells were then starved with 200 μl DMEM/well overnight. Immediately prior to the experiment, DMEM was removed and cells quickly washed with PBS. Cells were incubated in PBS in the presence or absence of antibodies to be tested or reference compounds, 10 min at 37° C. prior to the addition of coelenterazine with or without HGF in a final volume of 50 μl. After incubation for further 10 minutes at 37° C., light-emission acquisition at 485 nm and 530 nm was initiated using the Mithras luminometer (Berthold) (1s/wave length/well repeated 15 times).

[0299] BRET ratio has been defined previously [Angers et al., Proc. Natl. Acad. Sci. USA, 2000, 97:3684-3689] as: [(emission at 530 nm)−(emission at 485 nm)×Cf]/(emission at 485 nm), where Cf corresponds to (emission at 530 nm)/(emission at 485 nm) for cells expressing Rluc fusion protein alone in the same experimental conditions. Simplifying this equation shows that BRET ratio corresponds to the ratio 530/485 nm obtained when the two partners were present, corrected by the ratio 530/485 nm obtained under the same experimental conditions, when only the partner fused to R. reniformis luciferase was present in the assay. For the sake of readability, results are expressed in milliBRET units (mBU); mBU corresponds to the BRET ratio multiplied by 1000.

[0300] After this second in vitro test, the antibody 224G11 i) without intrinsic activity as a whole molecule in the functional test of proliferation, ii) inhibiting significantly BxPC3 proliferation and iii) inhibiting c-Met dimerization was selected. In the experiments, the 5D5 Mab, generated by Genentech, and available at the ATCC, was added as a control for the intrinsic agonistic activity.

EXAMPLE 2: HUMANIZATION PROCESS OF MOUSE 224G11 MAB BY CDR-GRAFTING

[0301] 1°) Humanization of the Light Chain Variable Domain (VL)

[0302] As a preliminary step, the nucleotide sequence of 224G11 VL was compared to the murine germline gene sequences included in the IMGT database (http://imgt.cines.fr). Murine IGKV3-5*01 and IGKJ4*01 germline genes showing a sequence identity of 99.31% for the V region and 94.28% for the J region, respectively, have been identified. Regarding these high homologies, the 224G11VL nucleotide sequence has been used directly to search for human homologies, instead of corresponding mouse germlines.

[0303] In a second step, the human germline gene displaying the best identity with the 224G11VL has been searched to identify the best human candidate for the CDR grafting. For optimization of the selection, alignments between the amino acid sequences have been performed. The human IGKV4-1*01 germline gene yielded a sequence identity of 67.30%, but showed a different length for CDR1 (10 amino acids in 224G11 VL and 12 amino acids in IGKV4-1*01). For the J region, the human IGKJ4*02 germline gene (sequence identity of 77.14%) was selected.

[0304] In a next step, mouse 224G11 VL CDR regions were engrafted into the above selected human framework sequences. Each amino acid position was analyzed for several criteria such as participation in VH/VL interface, in antigen binding or in CDR structure, localization of the residue in the 3D structure of the variable domain, CDR anchors, residues belonging to the Vernier zone. Three humanized versions, corresponding to SEQ ID No. 8, SEQ ID No. 9 and SEQ ID No. 10 were constructed, and containing respectively four (4, 39, 40, 84), two (39, 40) or one (40) murine residues in their FR regions and the CDRs corresponding to mouse 224G11 VL.)

[0305] 2°) Humanization of the Heavy Chain Variable Domain (VH)

[0306] As a preliminary step, the nucleotidic sequence of the 224G11 VH was compared to the murine germline genes sequences included in the IMGT database (http://imgt.cines.fr).

[0307] Murine IGHV1-18*01, IGHD2-4*01 and IGHJ2*01 germline genes with a sequence identity of 92.70% for the V region, 75.00% for the D region and 89.36% for the J region, respectively, have been identified. Regarding these high homologies, it has been decided to use directly the 224G11VH nucleotide sequences to search for human homologies, instead of corresponding mouse germlines.

[0308] In a second step, the human germline gene displaying the best identity with the 224G11 VH has been searched to identify the best human candidate for the CDR grafting. To this end, the nucleotidic sequence of 224G11 VH has been aligned with the human germline genes sequences belonging to the IMGT database. The human IGHV1-2*02 V sequence exhibited a sequence identity of 75.00% at the nucleotide level and 64.30% at the amino acid level. Looking for homologies for the J region led to the identification of the human IGHJ4*04 germline gene with a sequence identity of 78.72%.

[0309] In a next step, mouse 224G11 VH CDR regions were engrafted into the above selected human framework sequences. Each amino acid position was analyzed for several criteria such as participation in VH/VL interface, in antigen binding or in CDR structure, localization of the residue in the 3D structure of the variable domain, CDR anchors, residues belonging to the Vernier zone. One fully humanized form, corresponding to SEQ ID 4 was constructed; it contains exclusively human residues in its FR regions and the CDRs corresponding to mouse 224G11 VH.

EXAMPLE 3: ENGINEERING OF IMPROVED HINGE MUTANTS

[0310] It is well known by the skilled artisan that the hinge region strongly participates in the flexibility of the variable domain of immunoglobulins (see Brekke et al., 1995; Roux et al., 1997). During the chimerization process of 224G11 Mab, the mouse constant domain IGHG1 was replaced by the equivalent IGHG1 portion of human origin. Since the amino acid sequence of the hinge region were highly divergent, “murinization” of the hinge region was performed in order to keep its length and rigidity. Since the human IGHG2 hinge region corresponds to the closest homologue of the mouse IGHG1 hinge, this sequence was as well considered. A series of 7 different hinge sequences were constructed (SEQ ID Nos. 22 to 28) by incorporating portions of the mouse IGHG1 and the human IGHG2 hinges into the human IGHG1 hinge portion.

[0311] Another series of hinge mutants was designed and constructed (SEQ ID Nos. 58 to 72) to evaluate the influence of either an additional cysteine and its position along the hinge domain, deletion of 1, 2, 3 or 4 amino acids along the hinge domain and a combination of these two parameters (cysteine addition and amino acid deletion).

EXAMPLE 4: PRODUCTION OF HUMANIZED 224G11 MAB AND ENGINEERED HINGE MAB FORMATS

[0312] All above described Mab forms containing either chimeric, humanized and/or engineered hinge regions were produced upon transient transfection and by using the HEK293/EBNA system with a pCEP4 expression vector (InVitrogen, US).

[0313] The entire nucleotide sequences corresponding to the humanized versions of the variable domain of 224G11 Mab light (SEQ ID No. 18, SEQ ID No. 19 and SEQ ID No. 20) and heavy (SEQ ID No. 14) chains were synthesized by global gene synthesis (Genecust, Luxembourg). They were subcloned into a pCEP4 vector (InVitrogen, US) carrying the entire coding sequence of the constant domain [CH1-Hinge-CH2-CH3] of a human IgG1 or IgG2 immunoglobulin. Modification of the hinge region was performed by exchanging a {NhelI-Bcl1} restriction fragment by the equivalent portion carrying the desired modifications, each respective {Nhel-Bcl1} fragment being synthesized by global gene synthesis (Genecust, LU). All cloning steps were performed according to conventional molecular biology techniques as described in the Laboratory manual (Sambrook and Russel, 2001) or according to the supplier's instructions. Each genetic construct was fully validated by nucleotide sequencing using Big Dye terminator cycle sequencing kit (Applied Biosystems, US) and analyzed using a 3100 Genetic Analyzer (Applied Biosystems, US).

[0314] Suspension-adapted HEK293 EBNA cells (InVitrogen, US) were routinely grown in 250 ml flasks in 50 ml of serum-free medium Excell 293 (SAFC Biosciences) supplemented with 6 mM glutamine on an orbital shaker (110 rpm rotation speed). Transient transfection was performed with 2.Math.10.sup.6 cells/ml using linear 25 kDa polyethyleneimine (PEI) (Polysciences) prepared in water at a final concentration of 1 mg/ml mixed and plasmid DNA (final concentration of 1.25 μg/ml for heavy to light chain plasmid ratio of 1:1). At 4 hours post-transfection, the culture was diluted with one volume of fresh culture medium to achieve a final cell density of 10.sup.6 cells/ml. Cultivation process was monitored on the basis of cell viability and Mab production. Typically, cultures were maintained for 4 to 5 days. Mabs were purified using a conventional chromatography approach on a Protein A resin (GE Healthcare, US).

[0315] All different forms of Mabs were produced at levels suitable with functional evaluations. Productivity levels are typically ranging between 15 and 30 mg/l of purified Mabs.

EXAMPLE 5: EVALUATION OF C-MET PHOSPSHORYLATION STATUS BY A PHOSPHO-C-MET-SPECIFIC ELISA ASSAY

[0316] This functional assay allows to monitor modulation c-Met phosphorylation status either by Mabs alone or in the co-presence of HGF.

[0317] A549 cells were seeded in a 12MW plate in complete growth medium [F12K+10% FCS]. Cells were starved for 16 hours before stimulation with HGF [100 ng/ml], and each Mab to be tested was added at its final concentration of 30 μg/ml 15 minutes prior to ligand stimulation. Ice-cold lysis buffer was added 15 minutes after the addition of HGF to stop the phosphorylation reaction. Cells were scaped mechanically and cell lysates were collected by centrifugation at 13000 rpm for 10 min. at 4° C. and correspond to the supernatant phase. Protein content was quantified using a BCA kit (Pierce) and stored at −20° C. until use. The phosphorylation status of c-Met was quantified by ELISA. A goat anti-c-Met Mab (R&D, ref AF276) was used as a capture antibody (overnight coating at 4° C.) and after a saturation step with a TBS-BSA 5% buffer (1 hour at room temperature (RT)), 25 μg of protein lysates were added to each well of the coated 96MW plate. After a 90 minutes incubation at RT, plates were washed four time and the detection antibody was added (anti-phospho-c-Met Mab, directed against the phopshorylated Tyr residues at position 1230, 1234 and 1235). After an additional 1 hour incubation and 4 washes, an anti-rabbit antibody coupled to HRP (Biosource) was added for 1 hour at RT, and the luminescence detection was performed by adding Luminol. Luminescence readings were on a Mithras LB920 multimode plate reader (Berthold).

[0318] Both basal and HGF [100 ng/ml]-induced c-Met receptor phosphorylation level were unaffected neither by PBS treatment, nor by the addition of mouse or human Mabs which do not target human c-Met receptor (FIG. 1). On the other hand, mouse (m) 224G11 Mab strongly inhibited HGF [100 ng/ml]-induced c-Met phosphorylation (FIG. 2B) without altering by itself receptor phosphorylation (FIG. 2A). Surprisingly, the chimeric form of 224G11 Mab (224G11chim/IgG1), meaning variable domain (VH+VL) from m224G11 combined with human constant domain IgG1/kappa yielded strong (17% of maximal HGF effect, FIG. 2A) agonist activity associated with a reduced antagonist efficacy (54% inhibition of HGF maximal effect compared to the m224G11 that yields 75% inhibition of HGF maximum effect, FIG. 2B). Three humanized forms of 224G11 Mab, [224G11]Hz1/IgG1, [224G11]Hz2/IgG1 and [224G11]Hz3/IgG1, also constructed on a human IgG1/kappa backbone, yielded also decreased antagonist efficacy and significant agonist activity (11 to 24% of maximal HGF level) as compared to mouse 224G11 (FIGS. 2A and 2B). A series of engineered versions of the heavy chain hinge domain were constructed and assayed in the c-Met receptor phosphorylation assay. As shown in FIG. 3A, an important reduction of the agonist effect associated with the hIgG1/kappa isotype was observed for both the IgG2-based construct and for engineered IgG1/kappa constructs [MH, MUP9H and TH7]. A concomitant increase in antagonist efficacy was as well obtained. The hIgG1/kappa-based TH7 hinge mutant, with the most human sequence, was selected to complete the humanization process. In a next step, three humanized versions of 224G11 Mab variable domain were generated by combination to either a human IgG2/kappa or an IgG1/kappa-based TH7 engineered hinge constant domain. For the hIgG2/kappa humanized constructs, the humanized version Hz3 yielded strong agonism (FIG. 4A), and for all three humanized versions, the antagonist efficacy was below that observed with murine 224G11 Mab and comparable to the chimeric hIgG1-based Mab (56-57% inhibition of HGF effect, FIG. 4B). On the other hand, combination of the three humanized versions Hz1, Hz2 or Hz3 to the engineered IgG1/TH7 mutant almost fully restored the properties of mouse 224G11 Mab in terms of weak agonist activity (5-6% of HGF effect) and strong antagonist efficacy (68 to 72% inhibition of HGF effect) of c-Met receptor phosphorylation (FIGS. 5A and 5B). These variants were highly improved as compared to chimeric IgG1-based 224G11 Mab but also to IgG2-based humanized forms.

[0319] A second series of engineered versions of the heavy chain hinge domain was constructed and assayed in the c-Met receptor phosphorylation assay. As shown in FIG. 17A, all those new versions (c224G11[C2], c224G11[C3], c224G11[C5], c224G11[C6], c224G11[C7], c224G11[Δ1-3], c224G11[C7Δ6], c224G11[C6Δ9], c224G11[C2Δ5-7], c224G11[C5Δ2-6], c224G11[C9Δ2-7] and c224G11[Δ5-6-7-8]) exhibited weaker agonist effect than c224G11 since their agonism activities are comprised between 6 and 14% of the HGF effect compared to 23% for c224G11. As c224G11[TH7], all those new versions exhibited a concomitant increase in antagonist efficacy [FIG. 17B]. Those results showed that engineering of the heavy chain domain by point mutation and/or deletion could modify agonistic/antagonistic properties of an antibody.

EXAMPLE 6: BRET ANALYSIS

[0320] In a first set of experiments, it had been control that irrelevant mouse IgG1, human IgG1 and human IgG2 had no effect of HGF induced BRET signal in both BRET models (representative experiment out of 12 independent experiments; FIG. 6). These Mabs are forthwith cited as controls.

[0321] The effect of a IgG1 chimeric form of mouse 224G11 Mab ([224G11]chim) on both c-Met dimerization and c-met activation BRET model was evaluated. While mouse 224G11 Mab inhibited 59.4% of the HGF induced BRET signal on c-Met dimerization model, [224G11]chim Mab inhibited only 28.9% (FIG. 7A). [224G11]chim antibody was also less effective in inhibiting HGF induced c-Met activation since [224G11]chim and m224G11 antibodies inhibited respectively 34.5% and 56.4% of HGF induced BRET signal (FIG. 7B). Moreover, m224G11 alone had no effect on c-Met activation while [224G11]chim had a partial agonist effect on c-Met activation corresponding to 32.9% of the HGF induced signal. This partial agonist effect of the [224G11]chim was also seen on c-Met dimerization BRET model since [224G11]chim alone induced a BRET increase corresponding to 46.6% of HGF-induced signal versus 21.3% for m224G11 (FIG. 7A).

[0322] In FIGS. 8A and 8B, hinge mutated chimeric forms of 224G11 antibody showed a greater inhibitory effect on HGF induced BRET signal than [224G11]chim since they showed a 59.7%, 64.4%, 53.2% and 73.8% inhibition of the HGF induced activation BRET signal (FIG. 8B) and 61.8%, 64.4% 52.5% and 64.4% inhibition of the HGF induced c-Met dimerization BRET signal (FIG. 8A) for [224G11][MH chim], [224G11][MUP9H chim], [224G11][MMCH chim] and [224G11][TH7 chim] respectively. Contrary to [224G11]chim, which had a partial agonist effect on c-Met activation, hinge mutated chimerical forms of 224G11 antibody showed no significant effect on c-Met activation alone (5.1%, 7.6%, −2.0% and -6.9% respectively) as observed for m224G11.

[0323] In FIG. 9B, like the [224G11] [TH7 chim], the 3 humanized versions of 224G11 IgG1 antibody with the TH7 hinge induced no significant increased of BRET signal in activation model when tested alone and showed a strong inhibition of HGF induced BRET signal: 59.9%, 41.8% and 57.9% for the Hz1, Hz2 and Hz3 forms respectively. Moreover, [224G11] [TH7 Hz1], [224G11] [TH7 Hz2] and [224G11] [TH7 Hz3] inhibited HGF induced BRET signal on dimerization model of 52.2%, 35.8% and 49.4% respectively (FIG. 9A).

[0324] Contrary to [224G11]chim, the chimeric form of 224G11 IgG2 antibody ([224G11] [IgG2 chim]) showed no partial agonist effect alone and inhibited 66.3% of the HGF effect on c-Met activation model (FIG. 10B). On c-Met dimerization model, [224G11] [IgG2 chim] inhibited 62.4% of the HGF induced BRET signal (FIG. 10A).

[0325] The agonist efficacy of the second series of engineered versions of the heavy chain hinge domain was evaluated in c-Met activation BRET model (FIG. 18). In contrast to c224G11, which had a partial agonist effect on c-Met activation, c224G11[C2], c224G11[C3], c224G11[C5], c224G11[C6], c224G11[C7], c224G11[Δ1-3], c224G11[C7Δ6], c224G11[C6Δ9], c224G11[C2Δ5-7], c224G11[C5Δ2-6], c224G11[C9Δ2-7] and c224G11[Δ5-6-7-8] hinge mutated chimeric forms of 224G11 antibody showed no significant effect on c-Met activation alone.

EXAMPLE 7: C-MET RECOGNITION BY CHIMERIC AND HUMANIZED 224G11 FORMS

[0326] A direct ELISA has been set up to determine the binding ability of the various chimeric and humanized forms on the recombinant c-Met. Briefly recombinant dimeric c-Met from R&D Systems was coated at 1.25 μg/ml on 96-well Immunlon II plates. After an overnight incubation at 4° C., wells were saturated with a 0.5% gelatine/PBS solution. Plates were then incubated for 1 hour at 37° C. before addition of 2 fold dilutions of antibodies to be tested. Plates were incubated an additional hour before addition of a goat anti-mouse IgG HRP for detecting the murine antibody and a goat anti-human Kappa light chain HRP for chimeric and humanized antibody recognition. Plates were incubated for one hour and the peroxydase substrate TMB Uptima was added for 5 mn before neutralization with H.sub.2SO.sub.4 1 M. Results presented in FIG. 11 showed that all tested forms were comparable for c-Met recognition.

EXAMPLE 8: EFFECT OF MURINE AND CHIMERIC 224G11 ON HGF-INDUCED PROLIFERATION OF NCI-H441 CELLS IN VITRO

[0327] NCI-H441 cells from ATCC were routinely cultured in RPMI 1640 medium (Invitrogen Corporation, Scotland, UK), 10% FCS (Invitrogen Corporation), 1% L-Glutamine (Invitrogen corporation). For proliferation assays, cells were split 3 days before use so that they were in the confluent phase of growth before plating. NCI-H441 cells were plated in 96-well tissue culture plates at a density of 3.75×10.sup.4 cells/well in 200 μl of serum free medium (RPMI 1640 medium plus 1% L-Glutamine). Twenty four hours after plating, antibodies to be tested were added to NCI-H441 and incubated at 37° C. for thirty minutes before adding HGF at a final concentration of 400 ng/ml (5 nM) for 142 additional hours. The dose range tested for each antibody is from 10 to 0.0097 μg/ml (final concentration in each well). In this experiment, a murine IgG1 Mab was added as a murine isotype control and the tested antibodies were the following one: m224G11 and its human IgG1 chimeric form identified as [224G11]chim. Wells plated with cells alone −/+HGF were also included. Then cells were pulsed with 0.25 μCi of [.sup.3H]Thymidine (Amersham Biosciences AB, Uppsala, Sweden) for 7 hours and 30 minutes. The magnitude of [.sup.3H]Thymidine incorporated in trichloroacetic acid-insoluble DNA was quantified by liquid scintillation counting. Results are expressed as non transformed cpm data to better evaluate the potential intrinsic agonist activity that could occur with anti-c-Met Mabs when added alone to tumour cell.

[0328] Results described in FIG. 12 demonstrated that, as expected, the murine antibody m224G11 displayed no agonist effect when added alone to cancer cells whatever the tested dose. No significant inhibition of the HGF-induced proliferation was observed with the isotype control regarding to the cpm variations observed for this compound in this experiment. When added alone, the m224G11 antibody did not show any agonist effect compared to the mIgG1 isotype control Mab or cells alone. A dose dependent anti-proliferative activities reaching 78% was observed for m224G11 (% inhibition calculation: 100−[(cpm cells+Mab to be tested-mean cpm background mIgG1)×100/(mean cpm cells+HGF−mean cpm cells alone)]). Surprisingly, the chimeric form of the 224G11 Mabs induced a significant, dose dependent agonist effect when added alone. This agonist effect had an impact on the in vitro inhibition of HGF− induced proliferation that shifted from 78% for the murine 224G11 to 50% for its chimeric form. To determine whether such “lower” in vitro intrinsic agonist activity was compatible with an unchanged in vivo effect, both m224G11 and [224G11]chim were produced for in vivo testing. As, in previous studies, the 30 μg/mice dose had demonstrated a significant in vivo activity, that dose was selected for in vivo evaluation.

EXAMPLE 9: IN VIVO COMPARISON OF MURIN AND CHIMERIC 224G11 MABS ON THE NCI-H441 XENOGRAFT MODEL

[0329] NCI-H441 is derived from papillary lung adenocarcinoma, expresses high levels of c-Met, and demonstrates constitutive phosphorylation of c-Met RTK.

[0330] To evaluate the in vivo effect of antibodies on the NCI-H441 xenograft model, six to eight weeks old athymic mice were housed in sterilized filter-topped cages, maintained in sterile conditions and manipulated according to French and European guidelines. Mice were injected subcutaneously with 9×10.sup.6 cells. Then, six days after cell implantation, tumors were measurable (approximately 100 mm.sup.3), animals were divided into groups of 6 mice with comparable tumor size and treated first with a loading dose of 60 μg of antibody/mice and then twice a week with 30 μg/dose of each antibody to be tested. The mice were followed for the observation of xenograft growth rate. Tumor volume was calculated by the formula: π(Pi)/6×length×width×height. Results described in FIG. 13 demonstrate that the murine Mab devoided of agonist activity in vivo behave, as expected, as potent antagonist even at the low tested dose. In contrast to what observed with the murine Mab, the chimeric one displayed a very transient in vivo activity and tumor completely escaped to the treatment at D20 post cell injection. This experiment demonstrates clearly that the increase of in vitro agonist effect that resulted in a decrease of antagonist activity was also responsible for a significant in vivo loss of antagonist activity.

EXAMPLE 10: EFFECT OF THE MURINE 224G11 MAB AND OF VARIOUS CHIMERIC AND HUMANIZED VERSIONS OF THIS ANTIBODY ON HGF-INDUCED PROLIFERATION OF NCI-H441 CELLS IN VITRO

[0331] NCI-H441 cells from ATCC were routinely cultured in RPMI 1640 medium (Invitrogen Corporation, Scotland, UK), 10% FCS (Invitrogen Corporation), 1% L-Glutamine (Invitrogen Corporation). For proliferation assays, cells were split 3 days before use so that they were in the confluent phase of growth before plating. NCI-H441 cells were plated in 96-well tissue culture plates at a density of 3.75×10.sup.4 cells/well in 200 μl of serum free medium (RPMI 1640 medium plus 1% L-Glutamine). Twenty four hours after plating, antibodies to be tested were added to NCI-H441 and incubated at 37° C. for thirty minutes before adding HGF at a final concentration of 400 ng/ml (5 nM) for 142 additional hours. The dose range tested for each antibody is from 10 to 0.0097 μg/ml (final concentration in each well). In this experiment, murine IgG1 Mab was added as a murine isotype control and as an agonist negative control. The tested antibodies were the following one: i) m224G11, ii) its human IgG1 chimeric forms respectively identified as [224G11] chim, [224G11] [MH chim], [224G11] [MUP9H chim], [224G11] [MMCH chim], [224G11] [TH7 chim] iii) its humanized IgG1 forms respectively described as [224G11] [Hz1], [224G11] [Hz2], [224G11] [Hz3]. Wells plated with cells alone −/+HGF were also included. The 5D5 whole antibody from Genentech commercially available at the ATCC as an hybridoma cell line was introduced as a full agonist positive control and thereafter called m5D5. Then cells were pulsed with 0.25 μCi of [.sup.3H]Thymidine (Amersham Biosciences AB, Uppsala, Sweden) for 7 hours and 30 minutes. The magnitude of [.sup.3H]Thymidine incorporated in trichloroacetic acid-insoluble DNA was quantified by liquid scintillation counting. Results are expressed as non transformed cpm data to better evaluate the potential intrinsic agonist activity that could occur with anti-c-Met Mabs when added alone to tumour cell.

[0332] Results described in FIG. 14A demonstrated that as expected neither the isotype control nor the m224G11 displayed any agonist activity on NCI-H441 proliferation. The isotype control was without effect on HGF-induced cell proliferation whereas m224G11 showed a 66% inhibition when added at the final concentration of 10 μg/ml. The m5D5 used as an agonist control showed, as expected, a full dose dependent agonist effect when added alone to the cells. As already observed, the [224G11] chim Mab displayed a significant dose-dependent agonist effect and, a decreased inhibitory activity of this chimeric form was observed: 19% instead of 66% for the murine form. When added alone, the 3 IgG1 humanized Mabs demonstrated dose dependent agonist effects compared to the m224G11 form. [224G11] [Hz1], [224G11] [Hz2] and [224G11] [Hz3] had comparable antagonist activities about 46, 30 and 35%. These activities are significantly lower than the one observed for m224G11. In FIG. 14B, various IgG1 chimeric forms were tested. Compared to [224G11] chim form which displayed a dose-dependent agonist effect when added alone to NCI-H441 cells, the [224G11] [MH chim], [224G11] [MUP9H chim], [224G11] [MMCH chim], [224G11] [TH7 chim] forms were without significant intrinsic agonist effect. Their antagonist activity was higher than the one observed for the m224G11 Mab (57%) with inhibitions reaching 79, 78, 84 and 93% respectively for [224G11] [MH chim], [224G11] [MUP9H chim], [224G11] [MMCH chim] and [224G11] [TH7 chim].

EXAMPLE 11: IN VITRO EFFECT OF VARIOUS IGG1 HUMANIZED FORM OF THE 224G11 MAB

[0333] NCI-H441 cells from ATCC were routinely cultured in RPMI 1640 medium (Invitrogen Corporation, Scotland, UK), 10% FCS (Invitrogen Corporation), 1% L-Glutamine (Invitrogen Corporation). For proliferation assays, cells were split 3 days before use so that they were in the confluent phase of growth before plating. NCI-H441 cells were plated in 96-well tissue culture plates at a density of 3.75×10.sup.4 cells/well in 200 μl of serum free medium (RPMI 1640 medium plus 1% L-Glutamine). Twenty four hours after plating, antibodies to be tested were added to NCI-H441 and incubated at 37° C. for thirty minutes before adding HGF at a final concentration of 400 ng/ml (5 nM) for 142 additional hours. The dose range tested for each antibody is from 10 to 0.0097 μg/ml (final concentration in each well). In this experiment, murine IgG1 Mab was added as a background negative control for agonist activity and the tested antibodies were the following one: i) m224G11, ii) its human IgG1 chimeric forms respectively identified as [224G11] chim, [224G11] [TH7 chim] iii) its humanized IgG1 forms respectively described as [224G11] [TH7 Hz1], [224G11] [TH7 Hz3]. Wells plated with cells alone −/+HGF were also included. The 5D5 whole antibody from Genentech commercially available at the ATCC as an hybridoma cell line was introduced as a full agonist positive control and thereafter called m5D5. Then cells were pulsed with 0.25 μCi of [.sup.3H]Thymidine (Amersham Biosciences AB, Uppsala, Sweden) for 7 hours and 30 minutes. The magnitude of [.sup.3H]Thymidine incorporated in trichloroacetic acid-insoluble DNA was quantified by liquid scintillation counting. Results are expressed as non transformed cpm data to better evaluate the potential intrinsic agonist activity that could occur with anti-c-Met Mabs when added alone to tumour cell.

[0334] FIG. 15 showed that the m224G11 Mab displayed the usual inhibitory effect (74% inhibition). The chimeric IgG1 form [224G11] chim had as expected a dose dependent intrinsic agonist effect and a lower antagonist effect compared to the murin form: 33% versus 74% inhibition. The [224G11] [TH7 chim] had a very weak agonist activity in this experiment. However it displayed a high inhibitory effect (81%) close to the one noticed for the murine Mab. The 2 humanized forms had no intrinsic agonist effect and had an antagonist activity close to the ones observed for the murine Mab or the [224G11] [TH7 chim] with respectively 67 and 76% inhibition for [224G11] [TH7 Hz1] and [224G11] [TH7 Hz3].

EXAMPLE 12: IN VIVO COMPARISON OF MURIN, CHIMERIC AND HUMANIZED 224G11 MABS BEARING EITHER THE WILD TYPE OR THE TH7-ENGINEERED HINGE (NCI-H441 XENOGRAFT MODEL)

[0335] NCI-H441 is derived from papillary lung adenocarcinoma, expresses high levels of c-Met, and demonstrates constitutive phosphorylation of c-Met RTK.

[0336] To evaluate the necessity of hinge engineering to save in vivo activity of the 224G11 murine antibody, six to eight weeks old athymic mice were housed in sterilized filter-topped cages, maintained in sterile conditions and manipulated according to French and European guidelines. Mice were injected subcutaneously with 9×10.sup.6 NCI-H441 cells. Then, six days after cell implantation, tumors were measurable (approximately 100 mm.sup.3), animals were divided into groups of 6 mice with comparable tumor size and treated first with a loading dose of 2 mg of antibody/mice and then twice a week with a 1 mg/dose of each antibody to be tested. Ten antibodies were evaluated in this experiment including the m224G11, the chimeric form displaying the wild type hinge (c224G11), the TH7-engineered chimeric form (224G11[TH7 chim]), three humanized form bearing the wild type hinge (224G11[IgG1 Hz1], 224G11[IgG1 Hz2] and 224G11[IgG1 Hz3]) and the three corresponding TH7-engineered forms (224G11[TH7 Hz1], 224G11[TH7 Hz2] and 224G11[TH7 Hz3]). Mice were followed for the observation of xenograft growth rate.

[0337] Tumor volume was calculated by the formula: π(Pi)/6×length×width×height.

[0338] Results described in FIG. 16 demonstrate that the murine Mab devoid of any agonist activity in vitro behave, as expected, as potent in vivo antagonist. In contrast to what observed with the murine Mab, both chimeric and humanized forms bearing the wild type hinge displayed only a very transient in vivo activity. In any cases the substitution of the wild type hinge by the TH7-engineered one resulted in a complete restoration of the in vivo activity observed with murine antibodies. This experiment demonstrates clearly that the increase of in vitro agonist effect that resulted in a decrease of antagonist activity was also responsible of a significant in vivo loss of antagonist activity. It also demonstrates that the use of a TH7-engineered region instead of the wild type one is needed for keeping the in vivo properties of the murine Mab.

EXAMPLE 13: EFFECT OF M224G11 AND ITS HUMANIZED FORM H224G11 ON C-MET DOWNREGULATION IN VITRO

[0339] In the following examples, for the avoidance of doubt, the expression h224G11 refers to the humanized form 224G11 [TH7 Hz3] of the antibody of the invention.

[0340] Two cell lines have been selected to address the activity of anti-c-Met antibodies on c-Met receptor degradation. A549 (#HTB-174) and NCI-H441 (#CCL-185) are two NSCLC cell lines from the ATCC collection. NCI-H441 cells were seeded in RPMI 1640+1% L-glutamine+10% heat-inactivated FBS, at 3×10.sup.4 cells/cm.sup.2 in six-well plates for 24 h at 37° C. in a 5% CO.sub.2 atmosphere. A549 cells were seeded in F12K+10% heat-inactivated FBS, at 2×10.sup.4 cells/cm.sup.2 in six-well plates for 24 h at 37° C. in a 5% CO.sub.2 atmosphere.

[0341] Then, cells were washed twice with phosphate buffer saline (PBS) before being serum-starved for 24 additional hours. Anti-c-Met antibodies (10 μg/ml), irrelevant mIgG1 (10 μg/ml), or HGF (400 ng/mL) were added in serum-free DMEM medium at 37° C. After either 4 hours or 24 hours of incubation, the medium was gently removed and cells washed twice with cold PBS. Cells were lysed with 500 μL it of ice-cold lysis buffer [50 mM Tris-HCl (pH 7.5); 150 mM NaCl; 1% Nonidet P40; 0.5% deoxycholate; and 1 complete protease inhibitor cocktail tablet plus 1% antiphosphatases]. Cell lysates were shaken for 90 min at 4° C. and cleared at 15 000 rpm for 10 minutes. At this stage, cell lysates could be stored at −20° C. until needed for western blot analysis. Protein concentration was quantified using BCA. Whole cell lysates (5 μg in 20 μl) were separated by SDS-PAGE and transferred to nitrocellulose membrane. Menbranes were saturated for 1 h at RT with TBS-Tween 20 0.1% (TBST); 5% non-fat dry milk and probed with anti-c-Met antibody (dilution 1/1000) overnight at 4° C. in TBST-5% non-fat dry milk. Antibodies were diluted in tris-buffered saline-0.1% tween 20 (v/v) (TBST) with 1% non-fat dry milk. Then, membranes were washed with TBST and incubated with peroxydase-conjugated secondary antibody (dilution 1:1000) for 1 h at RT. Immunoreactive proteins were visualized with ECL (Pierce #32209). After c-Met visualization, membranes were washed once again with TBST and incubated for 1 h at RT with mouse anti-GAPDH antibody (dilution 1/200 000) in TBST-5% non-fat dry milk. Then, membranes were washed in TBST and incubated with peroxydase-conjugated secondary antibodies, for 1 h at RT. Membranes were washed and GAPDH was revealed using ECL. Band intensity was quantified by densitometry.

[0342] Results presented in FIGS. 19A and 20A demonstrated that both m224G11 and h224G11 are able to significantly downregulate c-Met, in a dose-dependant way, in both A549 and NCI-H441 cell lines. The downregulation is already significant after a 4 hour incubation time and still increased at 24 hour. Histograms presented in FIGS. 19A and 20A corresponds to mean values or respectively 4 and 3 independent experiments. Western blot images corresponding to one significant experiment were included in FIGS. 19B and 20B.

EXAMPLE 14: EFFECT OF M224G11 AND ITS HUMANIZED FORM H224G11 ON C-MET SHEDDING IN VITRO

[0343] Soluble shedded forms of the c-Met receptor occur naturally in the serum of mice xenografted with human tumor or in serum of human patient carrying tumors expressing c-Met. Moreover, antibodies directed against c-Met such as the DN30 Mab, are described as shedding inducers of c-Met in in vitro experiments. To determine whether the m224G11 as such a property, cells were seeded in six-well plates in 10% FCS medium. When they reached approximately 80% confluence, medium was removed and fresh complete culture medium +/−compounds to be tested was added. Cells were incubated 72 additional hours with either m224G11, an isotype control mIgG1 or PBS. PMA (phorbol meristate acetate) was introduced as a shedding inducer. HGF was also tested on cells to determine the impact of c-Met ligand on natural occurring shedding. Then supernatants were collected and filtered on 0.2 μm before use in an ELISA test which soluble forms of c-Met were captured with an anti-c-Met antibody that does not recognize the same epitope as either m224G11 or the c11E1 (FIG. 21). Moreover, cells from each well were washed once with PBS and lysed to determine protein concentration. For the ELISA, 224D10 was used as a capture antibody and after plate saturation, filtered supernatants from six well plates were added in the ELISA test. A monomeric c-Met form was used as a positive control. After supernatant incubation, plates were washed to remove the unbound c-Met and c11E1 was used to detect c-Met captured by the 224G11 Mab. The revelation of the test was finally performed by addition of an HRP-conjugated anti-hFc polyclonal antibody.

[0344] Results shown in FIG. 22 indicate that a natural shedding of c-Met occurred when cells were cultured for 72 hours in vitro. No effect of the mIgG1 was observed. However, the addition of m224G11 seemed to inhibit c-Met shedding. These results were confirmed for 3 other cells lines (Hs746T, EBC1 and MKN45) in FIG. 23. In that second experiment, the PMA, added as a positive shedding inducer, increased significantly, as expected, c-Met shedding at least in 2 cell lines (Hs746T and MKN45). Finally, in a third experiment (FIG. 24), HGF was introduced as a control. No additional shedding was induced by HGF compared as cells alone or cells+mIgG1. Once again, a significant inhibition of c-Met shedding was observed with m224G11.

EXAMPLE 15: INTRINSIC EFFECT OF H224G11 AB ON VARIOUS CELL LINES

[0345] In previous experiments described in this patent, it has been demonstrated that in contrast to what was observed with other antibodies such as 5D5, the m224G11 and its humanized form h224G11 do not display significant intrinsic activity tumor cell lines. To extend this property to other cell lines, western blot and phospho-ELISA experiments were performed with the antibody alone, added for various times, on a set of cancer cell lines, with variable levels of c-Met expression, including Hs746T, NCI-H441, Hs578T, NCI-H125, T98G, MDA-MB-231, PC3. The same test was also performed in a normal cell: HUVEC.

[0346] Method for the phospho cMet ELISA assay was already described in example 5 of the present patent application. For the western analysis, protein lysates were made from pelleted cells by incubation in lysis buffer with proteases and phosphatase inhibitors [10 nM Tris (pH 7.4), 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 0.5% Nonidet P40, 100 mM sodium fluoride, 10 mM sodium pyrophosphate, 2 mM sodium orthovanadate, 2 mM PMSF, 10 mg/ml leupeptin, 10 mg/ml aprotinin] at 4° C. Protein lysates were cleared of cellular debris by centrifugation, resolved by electrophoresis on 8% SDS-PAGE gels, and electrotransferred to a nitrocellulose membrane. For c-Met experiments, lysates were immunoprecipitated for specific protein of interest before electrophoresis and transfer.

[0347] Results presented in FIGS. 25 to 32 demonstrate once again that no intrinsic activity of the h224G11 antibody was observed in the tested cells.

EXAMPLE 16: IN VIVO COMPARISON OF THE MURIN WILD TYPE 224G11 WITH A CHIMERIC HINGE-ENGINEERED 224G11 FORM DESCRIBED AS 224G11[C2D5-7] (NCI-H441 XENOGRAFT MODEL)

[0348] NCI-H441 is derived from papillary lung adenocarcinoma, expresses high levels of c-Met, and demonstrates constitutive phosphorylation of c-Met RTK.

[0349] To evaluate the necessity of hinge engineering to save in vivo activity of the 224G11 murine antibody, six to eight weeks old athymic mice were housed in sterilized filter-topped cages, maintained in sterile conditions and manipulated according to French and European guidelines. Mice were injected subcutaneously with 9×10.sup.6 NCI-H441 cells. Then, six days after cell implantation, tumors were measurable (approximately 100 mm.sup.3), animals were divided into groups of 6 mice with comparable tumor size and treated first with a loading dose of 2 mg of antibody/mice and then twice a week with a 1 mg/dose of each antibody to be tested. Mice were followed for the observation of xenograft growth rate. Tumor volume was calculated by the formula: π(Pi)/6×length×width×height. Results described in FIG. 33 demonstrate that the murine Mab devoid of any agonist activity in vitro behave, as expected, as a potent in vivo antagonist. As suggested by the results obtained in vitro, in phosphorylation assays, the c224G11[C2D5-7] hinge-engineered antibody, that did not display a significant agonist effect, demonstrate a strong in vivo activity, comparable to the one of the m224G11 on the NCI-H441 xenograft model.

EXAMPLE 17: EVALUATION OF H224G11 IN AN ADCC TEST

[0350] As h224G11 is of IgG1 isotype, ADCC could be part of its in vivo efficacy in human. An in vitro [.sup.51Cr] release cytotoxicity assay was performed using either Hs746T or NCI-H441 cells as target cells and NK cells purified from human peripheral blood mononuclear lymphocytes.

[0351] Briefly, one million Hs746T or NCI-H441 target cells were incubated with or without 20 μg of h224G11 Ab in presence of 100 μCi of .sup.51Chromium (Perkin Elmer) for 1 hr. Then, 4×10.sup.3 cells were plated with an increasing number of human natural killer (NK) cells isolated from peripheral blood mononuclear cells (PBMC) using a negative selection (Stemcell Technologies). Cells were incubated together for 4 additional hours at 37° C. Percent of cell lysis was calculated following the formula: [(experimental .sup.51Cr release−spontaneous .sup.51Cr release)/(full .sup.51Cr release−spontaneous .sup.51Cr release)]×100. Spontaneous release represents the counts obtained when the target cells were cultured in absence of natural killer cells. Full release represents the counts obtained when the target cells were lysed with 1% Triton X-100. h224G11 significantly enhanced lysis of both Hs746T (FIG. 34a) and NCI-H441 (FIG. 34b) cells by 62.9% and 63.2%, respectively, at a ratio NK/Target cells of 100.

EXAMPLE 18: IMMUNOHISTOCHEMICAL STUDIES (IHC)

[0352] Procedures of Paraffin Embedded Tumors IHC Staining: 8 to 12 μM sections of frozen tumor were and immediately fixed in pre cooled acetone −20° C. for 3 minutes. Slides were then cooled at room temperature for 30 minutes to 1 hour. After 2 washes in PBS the Endogenous peroxidase activity was blocked using Peroxidase Blocking Reagent (Dako K4007) for five minutes. Sections were washed with PBS and incubated in avidin/biotin blocking reagent (Dako X0590) just before saturation of the non specific sites in PBS-BSA 4% for 30 minutes at room temperature. Then, slides were incubated with the biotinylated h224G11 (50 to 10 μg/ml) or human biotinylated IgG1/kappa (50 to 10 μg/ml, the Binding Site) as negative control 2 hours at room temperature.

[0353] Sections were washed with PBS and incubated with Streptavidin-peroxydase complex universal (Dako K0679) for 30 to 45 minutes. 3-Amino-9-Ethylcarbazole was used for development of a red reaction product (Sigma). The slides were immersed in hematoxylin for 4 minutes to counterstain (Dako S3309).

[0354] Results are represented in FIG. 35.

[0355] h224G11 differentially stains the cell membrane of various tumor types. In this immunohistochemistry procedure, the red reaction product correlates to positive staining of the cell membrane and lack of red reaction product correlates to negative staining and no visualization of the cell membrane. The IgG control, human IgG1/kappa is an isotype matched control.