Anti-CEACAM5 antibodies and uses thereof

Abstract

The present invention discloses antibodies which bind human and Macaca fascicularis CEACAM5 proteins, as well as isolated nucleic acids, vectors and host cells comprising a sequence encoding said antibodies. The invention also discloses immunoconjugates comprising said antibodies conjugated or linked to a growth-inhibitory agent, and to pharmaceutical compositions comprising antibodies, or immunoconjugates of the invention. The antibodies or immunoconjugates of the invention are used for the treatment of cancer or for diagnostic purposes.

Claims

1. An isolated antibody or antigen binding fragment thereof wherein said antibody or antigen binding fragment thereof binds to human CEACAM5 protein and wherein said antibody or antigen binding fragment thereof comprises a CDR1-H comprising an amino acid sequence of SEQ ID NO:7, a CDR2-H comprising an amino acid sequence of SEQ ID NO:8, a CDR3-H comprising an amino acid sequence of SEQ ID NO:9, a CDR1-L comprising an amino acid sequence of SEQ ID NO:10, a CDR2-L comprising an amino acid sequence of NTR or an amino acid sequence of NTK, and a CDR3-L comprising an amino acid sequence of SEQ ID NO:12.

2. The isolated antibody or antigen binding fragment thereof according to claim 1, wherein said antibody or antigen binding fragment thereof comprises a heavy chain variable domain comprising an amino acid sequence of sequence SEQ ID NO:33 and a light chain variable domain comprising an amino acid sequence of SEQ ID NO:34.

3. The isolated antibody or antigen binding fragment thereof according to claim 1, wherein said antibody or antigen binding fragment thereof is a chimeric or a humanised antibody.

4. The isolated antibody or antigen binding fragment thereof according to claim 3, wherein said antibody or antigen binding fragment thereof comprises a heavy chain comprising an amino acid sequence of SEQ ID NO:43 and a light chain comprising an amino acid sequence of SEQ ID NO:44.

5. The isolated antibody or antigen binding fragment thereof according to claim 3, wherein said antibody or antigen binding fragment thereof comprises a) a heavy chain variable domain comprising an amino acid sequence of SEQ ID NO:51 and a light chain variable domain comprising an amino acid sequence of SEQ ID NO:17; or b) a heavy chain variable domain comprising an amino acid sequence of SEQ ID NO:51 and a light chain variable domain comprising an amino acid sequence of SEQ ID NO:55; or c) a heavy chain variable domain comprising an amino acid sequence of SEQ ID NO:5 and a light chain variable domain comprising an amino acid sequence of SEQ ID NO:23; or d) a heavy chain variable domain comprising an amino acid sequence of SEQ ID NO:5 and a light chain variable domain comprising an amino acid sequence of SEQ ID NO:29.

6. The isolated antibody or antigen binding fragment thereof according to claim 3, wherein said antibody or antigen binding fragment thereof comprises a) a heavy chain comprising an amino acid sequence of SEQ ID NO:74 and a light chain comprising an amino acid of SEQ ID NO:75; or b) a heavy chain comprising an amino acid sequence of SEQ ID NO:87 and a light chain comprising an amino acid of SEQ ID NO:88; or c) a heavy chain comprising an amino acid sequence of SEQ ID NO:89 and a light chain comprising an amino acid of SEQ ID NO:90.

7. A pharmaceutical composition comprising an antibody according to claim 1, and a pharmaceutically acceptable carrier.

8. An immunoconjugate comprising an antibody according to claim 1 conjugated or linked to at least one growth inhibitory agent, wherein the growth inhibitory agent is (N.sup.2-deacetyl-N.sup.2-(3-mercapto-1-oxopropyl)-maytansine) (DM1) or N.sup.2-deacetyl-N-.sup.2(4-methyl-4-mercapto-1-oxopentyl)-maytansine (DM4).

9. The immunoconjugate according to claim 8, wherein the linker is selected from the group consisting of N-succinimidyl pyridyldithiobutyrate (SPDB), 4-(Pyridin-2-yldisulfanyl)-2-sulfo-butyric acid (sulfo-SPDB), and succinimidyl (N-maleimidomethyl) cyclohexane-1-carboxylate (SMCC).

10. The immunoconjugate according to claim 8, wherein the growth inhibitory agent is DM4 and the linker is SPDB.

11. The immunoconjugate according to claim 10, wherein the antibody comprises a heavy chain variable domain comprising an amino acid sequence of SEQ ID NO:51 and a light chain variable domain comprising an amino acid sequence of SEQ ID NO:17.

12. The immunoconjugate according to claim 10, wherein the antibody comprises a heavy chain variable domain comprising an amino acid sequence of SEQ ID NO:51 and a light chain variable domain comprising an amino acid sequence of SEQ ID NO:55.

13. The immunoconjugate according to claim 10, wherein the antibody comprises a heavy chain variable domain comprising an amino acid sequence of SEQ ID NO:5 and a light chain variable domain comprising an amino acid sequence of SEQ ID NO:23.

14. The immunoconjugate according to claim 10, wherein the antibody comprises a heavy chain variable domain comprising an amino acid sequence of SEQ ID NO:5 and a light chain variable domain comprising an amino acid sequence of SEQ ID NO:29.

15. The immunoconjugate according to claim 10, wherein the antibody comprises a heavy chain comprising an amino acid sequence of SEQ ID NO:74 and a light chain comprising an amino acid of SEQ ID NO:75.

16. The immunoconjugate according to claim 10, wherein the antibody comprises a heavy chain comprising an amino acid sequence of SEQ ID NO:87 and a light chain comprising an amino acid of SEQ ID NO:88.

17. The immunoconjugate according to claim 10, wherein the antibody comprises a heavy chain comprising an amino acid sequence of SEQ ID NO:89 and a light chain comprising an amino acid of SEQ ID NO:90.

18. The immunoconjugate according to claim 8, wherein the immunoconjugate is characterised by a drug-to-antibody ratio (DAR) ranging from 1 to 10.

19. An immunoconjugate comprising an antibody which binds to human CEACAM5 protein, said antibody comprises a heavy chain comprising an amino acid sequence of SEQ ID NO:87 and a light chain comprising an amino acid of SEQ ID NO:88, wherein said antibody is conjugated or linked to at least one growth inhibitory agent, wherein the growth inhibitory agent is N2-deacetyl-N-2(4-methyl-4-mercapto-1-oxopentyl)-maytansine (DM4) and wherein the linker is N-succinimidyl pyridyldithiobutyrate (SPDB).

20. An immunoconjugate comprising an antibody which binds to human CEACAM5 protein, said antibody comprises a heavy chain comprising an amino acid sequence of SEQ ID NO:89 and a light chain comprising an amino acid of SEQ ID NO:90, wherein said antibody is conjugated or linked to at least one growth inhibitory agent, wherein the growth inhibitory agent is N2-deacetyl-N-2(4-methyl-4-mercapto-1-oxopentyl)-maytansine (DM4) and wherein the linker is N-succinimidyl pyridyldithiobutyrate (SPDB).

Description

DESCRIPTION OF THE FIGURES

(1) FIG. 1: Evaluation of the selectivity of the anti-CEACAM5 antibodies.

(2) FIG. 2: Domain mapping of the anti-CEACAM5 antibodies on human CEACAM5.

(3) FIG. 3: Domain mapping of the anti-CEACAM5 antibodies on cynomolgus CEACAM5.

(4) FIG. 4: Evaluation of the anti-tumor activity of chMAb4-SPDB-DM4, chMAb1-SPDB-DM4, chMAb5-SPDB-DM4, and chMAb2-SPDB-DM4 conjugates against primary human colon adenocarcinoma CR-IGR-034P in SCID female mice.

(5) FIG. 5: Evaluation of the anti-tumor activity of huMAb2-3-SPDB-DM4, huMAb2-4-SPDB-DM4 and chMAb2-SPDB-DM4 conjugates against primary human colon adenocarcinoma CR-IGR-034P in SCID female mice.

(6) FIG. 6: Evaluation of the anti-tumor activity of huMAb2-3-SPDB-DM4 conjugate against primary human stomach adenocarcinoma STO-IND-006 in SCID female mice.

(7) FIG. 7: Sequence alignments of the VH and VL regions of the following antibodies MAb1 VH: SEQ ID NO: 31, VL: SEQ ID NO: 32, MAb2 VH: SEQ ID NO: 33, VL: SEQ ID NO: 34, MAb3 VH: SEQ ID NO: 35, VL: SEQ ID NO: 36, MAb4 VH: SEQ ID NO: 37, VL: SEQ ID NO: 38, and MAb5 VH: SEQ ID NO: 39, VL: SEQ ID NO: 40).

(8) FIG. 8: HRMS analysis of chMAb1-SPDB-DM4 conjugate.

(9) FIG. 9: HRMS analysis of chMAb2-SPDB-DM4 conjugate.

(10) FIG. 10: HRMS analysis of chMAb4-SPDB-DM4 conjugate.

(11) FIG. 11: HRMS analysis of chMAb5-SPDB-DM4 conjugate.

(12) FIG. 12: HRMS analysis of huMAb2-2-SPDB-DM4 conjugate.

(13) FIG. 13: HRMS analysis of huMAb2-1-SPDB-DM4 conjugate.

(14) FIG. 14: HRMS analysis of huMAb2-3-SPDB-DM4 conjugate.

(15) FIG. 15: HRMS analysis of huMAb2-4-SPDB-DM4 conjugate.

(16) FIG. 16: Binding activity of humanized variants of MAb2 to human and monkey CEACAM5 extracellular domain.

(17) FIG. 17: Stability of binding of humanized variants of MAb2 to human and monkey CEACAM5 extracellular domain.

(18) FIG. 18: Evaluation of the anti-tumor activity of huMAb2-3-SPDB-DM4 conjugate against primary human lung adenocarcinoma LUN-NIC-0014 in SCID female mice.

(19) FIG. 19: Evaluation of the anti-tumor activity of huMAb2-3-SPDB-DM4 and huMAb2-3-sulfo-SPDB-DM4 conjugates against primary human colon adenocarcinoma CR-IGR-034P in CD1 nude female mice.

(20) FIG. 20: HRMS analysis of huMAb2-3-sulfoSPDB-DM4.

(21) FIG. 21: HRMS analysis of huMAb2-3-SMCC-DM1.

(22) FIG. 22: Heavy Chain variable domain alignment of MAb2 SEQ ID NO: 33, MAb4 SEQ ID NO: 37, MAb5 SEQ ID NO: 39, humanized VH1a SEQ ID NO: 5, humanized VH1 SEQ ID NO: 51 and humanized VHg2 SEQ ID NO: 74.

(23) FIG. 23: Light Chain variable domain alignment of MAb2 SEQ ID NO: 34, MAb4 SEQ ID NO: 38, MAb5 SEQ ID NO: 40, humanized VL1 SEQ ID NO: 17, humanized VL1a SEQ ID NO: 23, humanized VL1c SEQ ID NO: 29, humanized VL1d SEQ ID NO: 55, and humanized VLg5 SEQ ID NO: 75.

EXAMPLES

(24) The present invention is further illustrated by the following examples which should not be construed as further limiting.

(25) The contents of the Sequence Listing, figures and all references, patents and published patent applications cited throughout this application are expressly incorporated herein by reference in their entirety.

Example 1

Preparation of Recombinant Extracellular Domains of CEACAM Proteins

(26) In this example, the extracellular protein domains (ECD) of CEACAM from human (h) or cynomolgus monkey (c) origin have been prepared by transient expression in human embryonic kidney HEK293 cells with plasmids allowing expression of the respective cDNA as outlined on Table 1.

(27) Each expression plasmid was complexed with 293Fectin (Life Technologies) and added to suspension-cultivated 293-F cells (derived from HEK293 cells). Eight days post-transfection, the culture supernatants were collected and the corresponding soluble protein was purified by IMAC (GE Healthcare) to generate a protein batch (see Table 1).

(28) TABLE-US-00003 TABLE 1 Description of the recombinant extracellular domains of CEACAM proteins cDNA Protein sequence Sequence name Protein desription origin identifier hCEACAM1 human CEACAM1 ECD NP_001703.2 SEQ ID (35-428) NO: 56 cCEACAM1 M. fascicularis CEACAM1 cloned SEQ ID ECD (35-428) internally NO: 57 hCEACAM5 human CEACAM5 ECD AAA51967.1 SEQ ID (35-685) NO: 58 cCEACAM5 M. fascicularis CEACAM5 cloned SEQ ID ECD (35-688) internally NO: 59 hCEACAM6 human CEACAM6 ECD NP_002474.3 SEQ ID (35-327) NO: 60 cCEACAM6 M. fascicularis CEACAM6 cloned SEQ ID ECD (35-327) internally NO: 61 hCEACAM8 human CEACAM8 ECD NP_001807.2 SEQ ID (35-332) NO: 62 cCEACAM8 M. fascicularis CEACAM8 cloned SEQ ID ECD (35-332) internally NO: 63 hCEACAM7 human CEACAM7 ECD NP_008821.1 SEQ ID (36-248) NO: 64 hCEACAM5 human CEACAM5 N-A1-B1 AAA51967.1 SEQ ID NA1B1 domain (35-320) NO: 65 hCEACAM5 human CEACAM5 A2-B2 AAA51967.1 SEQ ID A2B2 domain (321-498) NO: 66 hCEACAM5 human CEACAM5 A3-B3 AAA51967.1 SEQ ID A3B3 domain (499-685) NO: 67 cCEACAM5 M. fascicularis CEACAM5 cloned SEQ ID NA1B1 N-A1-B1 domain (35-320) internally NO: 68 cCEACAM5 M. fascicularis CEACAM5 cloned SEQ ID A2B2 A2-B2 domain (321-498) internally NO: 69 cCEACAM5 M. fascicularis CEACAM5 cloned SEQ ID A3B3 A3-B3 domain (499-688) internally NO: 70

Example 2

Generation of Monoclonal Mouse Anti-CEACAM5 Antibodies

(29) In this example, monoclonal antibodies have been generated following mice immunization according to a protocol that led to the generation of antiCEACAM5 mAb.

Example 2.1

Immunization & Hybridoma generation

(30) Immunizations, fusion and screening were performed using P3X63-Ag8.653 myeloma cells with either the extracellular domain of human CEACAM5, the extracellular domain of cynomolgus CEACAM5 or with human tumoral UMC11 cells as described in Wennerberg A. E et al., 1993. Am. J. Pathol., 143(4), 1050-1054 and Kilpatrick et al. 1997. Hybridoma 16: 381389.

(31) Using the RIMMS method as described by Kilpatrick et al. (1997. Hybridoma 16: 381389), 6-8 weeks old female BALB/c mice (S082342; Charles River Labs, Bar Harbor, Me.) each received four rounds of immunization over a course of 14 days at intervals of 3-4 days. Antigens emulsified in Titermax's adjuvant (TierMax Gold Adjuvant; Sigma #T2684) was administered subcutaneously to six sites proximal to draining lymph nodes, along the back of the mice and to six juxtaposed sites along abdomen. Four days after the last injection, mice were sacrified. Bilateral popliteal, superficial inguinal, axillary and branchial lymph nodes were isolated aseptically and washed with fresh RPMI medium.

(32) Using the classical method as described by Wennerberg A. E et al. (1993. Am. J. Pathol., 143(4), 1050-1054), 6-8 weeks old female BALB/c mice (S082342; Charles River Labs, Bar Harbor, Me.) each received three rounds of immunization over a course of 41 days. Antigens were administered intraperitonealy to ventral site of mice. Three days after the last injection, mice were sacrified and spleens were isolated aseptically and washed with fresh RPMI medium.

(33) Lymphocytes were released from the lymph nodes or from the spleens and single-cell suspension was washed twice with RPMI medium before being fused with P3X63-AG8.653 myeloma cells using polyethylene glycol. After fusion, the cell mixture was incubated in an incubator at 37 C. for 16-24 hours. The resulting cells preparation was transferred into selective semi-solid medium and aseptically plated out into 100 mm Petri plates and incubated at 37 C. Ten days after initiation of selection, the plates were examined for hybridoma growth, and visible colonies were picked-up and placed into 96-well plates containing 200 L of growth medium. The 96-well plates were kept in an incubator at 37 C. for 2 to 4 days.

Example 2.2

Screening and In Vitro Characterization of Murine Anti-CEACAM5 Antibodies

(34) Primary screening for anti-CEACAM5 IgG production was performed by Enzyme-linked immunosorbent assay (ELISA) using human CEACAM5 protein (prepared as described in Example 1) as capturing antigen and by FACS using several human tumoral cells (H460, MKN45, SW1463, SKMEL28 and UMC11). For ELISA assay, plates were coated with human CEACAM5 protein at 0.25 g/well in PBS and 100 L/well of anti-CEACAM5 antibodies were added to the plate. The plate was incubated at 37 C. for 1 h and washed five times with PBS containing 0.05% Tween-20 (PBS-T). Then, 100 L of a 1:50,000 dilution of rabbit anti-mouse IgG conjugated with horseradish peroxidase (Sigma; #A9044) was added to each well. Following incubation at 37 C. for 1 h in darkness, plates were washed with PBS-T five times. Antibody binding was visualized by adding TMB-H2O2 buffer and read at a wavelength of 450 nm. For FACS assay, human tumoral cells were coated at 40,000 cells/well on 96-well High Bind plate (MSD L15XB-3) and 100 L/well of anti-CEACAM5 antibodies were added for 45 min at 4 C. and washed three times with PBS 1% BSA. 100 L/well of goat anti-mouse IgG conjugated with Alexa647 (Invitrogen; # A2135) was added for 45 min at 4 C. and washed three times with PBS 1% BSA. Antibody binding was evaluated after centrifugation and resuspension of cells by adding 200 l/well PBS 1% BSA and read using Guava easyCyte 8HT Flow Cytometry System.

(35) For evaluating specificity to CEACAM5 of anti-CEACAM5 antibodies, 96-well plates were coated with recombinant human CEACAM1, CEACAM6, CEACAM7 and CEACAM8 proteins (prepared as described in Example 1) using the same coating conditions described previously. Anti-CEACAM5 antibodies were added to the plates and detected by using rabbit anti-mouse IgG conjugated with horseradish peroxidase (Sigma; #A9044). Antibody binding was visualized by adding TMB-H2O2 buffer and read at a wavelength of 450 nm. The results presented on FIG. 1 show that the anti-CEACAM5 antibodies are selective for human CEACAM5 v. human CEACAM1, CEACAM6, CEACAM7 and CEACAM8.

Example 2.3

mAb Binding Characterization

(36) The apparent affinity of anti-CEACAM5 antibodies to hCEACAM5 expressed on the surface of human MKN45 (DSMZ, ACC 409) tumoral cells were determined by Guava easyCyte 8HT Flow Cytometry System. MKN45 tumoral cells were coated at 40,000 cells/well on 96-well High Bind plate (MSD L15XB-3) and 100 L/well of anti-CEACAM5 antibodies were added in 2-fold serial dilutions starting at 20 g/ml up to 12 dilutions in assay diluent for 45 min at 4 C. and washed three times with PBS 1% BSA. 100 L/well of goat anti-mouse IgG conjugated with Alexa647 (Invitrogen; # A2135) was added for 45 min at 4 C. and washed three times with PBS 1% BSA. The antibody binding was evaluated after centrifugation and resuspension of cells by adding 200 l/well PBS 1% BSA and read using Guava easyCyte 8HT Flow Cytometry System. Apparent KD and EC50 values were estimated using BIOST@T-BINDING and BIOST@T-SPEED softwares, respectively.

(37) TABLE-US-00004 TABLE 2 EC50 values obtained on MKN45 cells Antibodies MAb1 MAb2 MAb3 MAb4 MAb5 EC50 values 16 nM 3.4 nM 6.2 nM 4.9 nM 0.73 nM

(38) Domain mapping of anti-CEACAM5 antibodies to human CEACAM5 and cynomolgus CEACAM5 proteins was determined by ELISA. 96-well plates were coated with recombinant human A1 (143-237), A1-B1 (143-320), A2-B2 (321-498) and A3-B3 (499-685) domains of CEACAM5 protein (prepared as described in Example 1) and with recombinant cynomolgus N-A1-B1 (1-320), A1-B1 (143-320), A2-B2 (321-498) and A3-B3 (499-688) domains of CEACAM5 protein (prepared as described in Example 1) using the same coating conditions described previously. Purified antibodies were added to the plates and detected by using rabbit anti-mouse IgG conjugated with horseradish peroxidase (Sigma; #A9044). Antibody binding was visualized by adding TMB-H2O2 buffer and read at a wavelength of 450 nm. The results are presented on FIGS. 2 and 3 and show that the anti-CEACAM5 antibodies bind to the A3-B3 domain of human and cynomolgus CEACAM5 proteins.

(39) Isotypes of individual mAbs were determined using a mouse IgG isotyping kit according to the manufacturer's instructions (SEROTEC ref. MMT1). The five CEACAM5-specific mAbs were of the IgG1, k isotype.

Example 3

Characterization of Murine Anti-CEACAM5 Antibodies

Example 3.1

In Vitro Characterization of Murine Anti-CEACAM5 Antibodies

(40) Mouse hybridoma expressing CEACAM5-specific Abs were produced into T500 flask and conditioned media collected after 7 days of growth. CEACAM5-specific Abs were purified by passing the conditioned media through a Protein-G column, washed and eluted with Glycine/HCl 100 mM pH 2.7 buffer. The eluate was dialyzed against PBS before sterile filtration and stored at 4 C.

(41) All CEACAM5-specific mAbs were assessed for their ability to bind human and primate CEACAM5 protein by ELISA. Plates were coated with human or primate CEACAM5 protein, anti-hCEACAM5 mAbs were added to the plate and detected with rabbit anti-mouse IgG conjugated with horseradish peroxidase (Sigma; #A9044). The antibody binding was visualized by adding TMB-H2O2 buffer and read at a wavelength of 450 nm.

(42) TABLE-US-00005 TABLE 3 EC50 values corresponding to binding ability of CEACAM5-specific mAbs to primate CEACAM5 proteins Antibodies MAb1 MAb2 MAb3 MAb4 MAb5 EC50 (nM) 0.53 0.14 0.36 0.08 0.40 hCEACAM5 EC50 (nM) 1.18 0.07 3.72 0.05 0.45 cCEACAM5 Ratio c/h 2.2 0.5 10 0.6 1.1

Example 3.2

Apparent Affinity and Antibody Binding Capacity of Anti-CEACAM5 Antibodies to Advanced Human Primary Colon Tumor Cells CR-IGR-034P by Flow Cytometry

(43) Advanced human primary colon tumor CR-IGR-034P (Julien et al., Clin Cancer Res Oct. 1, 2012 18:5314-5328) was obtained from Patient-derived xenograft in mice. Tumor CR-IGR-034P was enzymatically dissociated using collagenase Type IV (Invitrogen; #17104-019) and deoxyribonuclease I (Invitrogen; #18047-019) for 1 h at 4 C. Cell viability was estimated by Viacount application using Guava easyCyte 8 HT Flow Cytometry System. For apparent affinity estimation, CR-IGR-034P tumoral cells were coated at 40,000 cells/well on 96-well High Bind plate (MSD L15XB-3) and 100 L/well of anti-CEACAM5 antibodies were added in 2-fold serial dilutions starting at 20 g/ml up to 12 dilutions in assay diluent for 45 min at 4 C. and washed three times with PBS 1% BSA. 100 L/well of goat anti-mouse IgG conjugated with Alexa647 (Invitrogen; # A2135) or goat anti-human IgG conjugated with Alexa488 (Invitrogen; # A11013) was added for 45 min at 4 C. and washed three times with PBS 1% BSA. The antibody binding was evaluated after centrifugation and resuspension of cells by adding 200 l/well PBS 1% BSA and read using Guava easyCyte 8HT Flow Cytometry System. Apparent KD and EC50 values were estimated using BIOST@T-BINDING and BIOST@T-SPEED softwares, respectively.

(44) Antibody binding capacity of anti-CEACAM5 antibodies was determined using Mouse IgG Calibrator kit (Biocytex #7208) or Human IgG Calibrator Kit (Biocytex #CP010) according to the manufacturer's instructions.

(45) TABLE-US-00006 TABLE 4 KD and EC50 values obtained on advanced human primary colon tumor cells CR-IGR-034P Antibodies MAb1 MAb2 MAb3 MAb4 MAb5 KD value 1.92 nM 0.38 nM 1.01 nM 0.16 nM 0.5 nM EC50 value 1 nM 0.53 nM 2.8 nM 0.2 nM 1.4 nM

Example 3.3

Internalization Activity of Murine CEACAM5-Specific Antibodies

(46) To evaluate the internalization of the anti-CEACAM5 antibodies MAb1, MAb2, MAb3, MAb4 and MAb5, viable MKN45 cells were incubated for 24 h at 37 C./5% CO2 (or 4 C. on ice for negative control) with 10 g/ml of AlexaFluor488-pre-labeled anti-CEACAM5 antibodies. Then, one part of the wells were rinced with culture medium and the extracellular AF-labeled antibodies bound to the cells were quenched by incubating the cells with anti-AlexaFluor 488 antibody (50 g/mL) on ice for 30 min (intracellular fluorescence level). The other part of the wells was only incubated with culture medium in the same time condition (total fluorescence level).

(47) The cells were then detached and washed, and collected in culture medium before flow cytometry analysis using a MACSQUANT Vyb analyzer. The cellular-associated fluorescence of 110.sup.4 cells was measured, and the mean fluorescent intensity of gated viable cells was quantified. The internalization ratio (%) is defined by dividing the quenched cell-associated fluorescence by the total cell-associated fluorescence multiplicated by 100. Data are expressed as the meanstandard deviation (SD)

(48) TABLE-US-00007 TABLE 5 Anti-CEACAM5 murine antibody internalization at 24 hrs in MKN45 cell line Internalization 24 hrs, 37C/5% CO2 Antibody % StD MAb1 49.9 5.1 MAb2 45.0 5.5 MAb3 51.1 3.5 MAb4 42.5 6.7 MAb5 51.7 3.1

(49) The five CEACAM5-specific antibodies undergo internalization after binding of CEACAM5 expressed at the cell surface membrane, supporting their use in the field of antibody immunoconjugates to specifically address cytotoxic to cancer cells. The anti-CEACAM5 antibodies MAb1, MAb2, MAb3, MAb4 and MAb5 showed internalization in MKN45 human cancer cell line of 49.9%, 45%, 51.1%, 42.5%, 51.7%, respectively, after 24 hours of incubation.

Example 3.4

Cytotoxic Activity of the Corresponding Murine ADCs on MKN45 Cell Line

(50) The murine antibodies were conjugated in order to define their in vitro cytotoxic activity. In a 15 ml tube, at room temperature (23 C.), mAb, Buffer A/HEPES (4%), DMA (dimethylacetamide, 20% v/v), then 6 equivalent of SPDB linker are successively introduced under magnetic stirring. After one night at room temperature, DM4 (maytansinoid, 9.6 equivalent) in 15 mM DMA solution is added, and reacted 5 hours. Crude conjugation mixture is purified on Superdex 200 g 16/60 or G25 26/10 columns (PBS-Na pH7.4/5% NMP), concentrated on Amicon 15 @ 5000 g and filtered on Millex 0.22 m.

(51) The effect of the anti-CEACAM5 maytansinoid conjugates were then tested on tumor cell viability using the Cell Titer-Glo kit (Promega). To do so, MKN45 human gastric cancer cells were plated in 96-well plates and allowed to adhere during 4 hours in 37 C./5% CO2 atmosphere. Different concentrations of anti-CEACAM5 conjugates were added to the seeded cells. The cells were then incubated for 96 hours in the same atmosphere. Cell Titer-Glo reagent was then added to the wells for 10 min at room temperature and the luminescent signal was measured using an EnVision plate counter (Perkin-Elmer).

(52) TABLE-US-00008 TABLE 6 Cytotoxic activities of the CEACAM5-specific murine ADCs on CEACAM5 + MKN45 cell line Antibody Drug Cytotoxic activity Conjugate IC.sub.50 (nM) MAb1-SPDB-DM4 0.89 0.23 MAb2-SPDB-DM4 0.14 0.01 MAb3-SPDB-DM4 0.53 0.15 MAb4-SPDB-DM4 0.96 0.02 MAb5-SPDB-DM4 0.24 0.04

(53) The anti-CEACAM5 antibodies conjugated to maytansinoid (DM4) MAb1-SPDB-DM4, MAb2-SPDB-DM4, MAb3-SPDB-DM4, MAb4-SPDB-DM4 and MAb5-SPDB-DM4 showed in vitro cytotoxic activities with an IC50 of 0.89, 0.14, 0.53, 0.96 and 0.24 nM, respectively.

Example 4

Sequence Determination of Heavy and Light Chains of the Anti-CEACAM5 mAbs

(54) The sequences of the variable domains of the mAb were retrieved from the hybridoma and cloned into an expression vector to ensure that the cloned mAbs had the same characteristics as the initial murine mAbs.

(55) The derived amino acid sequences provided information in agreement with the data obtained on purified mAbs derived from the hybridoma by N-terminal sequencing and mass spectrometry (LC/MS) of the heavy and light chains (LC, HC) (see Table 7).

(56) TABLE-US-00009 TABLE 7 Mass spectrometry analysis of anti-CEACAM5 mAbs from hybridoma Mass (Da) by LC/MS in silico value Clone ID Chain from batch retrieved sequence MAb1 LC 23837 23836 HC (G0F) 50328 50330 MAb2i* LC 23467 23467 HC (G0F) 50288 50286 MAb3 LC 23907 23907 HC (G0F) 50372 50373 MAb4 LC 23731 23731 HC (G0F) 50370 50370 MAb5 LC 23659 23659 HC (G0F) 50329 50330 *:MAb2i is the antibody produced by one of the cloned hybridoma and from which the so-called MAb2 has been derived by introducing canonical residues in the framework regions of VL and VH, as explained in example 5.

Example 5

Antibody Drug Conjugate (ADC) (Chimer)

Example 5.1

Naked Chimer mAb

(57) The nucleic acid sequences of the variable domains VH, VL were cloned into expression vectors in fusion with the human IgG1 or the human Ckappa constant domain coding sequences respectively to then generate batches of chimer mAbs by transient expression in HEK293 as described in Example 1. Affinities to human and cynomolgus CEACAM5 remained similar for murine and chimer mAbs. On Table 8, affinities are illustrated by the EC50 obtained by ELISA with human or cynomolgus CEACAM5.

(58) TABLE-US-00010 TABLE 8 EC50 obtained with CEACAM5 for murine hybridoma and corresponding chimer mAbs EC50 obtained for murine hybridoma mAbs EC50 obtained for chimeric mAbs clone ID hCEACAM5 cCEACAM5 clone ID hCEACAM5 cCEACAM5 MAb//1 0.53 1.18 chMAb1 0.51 1.57 MAb2i 0.14 0.07 chMAb2 (lot 1) 0.16 0.13 chMAb2 (lot 2) 0.14 0.17 chMab2.sub.K52R 0.11 0.15 MAb3 0.36 3.72 chMAb3 Not done Not done MAb4 0.08 0.05 chMAb4 0.14 0.12 MAb5 0.4 0.45 chMAb5 0.18 0.13

(59) The sequences for the CDR regions were deduced from the protein sequence using the IMGT nomenclature. They correspond to SEQ ID NO: 1-4, 6, 7-10, 12, 13-16, 18, 19-22, 24, 25-28, 30.

(60) Of note, compared to the antibody produced by the cloned hybridoma (MAb2i), canonical residues have been introduced into clone MAb2 at positions 41G, 42K, and 45Q on VL, and at positions 5Q and 7S on VH.

(61) In addition, lysine at position 52 on the VL of clone MAb2 CEA-4 is located in the CDR2, has been replaced by arginine in clone Mab2.sub.K52R. A batch was generated in the same conditions as that corresponding to clone MAb2 and led to similar affinity to human and cynomolgus CEACAM5 extracellular domain as shown on Table 7. It highlighted that this point mutation in the CDR can be made without any impact on binding.

(62) The LC and HC sequences of the chimer mAb for clone MAb2 and clone Mab2.sub.K52R correspond to SEQ ID NO:43, 44, 54.

(63) chMAb2 was constructed as described in example 4. It is a chimer mAb derived from clone MAb2 with a human IgG1, Ck isotype. The sequences correspond to SEQ ID NO:43, and 44. A batch was prepared at 300 mg scale by transient expression in HEK293 followed by protein An affinity chromatography purification, see Table 7 for the binding data. It was the naked mAb used for the production of the ADC.

Example 5.2

Production and Characterisation of ADC

(64) In this example, immunoconjugates were prepared from naked chimer mAb. in vivo efficacy were then assessed.

(65) DAR Calculation:

(66) A conjugate comprises generally from 1 to 10 molecule(s) of the maytansinoid attached covalently to the antibody (so called, drug-to-antibody ratio or DAR). This number can vary with the nature of the antibody and of the maytansinoid used along with the experimental conditions used for the conjugation (like the ratio maytansinoid/antibody, the reaction time, the nature of the solvent and of the cosolvent if any). Thus the contact between the antibody and the maytansinoid leads to a mixture comprising several conjugates differing from one another by different drug-to-antibody ratios; optionally the naked antibody; optionally aggregates. The DAR that is determined is thus a mean value.

(67) The method used herein to determine the DAR consists in measuring spectrophotometrically the ratio of the absorbance at 252 nm and 280 nm of a solution of the substantially purified conjugate. In particular, said DAR can be determined spectrophotometrically using the measured extinction coefficients at respectively 280 and 252 nm for the antibody and for the maytansinoid (.sub.D280=5,180 M.sup.1 cm.sup.1 and .sub.D252=26,159 M.sup.1 cm.sup.1). The method of calculation is derived from Antony S. Dimitrov (ed), LLC, 2009, Therapeutic Antibodies and Protocols, vol 525, 445, Springer Science and is described in more details below:

(68) The absorbances for the conjugate at 252 nm (A252) and at 280 nm (A280) are measured either on the monomeric peak of the size exclusion chromatography (SEC) analysis (allowing to calculate the DAR(SEC) parameter) or using a classic spectrophotometer apparatus (allowing to calculate the DAR(UV) parameter). The absorbances can be expressed as follows:
A.sub.252=(c.sub.D.sub.D252)+(c.sub.A.sub.A252)
A.sub.280=(c.sub.D.sub.D280)+(c.sub.A.sub.A280) wherein: c.sub.D and c.sub.A are respectively the concentrations in the solution of the maytansinoid and of the antibody .sub.D252 and .sub.D280 are respectively the molar extinction coefficients of the maytansinoid at 252 nm and 280 nm .sub.A252 and .sub.A280 are respectively the molar extinction coefficients of the antibody at 252 nm and 280 nm.

(69) Resolution of these two equations with two unknowns leads to the following equations:
c.sub.D=[(.sub.A280A.sub.252)(.sub.A252A.sub.280)]/[(.sub.D252.sub.A280)(.sub.A252.sub.D280)]
c.sub.A=[A.sub.280(c.sub.D.sub.D280)]/.sub.A280

(70) The average DAR is then calculated from the ratio of the drug concentration to that of the antibody: DAR=c.sub.D/c.sub.A

(71) Deglycosylation and High Resolution Mass Spectrometry of Conjugates (HRMS)

(72) Deglycosylation is a technique of enzymatic digestion by means of glycosidase. The deglycosylation is made from 500 l of conjugated+100 l of Tris buffer HCl 50 mM+10 l of glycanase-F enzyme (100 units of freeze-dried enzyme/100 l of water). The medium is vortexed and maintained one night at 37 C. The deglycosylated sample is then ready to be analyzed in HRMS. Mass spectra were obtained on a Waters Q-Tof-2 system in electrospray positive mode (ES+). Chromatographic conditions are the following: column: 4 m BioSuite 250 URH SEC 4.6300 mm (Waters); solvents: A: ammonium formate 25 mM+1% formic acid: B: CH3CN; column temperature: 30 C.; flow rate 0.4 ml/min; isocratic elution 70% A+30% B (15 min).

(73) Analytical Size Exclusion Chromatography (SEC) Column: TSKgel G3000 SWXL 5 m column, 7.8 mm30 cm, TOSOH BIOSCIENCE, LLC Part #08541+guard column TSK-GEL SWXL 7 M, 40 mm6 mm, TOSOH BIOSCIENCE, LLC Part #08543 Mobile Phase: KCl (0.2M), KH2PO4 (0.052 M), K2HPO4 (0.107 M), iPrOH (20% in volume) Analysis Conditions: isocratic elution at 0.5 ml/min for 30 min Analysis performed on a Lachrom Elite HPLC system (Merck) using a L2455 DAD spectrophotometer detector.

(74) Buffers Contents Buffer A (pH 6.5): NaCl (50 mM), Potassium Phosphate buffer (50 mM), EDTA (2 mM) Buffer HGS (pH 5.5): histidine (10 mM), glycine (130 mM), sucrose 5% (w/v), HCl (8 mM)

(75) Abbreviations Used

(76) CV: Column Volume; DAR: Drug Antibody Ratio; DMA: dimethylacetamide; HEPES: 4-(2-hydroxyethyl)-1-piperazine-ethanesulfonic acid; HRMS: High Resolution Mass Spectroscopy; NHS: N-hydroxysuccinimide; Nitro-SPDB: butanoic acid, 4-[(5-nitro-2-pyridinyl)dithio]-, 2,5-dioxo-1-pyrrolidinyl ester (could be prepared as described in WO2004016801 patent); NMP: N-methylpyrrolidinone; RT: room temperature; SEC: Size Exclusion Chromatography ADC (chimers):

(77) chMAb1-SPDB-DM4

(78) Analytical Data: MW(Ab)=148438 g/mol; MW(DM4)=780.38 g/mol .sub.280nm(Ab)=213320; .sub.252nm(Ab)=73473 .sub.280nm(DM4)=5180 et .sub.252nm(DM4)=26159

(79) Under stirring, at RT, 3.59 ml of chMAb1 (C=5.72 mg/ml in PBS pH=7.4 buffer) are introduced in a vessel, followed by 0.312 ml of DMA and 0.046 ml of nitro-SPDB linker solution (5.0 Eq-15 mM solution in DMA). Solution is vortexed for 30 sec and then slowly stirred at RT for 3 hours. Under magnetic stirring, 3.8 ml of PBS pH7.5 buffer, 0.389 ml of DMA and 0.074 ml of DM4 solution (15 mM solution in DMA) were sucessively added. After 2.5 hours at RT, crude reaction mixture is purified on HiLoad 26/60 desalting column (Superdex 200 pg; GE Healthcare), pre-conditioned with 1CV of NaOH 1M, 2 CV of water and 2 CV of PBS pH7.4 buffer containing 5% of NMP in volume. Conjugate is eluted with PBS pH7.4 buffer containg 5% of NMP, and monomeric conjugate fractions are pooled, concentrated on Amicon Ultra-15 (Ultracel 10 k, Millipore) and filtered on 0.22 m filter.

(80) 7.6 ml of chMAb1-SPDB-DM4 conjugate (c=2.19 mg/ml) was thus obtained as a colorless clear solution. The conjugate is then analyzed for final drug load and monomeric purity: DAR (UV)=3.38; DAR (SEC)=3.34; RT=17.54 min; monomeric purity=99.8%.

(81) The result of HRMS analysis is shown on FIG. 8.

(82) chMAb2-SPDB-DM4

(83) Analytical Data: MW(Ab)=147900 g/mol; MW(DM4)=780.38 g/mol .sub.280nm(Ab)=201400; .sub.252nm(Ab)=70889 .sub.280nm(DM4)=5180 et .sub.252nm(DM4)=26159

(84) Under stirring, at RT, 3.8 ml of chMAb2 (C=5.08 mg/ml in PBS pH=7.4 buffer) are introduced in a vessel, followed by 0.337 ml of DMA and 0.0433 ml of nitro-SPDB linker solution (5.0 Eq-15 mM solution in DMA). Solution is vortexed for 30 sec and then slowly stirred at RT for 3 hours. Under magnetic stirring, 3.12 ml of PBS pH7.5 buffer, 0.319 ml of DMA and 0.069 ml of DM4 solution (15 mM solution in DMA) were sucessively added. After 2 hours at RT, crude reaction mixture is filtered on 0.45 m filter and purified on HiLoad 26/60 desalting column (Superdex 200 pg; GE Healthcare), pre-conditioned with 1CV of NaOH 1M, 2 CV of water and 2 CV of PBS pH7.4 buffer containg 5% of NMP in volume. Conjugate is eluted with PBS pH7.4 buffer containg 5% of NMP, and monomeric conjugate fractions are pooled, concentrated on Amicon Ultra-15 (Ultracel 10 k, Millipore) and filtered on 0.22 m filter.

(85) 7.5 ml of chMAb2-SPDB-DM4 conjugate (c=1.8 mg/ml) was thus obtained as a colorless clear solution. The conjugate is then analyzed for final drug load and monomeric purity: DAR (UV)=4.10; DAR (SEC)=4.05; RT=17.52 min; monomeric purity=99.9%.

(86) The result of HRMS analysis is shown on FIG. 9.

(87) chMAb4-SPDB-DM4

(88) Analytical Data: MW(Ab)=148124 g/mol; MW(DM4)=780.38 g/mol .sub.280nm(Ab)=204380; .sub.280nm252nm(Ab)=73142 .sub.280nm(DM4)=5180 et .sub.280nm252nm(DM4)=26159

(89) Under stirring, at RT, 3.63 ml of chMAb4 (C=5.69 mg/ml in PBS pH=7.4 buffer) are introduced in a vessel, followed by 0.316 ml of DMA and 0.0465 ml of nitro-SPDB linker solution (5.0 Eq-15 mM solution in DMA). Solution is vortexed for 30 sec and then slowly stirred at RT for 3 hours. Under magnetic stirring, 3.8 ml of PBS pH7.5 buffer, 0.389 ml of DMA and 0.074 ml of DM4 solution (15 mM solution in DMA) were sucessively added. After 2 hours at RT, crude reaction mixture is purified on HiLoad 26/60 desalting column (Superdex 200 pg; GE Healthcare), pre-conditioned with 1CV of NaOH 1M, 2 CV of water and 2 CV of PBS pH7.4 buffer containg 5% of NMP in volume. Conjugate is eluted with PBS pH7.4 buffer containg 5% of NMP, and monomeric conjugate fractions are pooled, concentrated on Amicon Ultra-15 (Ultracel 10 k, Millipore) and filtered on 0.22 m filter.

(90) 6.5 ml of chMAb4-SPDB-DM4 conjugate (c=2.20 mg/ml) was thus obtained as a colorless clear solution. The conjugate is then analyzed for final drug load and monomeric purity: DAR (UV)=3.87; DAR (SEC)=3.85; RT=17.52 min; monomeric purity=99.8%.

(91) The result of HRMS analysis is shown on FIG. 10.

(92) chMAb5-SPDB-DM4

(93) Analytical Data: MW(Ab)=148040 g/mol; MW(DM4)=780.38 g/mol .sub.280nm(Ab)=207360; .sub.280nm252nm(Ab)=72288 .sub.280nm(DM4)=5180 et .sub.280nm252nm(DM4)=26159

(94) Under stirring, at RT, 3.15 ml of chMAb5 (C=6.38 mg/ml in PBS pH=7.4 buffer) are introduced in a vessel, followed by 0.269 ml of DMA and 0.0453 ml of nitro-SPDB linker solution (5.0 Eq-15 mM solution in DMA). Solution is vortexed for 30 sec and then slowly stirred at RT for 3 hours. Under magnetic stirring, 4.1 ml of PBS pH7.5 buffer, 0.317 ml of DMA and 0.072 ml of DM4 solution (15 mM solution in DMA) were sucessively added. After 2 hours at RT, crude reaction mixture is filtered on 0.45 m filter and purified on HiLoad 26/60 desalting column (Superdex 200 pg; GE Healthcare), pre-conditioned with 1CV of NaOH 1M, 2 CV of water and 2 CV of PBS pH7.4 buffer containg 5% of NMP in volume. Conjugate is eluted with PBS pH7.4 buffer containg 5% of NMP, and monomeric conjugate fractions are pooled, concentrated on Amicon Ultra-15 (Ultracel 10 k, Millipore) and filtered on 0.22 m filter.

(95) 7.5 ml of AntiCEACAM5_hyb_1917 CEA4_VH5Q7S_VL41G42K45Q_IgG1-SPDB-DM4 conjugate (c=3.4 mg/ml) was thus obtained as a colorless clear solution. The conjugate is then analyzed for final drug load and monomeric purity: DAR (UV)=3.4; DAR (SEC)=3.4; RT=17.49 min; monomeric purity=99.8%.

(96) The result of HRMS analysis is shown on FIG. 11.

Example 5.3

In Vivo Efficacy

(97) Four chimeric conjugates (chMAb4-SPDB-DM4, chMAb1-SPDB-DM4, chMAb5-SPDB-DM4 and chMAb2-SPDB-DM4) were evaluated at 2 doses against measurable primary colon CR-IGR-034P tumors implanted s.c. in female SCID mice. Control groups were left untreated. The doses conjugates were given in mg/kg. They were administered at 5 and 2.5 by an intravenous (IV) bolus injection, on day 14 after tumor implantation.

(98) For the evaluation of anti-tumor activity of conjugates, animals were weighed daily and tumors were measured 2 times weekly by caliper. A dosage producing a 20% weight loss at nadir (mean of group) or 10% or more drug deaths, was considered an excessively toxic dosage. Animal body weights included the tumor weights. Tumor volume were calculated using the formula mass (mm.sup.3)=[length (mm)width (mm)2]/2. The primary efficacy end points are T/C, percent median regression, partial and complete regressions (PR and CR).

(99) Changes in tumor volume for each treated (T) and control (C) are calculated for each tumor by subtracting the tumor volume on the day of first treatment (staging day) from the tumor volume on the specified observation day. The median T is calculated for the treated group and the median C is calculated for the control group. Then the ratio T/C is calculated and expressed as a percentage: T/C=(delta T/delta C)100.

(100) The dose is considered as therapeutically active when T/C is lower than 40% and very active when T/C is lower than 10%. If T/C is lower than 0, the dose is considered as highly active and the percentage of regression is dated (Plowman J, Dykes D J, Hollingshead M, Simpson-Herren L and Alley M C. Human tumor xenograft models in NCI drug development. In: Feibig H H B A, editor. Basel: Karger; 1999 p 101-125):

(101) % tumor regression is defined as the % of tumor volume decrease in the treated group at a specified observation day compared to its volume on the first day of first treatment.

(102) At a specific time point and for each animal, % regression is calculated. The median % regression is then calculated for the group:

(103) $\%\mathrm{regression}\left(\mathrm{at}t\right)=\frac{{\mathrm{volume}}_{t0}-{\mathrm{volume}}_t}{{\mathrm{volume}}_{t0}}100$

(104) Partial regression (PR): Regressions are defined as partial if the tumor volume decreases to 50% of the tumor volume at the start of treatment.

(105) Complete regression (CR): Complete regression is achieved when tumor volume=0 mm.sup.3 (CR is considered when tumor volume cannot be recorded).

(106) Results:

(107) The results are presented on FIG. 4 and Table 9 (below). Using a single administration schedule at 2.5 and 5 mg/kg, all conjugates tested in this study did not induce toxicity.

(108) chMAb1-SPDB-DM4 was very active at 5 and 2.5 mg/kg with a T/C of 0 and 7% (p<0.0001 and p=0.0170 vs control), respectively. chMAb4-SPDB-DM4 and chMAb5-SPDB-DM4 were highly active at 5 mg/kg with T/C of 5 and 7% (p<0.0001 vs control), respectively and tumor regression of 25 and 65%, respectively. They were very active at 2.5 mg/kg with T/C of 7 and 2% (p=0.0152 and p=0.0020 vs control), respectively. chMAb2-SPDB-DM4 was highly active at 5 and 2.5 mg/kg with T/C of 10 and 8% (p<0.0001 vs control), respectively, tumor regression of 82 and 39%, respectively and 3 and 1 CR/6, respectively.

(109) From these results, all chimeric conjugates chMAb4-SPDB-DM4, chMAb1-SPDB-DM4, chMAb5-SPDB-DM4 and chMAb2-SPDB-DM4 were usable to develop a therapeutic ADC.

(110) TABLE-US-00011 TABLE 9 Evaluation of the anti-tumor activity of chMAb1-SPDB-DM4, chMAb2-SPDB-DM4, chMAb4-SPDB-DM4, and chMAb5-SPDB- DM4 conjugates against primary human colon adenocarcinoma CR-IGR-034P in SCID female mice. Average body weight change Median % Route/ Dosage in Schedule Drug in % per mouse Median of Dosage in mg/kg per in death at nadir T/C in regression Regressions Biostatistic Agent.sup.1 mL/kg injection days (Day) (day of nadir) % (day) (day) Partial Complete p value.sup.2 Comments chMAb1- IV 5 14 0/6 0.3 (D23) 0 (D33) 2/6 0/6 <0.0001 Very active SPDB- (10 mL/Kg) 2.5 14 0/6 1.2 (D22) 7 (D21) 0/6 0/6 =0.0170 Very active DM4 chMAb2- IV 5 14 0/6 1.1 (D29) 10 (D33) 82 (D33) 6/6 3/6 <0.0001 Highly SPDB- (10 mL/Kg) active DM4 2.5 14 0/6 1.3 (D57) 8 (D28) 39 (D28) 2/6 1/6 <0.0001 Highly active chMAb4- IV 5 14 0/6 1.9 (D22) 5 (D28) 25 (D28) 2/6 0/6 <0.0001 Highly SPDB- (10 mL/Kg) active DM4 2.5 14 0/6 1.8 (D21) 7 (D25) 0/6 0/6 =0.0152 Very active chMAb5- IV 5 14 0/6 1.8 (D29) 7 (D33) 65 (D33) 4/6 0/6 <0.0001 Highly SPDB- (10 mL/Kg) active DM4 2.5 14 0/6 0.8 (D23) 2 (D21) 0/6 0/6 =0.0020 Very active Control 14 3.6 (D29) .sup.1drug formulation: HGS(10 mM Histidine, 130 mM Glycine, 5% v/v Sucrose, 0.01% Tween 80) pH 7.4 .sup.2p-value: Dunett's test versus control after 2-way Anova with repeated measures on rank transformed changes of tumour volume from baseline.

Example 6

Humanization of the Anti-CEACAM5_MAb2 mAb

(111) In this example, humanized variants of parental murine IgG MAb2 have been designed in silico. The resulting mAbs were produced and provided similar characteristics as the chimeric IgG ch-MAb2.

Example 6.1

4D-Humanization Protocol

(112) a) Humanization Based on Molecular Dynamic Trajectories

(113) The VL & VH sequences of the murine MAb2 clone were compared against the protein data base (PDB) (Berman et al., Nucleic Acids Research, 2000, 28:235-242). The following templates were used: light and heavy chain framework3EHB (90.9% Framework light chain identity and 90.8% Framework heavy chain identity), L11I8M, L21F6L, L31P7K, H12QHR, H21IGT and H31P4B to build a homology model of anti-CEACAM5 LC and HC using Molecular Operating Environment (MOE) (v. 2011.10Chemical Computing Group, Quebec, Canada). The homology model was subsequently energy minimized using the standard procedures implemented in MOE.

(114) A molecular dynamics (MD) simulation of the minimized 3D homology model of the murine MAb2 was subsequently performed, with constraints on the protein backbone at 500 K temperature for 1.1 nanoseconds (ns) in Generalized Born implicit solvent. 10 diverse conformations were extracted from this first MD run every 100 picoseconds (ps) for the last 1 ns. These diverse conformations were then each submitted to a MD simulation, with no constraints on the protein backbone and at 300 K temperature, for 2.3 ns. For each of the 10 MD runs, the last 2,000 snapshots, one every ps, from the MD trajectory were then used to calculate, for each murine MAb2 amino acid, its root mean square deviations (rmsd) compared to a reference medoid position. By comparing the average rmsd on the 10 separate MD runs of a given amino acid to the overall average rmsd of all MAb2 murine amino acids, one decides if the amino acid is flexible enough, as seen during the MD to be considered as likely to interact with T-cell receptors and responsible for activation of the immune response. 32 amino acids were identified as flexible in the murine MAb2 antibody, excluding the CDR and its immediate 5 vicinity.

(115) The motion of the 60 most flexible murine MAb2 amino acids, during the 20 ns (102 ns) of molecular dynamic simulation, were then compared to the motion of the corresponding flexible amino acids of 49 human 3D homology models, for each of which were run the same MD simulations. These 49 human germline models have been built by systematically combining a representative panel of 7 human light chains (namely vk1, vk2, vk3, vk4, vlambda1, vlambda2, vlambda3) with a representative panel of 7 human heavy chains (namely vh1a, vh1b, vh2, vh3, vh4, vh5, vh6) (Nucleic Acid Research, 2005, Vol. 33, Database issue D593-D597).

(116) The vk1-vh6 combination showed the highest (72.6%) 4D similarity of its flexible amino acids compared to the flexible amino acids of the murine MAb2 antibody; this model was therefore used to humanize the MAb2 antibody focusing on the flexible amino acids. For the pairwise amino acid association between the murine MAb2 and vk1-vh6 amino acids, the 2 sequences were aligned based on the optimal 3D superposition of the alpha carbons f the 2 corresponding homology models.

(117) b) Stabilizing Mutations

(118) To improve the stability of VL and VH regions of the anti-CEACAM5 antibody, the amino acids of the light and heavy chains with low frequency of occurrence vs. their respective canonical sequences, excluding the CDRs, are originally proposed to be mutated into the most frequently found amino acids (Gth>0.5 kcal/mol; (Monsellier et al. J. Mol. Biol. 2006, 362, 580-593). A first list of consensus mutations for the LC and for the HC has been restricted to the amino acids found in the closest human model (i.e vk1-vh6). None of these mutations are located in the Vernier zone (Foote et al., J. Mol. Biol. 1992, 224, 487-499). Other criteria are taken into account to consider these consensus mutations for potentially stabilizing the anti-CECAM5 MAb2 antibody. These criteria are a favourable change of hydropathy at the surface or a molecular mechanics based predicted stabilization of the mutant. Stabilizing mutations reported to be successful in the literature (Bedouelle, H. J. Mol. Biol. 2006, 362, 580-593; Steipe B. J. Mol. Biol. 1994, 240, 188-192) were considered.

(119) c) Removal of Unwanted Sequence Motifs

(120) The following motifs of sequences were considered: Asp-Pro (acide labile bond), Asn-X-Ser/Thr (glycosylation, X=any amino acid but Pro), Asp-Gly/Ser/Thr (succinimide/iso-asp formation in flexible areas), Asn-Gly/His/Ser/Ala/Cys (exposed deamidation sites), Met (oxidation in exposed area). The resulting humanized sequences were blasted for sequence similarity against the Immune Epitope Data Base (IEDB) database ((PLos Biol (2005) 3(3)e91) http://www.immuneepitope.org) to ensure that none of the sequences contain any known B- or T-cell epitope listed in.

(121) d) Humanized VH and VL Regions

(122) Three versions for the light chain (VL1, VL1a, and VL1c) and three versions for the heavy chain are proposed (VH1, VH1a and VH1b). The particular combination of amino acid residues altered in each humanized MAb2 VL and VH variant are set forth in Table 10 and Table 11, respectively. The complete amino acid sequences of the humanized VH and VL domains are set forth in Table 12.

(123) The VL1 variant displays 5 mutations which derive from the direct comparison between the non-CDR most flexible amino acids of the anti-CEACAM5 MAb2 light chain and the vk1 human light chain sequence.

(124) The VL1a variant derives from VL1 and includes 4 new mutations that are consensus (vk1 sequence) and potentially stabilizing. Moreover, 1 of these mutations addresses a potentially problematic deamidation site (D.sub.17T.sub.18).

(125) The VL1c variant derives from VL1a with the introduction of 1 mutation R instead of K at position 52. Indeed, this K52 is located in the CDR L2 and could be a target for the conjugation process.

(126) The VH1 variant displays 7 mutations which derive from the direct comparison between the non-CDR most flexible amino acids of the anti-CEACAM5 heavy chain and the vh6 human heavy chain sequence.

(127) The VH1a variant derives from VH1 and includes 4 new mutations that are consensus (vh6 sequence) and potentially stabilizing.

(128) The humanized anti-CEACAM5 MAb2 antibody VL and VH domains were combined as follows: VL1 and VH1; VL1a and VH1a; VL1c and VH1a; VL1a and VH1b

(129) TABLE-US-00012 TABLE 10 Mutations of the VL variants of the anti-CEACAM5 MAb2 antibody Mouse MAb2 VL VL1 VL1d VL1a VL1c E17 D D D D T18 R R Q40 P P P P Q45 K K K K K52 R R Q70 D D K74 T T T T N76 S S S S G84 A A S85 T T

(130) TABLE-US-00013 TABLE 11 Mutations of the VH variants of the anti-CEACAM5 MAb2 antibody Mouse MAb2 VH VH1 VH1a G9 P P V10 G G K19 S S K43 R R R44 G F60 A D62 S S Q65 K K N84 K87 T T I89 V A113 S

(131) TABLE-US-00014 TABLE12 VHandVLaminoacid sequencesofexemplaryhumanized anti-CEACAM5antibodies. VHorVL SEQ variant Sequence IDNO. clone EVQLQESGPGLVKPGGSLSLSCAASGFVFS SEQID MAb2 SYDMSWVRQTPERRLEWVAYISSGGGITYF NO:51 VH1 PSTVKGRFTVSRDNAKNTLYLQMNSLTSED TAIYYCAAHYFGSSGPFAYWGQGTLVTVSA clone EVQLQESGPGLVKPGGSLSLSCAASGFVFS SEQID MAb2 SYDMSWVRQTPERGLEWVAYISSGGGITYA NO:5 VH1a PSTVKGRFTVSRDNAKNTLYLQMNSLTSED TAVYYCAAHYFGSSGPFAYWGQGTLVTVSS clone DIQMTQSPASLSASVGDTVTITCRASENIF SEQID MAb2 SYLAWYQQKPGKSPKLLVYNTKTLAEGVPS NO:17 VL1 RFSGSGSGTQFSLTISSLQPEDFGSYYCQH HYGTPFTFGSGTKLEIK clone DIQMTQSPASLSASVGDRVTITCRASENIF SEQID MAb2 SYLAWYQQKPGKSPKLLVYNTKTLAEGVPS NO:23 VL1a RFSGSGSGTDFSLTISSLQPEDFATYYCQH HYGTPFTFGSGTKLEIK clone DIQMTQSPASLSASVGDRVTITCRASENIF SEQID MAb2 SYLAWYQQKPGKSPKLLVYNTRTLAEGVPS NO:29 VL1c RFSGSGSGTDFSLTISSLQPEDFATYYCQH HYGTPFTFGSGTKLEIK clone DIQMTQSPASLSASVGDTVTITCRASENIF SEQID MAb2 SYLAWYQQKPGKSPKLLVYNTRTLAEGVPS NO:55 VL1d RFSGSGSGTQFSLTISSLQPEDFGSYYCQH HYGTPFTFGSGTKLEIK

Example 6.2

Sequence of Humanized Anti CEACAM5 mAb

(132) From the amino acid sequences of in silico VL and VH variants, the nucleic acid sequences were derived and synthesized by Geneart. The sequences were cloned into expression vectors in fusion with the human IgG1 or the human Ckappa constant domain coding sequences respectively.

Example 6.3

Production and In Vitro Characterization

(133) Batches of humanized mAbs were produced by transient expression in HEK293 and purified by protein A affinity chromatography. Structure and identity were confirmed by SDS-PAGE analysis, Size Exclusion Chromatography and Mass Spectrometry.

(134) Affinity to human and cynomolgus CEACAM5 was verified by ELISA, EC50 are provided on Table 13.

(135) TABLE-US-00015 TABLE 13 Affinity of humanized anti-CEACAM5 mAb to human and cynomolgus CEACAM5 Human Cynomolgus CEACAM5 CEACAM5 EC50 EC50 code mAb (nM) CV (nM) CV huMAb2-1 MAb2VL1VH1-IgG1 0.22 4.7 % 0.20 7.9% huMAb2-2 MAb2_VL1aVH1a-IgG1 0.20 9.2 % 0.17 5.0% huMAb2-3 MAb2_VL1cVH1a-IgG1 0.18 11 % 0.19 4.3% huMAb2-4 MAb2_VL1d VH1-IgG1 0.22 4.3 % 0.17 5.0% chMAb2 MAb2-IgG1 0.16 9.9 % 0.17 3.0%

(136) Specificity to human CEACAM5 versus human CEACAM1, CEACAM6, CEACAM7 and CEACAM8 was verified by ELISA. It was reported as the percentage of binding compared to full binding with human CEACAM5, see Table 14.

(137) TABLE-US-00016 TABLE 14 Percentage of binding of humanized anti-CEACAM5 mAb to human CEACAMs hCEA code mAb CAM5 CAM1 CAM6 CAM7 CAM8 huMAb2-1 MAb2_VL1VH1-IgG1 100% 0.3% 0.2% 0.3% 0.9% huMAb2-2 MAb2_VL1aVH1a- 100% 0.3% 0.3% 0.3% 0.5% IgG1 huMAb2-3 MAb2_VL1cVH1a- 100% 0.2% 0.3% 0.3% 0.6% IgG1 huMAb2-4 MAb2_VL1d VH1-IgG1 100% 0.3% 0.3% 0.3% 1.4% chMAb2 MAb2-IgG1 100% 0.3% 0.3% 0.3% 0.6%

(138) Epitope binding domain was verified by ELISA and showed that humanized variants recognized the A3-B3 domain specifically. It was reported as the percentage of binding compared to full binding with human CEACAM5 on Table 15.

(139) TABLE-US-00017 TABLE 15 Percentage of binding of humanized anti-CEACAM5 mAb to human CEACAM5 domains hCEACAM5 code mAb N-ter-A1-B1 A2-B2 A3-B3 huMAb2-1 MAb2_VL1VH1-IgG1 0.4% 0.3% 100% huMAb2-2 MAb2_VL1aVH1a-IgG1 0.4% 0.3% 100% huMAb2-3 MAb2_VL1cVH1a-IgG1 0.4% 0.4% 100% huMAb2-4 MAb2_VL1dVH1-IgG1 0.3% 0.3% 100% chMAb2 MAb2-IgG1 0.5% 0.3% 100%

(140) The binding kinetics of humanized anti-CEACAM5_MAb2 variants, compared with chimeric MAb2, to recombinant human CEACAM5 (hCEACAM5) and cynomolgus monkey CEACAM5 (cCEACAM5) were determined by surface plasmon resonance assay using a BIAcore 2000 (BIAcore Inc., Uppsala, N.J.).

(141) Briefly, a CM5 BIAcore biosensor chip was docked into the instrument and activated with 70 L of 1:1 NHS/EDC at room temperature. A mouse anti- human Fc IgG1 (BIAcore #BR-1008-39) (50 g/mL in 1M acetate buffer, pH5) were immobilized on the activated chips in all flow cells. The immobilization was carried out at a flow rate of 10 L/min up to saturation. The chip was then blocked by injection of 70 L of ethanolamine-HCl, pH8.5, followed by one wash with 3M MgCl2. To measure the binding of anti-CEACAM5 mAbs to the human CEACAM5 protein or cynomolgus CEACAM5 protein, antibodies were used at 1-5 g/mL in BIAcore running buffer (HBS-EP). Antigens (human CEACAM5 or cynomolgus CEACAM5) were injected from 1 to 500 nM. Following completion of the injection phase, dissociation was monitored in a BIAcore running buffer at the same flow rate for 600 sec. The surface was regenerated between injections using 25 L MgCl2 3M (230s). Individual sensorgrams were analyzed using BIAevaluation software.

(142) TABLE-US-00018 TABLE 16 binding of humanized anti-CEACAM5 mAb to human and monkey CEACAM5 Human Cynomolgus CEACAM5 CEACAM5 mAb KD (nM) KD (nM) huMAb2-1 9.8 41.7 huMAb2-2 24.5 96.0 huMAb2-3 11.7 73.5 huMAb2-4 6.9 38.6 chMAb2 9.9 52.3

(143) Specificity of humanized anti-CEACAM5_MAb2 variants, compared with chimeric MAb2, to cynomolgus CEACAM5 versus cynomolgus CEACAM1, CEACAM6 and CEACAM8 was verified by ELISA. It was reported as the percentage of binding compared to full binding with CEACAM5 or binding at the EC.sub.50, see Table 17 below

(144) TABLE-US-00019 TABLE 17 Percentage of binding of humanized anti-CEACAM5 mAb to cynomolgus CEACAMs Cynomolgus CEA code mAb CAM5 CAM1 CAME CAM8 huMAb2-1 MAb2_VL1VH1-IgG1 100% 0.3% 0.3% 3.6% huMAb2-2 MAb2_VL1aVH1a-IgG1 100% 0.3% 0.3% 0.9% huMAb2-3 MAb2_VL1cVH1a-IgG1 100% 0.3% 0.4% 1.2% huMAb2-4 MAb2_VL1dVH1-IgG1 100% 0.3% 0.3% 3.2% chMAb2 MAb2-IgG1 100% 0.2% 0.3% 1.2%

Example 6.4

Characterization of Humanized Variants of Mab2 Obtained by Grafting to Human Germline Frameworks

(145) In this example, humanized variants of Mab2 were obtained by a CRD-grafting approach. Further, the CDRs of the humanized antibody were submitted to an alanine-scanning approach to show that several positions could be substituted without affecting the binding to human and Macaca fascicularis CEACAM5.

(146) The sequence of a humanized version of Mab2 was generated first in silico by selecting human germline frameworks on the basis of structural homology with the murine antibody Mab2. For the light chain, the selected human frameworks are defined by genes IGKV1D-39*01 and IGKJ2*02 and for the heavy chain, by genes IGHV3-23*04 and IGHJ4*01. The six CDRs of Mab2.sub.K52R were grafted into these human frameworks. Three back-mutations were introduced, corresponding to positions 34 and 53 in the VL (SEQ ID NO. 34) (FR2-L and FR3-L regions, respectively) and position 50 in the VH (SEQ ID NO. 33) (FR2-H region), resulting in the following sequence, defined as MAb2_VLg5VHg2.

(147) TABLE-US-00020 TABLE18 VHandVLsequences ofMAb2_VLg5VHg2 VHorVL SEQ variant Sequence IDNO. MAb2_VHg2 EVQLVESGGGLVQPGGSLRLSCAASGFVFS SEQID SYDMSWVRQAPGKGLEWVSYISSGGGITYY NO:74 ADSVKGRFTISRDNSKNTLYLQMNSLRAED TAVYYCAAHYFGSSGPFAYWGQGTLVTVSS MAb2_VLg5 DIQMTQSPSSLSASVGDRVTITCRASENIF SEQID SYLAWYQQKPGKAPKLLIYNTRTLQSGVPS NO:75 RFSGSGSGTDFTLTISSLQPEDFATYYCQH HYGTPFTFGQGTKLEIK

(148) Several variants of MAb2_VLg5VHg2 were obtained by the single replacement of each amino acid of the six CDRs, preferentially by an alanine. When alanine is already found in MAb2_VLg5VHg2 CDRs, which occurs in H-CDR3, another amino acid was substituted (Val at residue 97, Arg at residue 98 and Asp at residue 108 of SEQ ID NO:74).

(149) From the in silico amino acid sequences of VL and VH variants, the nucleic acid sequences were derived and generated by gene synthesis. The sequences were cloned into a mammalian expression vector in fusion with the human IgG1 or the human Ckappa constant domain coding sequences respectively. Humanized MAb2_VLg5VHg2, single variants differing from it by one position, and a limited number of combination mutants, were produced by transient expression in HEK293 cells. Cell supernatants containing the secreted IgGs (20 to 70 g/ml) were diluted to 1 g/ml for use in binding assays to human CEACAM5 ECD, Macaca fascicularis ECD and A3-B3 domain of human CEACAM5.

(150) To evaluate impact of these modifications, IgGs binding was determined by measuring SPR signals with a Biacore T200 unit (GE Healthcare). Anti-human Fc antibodies were coupled to a Series S CM5 chip via amine coupling kit to reach a level of 10,000 response units (RU). Approximately 300 to 600 RU of each variant were captured by injecting supernatants at 1 g/mL with a contact time of 60 seconds and a flow rate of 10 L/min. All experiments were performed at 25 C. with HBS-EP+ (10 mM Hepes, 150 mM NaCl, 3 mM EDTA, 0.05% surfactant P20) as the running buffer. In a screening mode, the human CEACAM5/cynomolgus CEACAM5/human A3B3 domain was injected at 50 nM over the captured IgG variants at a flow rate of 30 L/min for 1 minute. A dissociation phase of 60 seconds was held before the surface was regenerated with 1 pulse injection of 3 M MgCl.sub.2 at a flow rate of 10 L/minute and a contact time of 30 seconds.

(151) For each experiment, response data were processed using a reference surface, thereby allowing correction for bulk refractive index changes and any non-specific binding. Data were double referenced using response from blank injections. According to the screening method described in an application note from GE Healthcare (Application note 28-9777-72 AA), two parameters were considered to rank the variants with respect to binding characteristics. First, the binding activity is estimated by the proportion of the theoretical maximum signal measured (percentage of Rmax, see FIG. 16). Second, the percentage of remaining signal is calculated using dissociation report points (the first one 10 seconds after the end of the injection and the second one 50 seconds after the end of injection) and reflects the stability of binding (see FIG. 17).

(152) Single alanine variants at the following positions demonstrated equivalent binding parameters, as compared to the original antibody, suggesting that the CDR amino acids at these positions are neutral for the binding: LC residues 27, 28, 29, 31, 51, 52, 89, 90, 93, 94, 96, 97 and HC residues 26 to 31, 51 to 58, 97, 103, 104, 107, 109. Behaviors of these variants are similar with human and monkey CEACAM5, thus maintaining their cross-reactivity. Binding to A3-B3 domain of CEACAM5 was also found unaffected. Some combinations of two neutral substitutions were also generated and were found to result in unaltered binding parameters, as illustrated with association of LC_T51A with LC_T94A, LC_S31A with HC_G54Y, or LC_T53I with HC_S53A.

(153) Conversely, at all the other CDR positions, substitution of alanine for the original amino acid was found to induce a complete loss of binding or dramatically altered binding parameters. Position 101 of the heavy chain or positions 32 and 91 of the light chain are examples shown on FIGS. 16 and 17), A second set of variants consisted in testing more conservative mutations at some such positions. Doing that, we found that the following conservative substitution were neutral for antigen binding: Tyr for Phe at residue 30 of MAb2_VLg5, Phe for Tyr at residue 92 of the MAb2_VLg5, Ser for Ala at residue 98 of MAb2_VHg2 and Phe for Tyr at residue 100 of MAb2_VHg2 (shown on figures)

Example 7

Humanized Variants of MAb2 Drug Conjugates

Example 7.1

Production and Characterisation

(154) huMAb2-2-SPDB-DM4

(155) Analytical Data: MW(Ab)=147360 g/mol; MW(DM4)=780.38 g/mol .sub.280nm(Ab)=201400; .sub.280nm(Ab)=71693 .sub.280nm(DM4)=5180; .sub.280nm(DM4)=26159

(156) Under stirring, at RT, 19.4 mg of huMAb2-2 (C=5.1 mg/ml in PBS pH=7.4 buffer) are introduced in a vessel, followed by 0.375 ml of DMA and 0.0439 ml of nitro-SPDB linker solution (5.0 Eq-15 mM solution in DMA). Solution is vortexed for 30 sec and then slowly stirred at RT for 2 hours. An extra volume of 0.0044 ml of nitro-SPDB linker solution (5.0 Eq-15 mM solution in DMA) is added. After 2 hours at RT under magnetic stirring, 2 ml of PBS pH=7.5 buffer and 0.0702 ml of DM4 solution (15 mM solution in DMA) were successively added. After 2 hours at RT, crude reaction mixture is filtered on 0.45 m filter and purified on HiPrep 26/10 desalting column (Sephadex G25, GE Healthcare), pre-conditioned with 1CV of NaOH 1M, 2 CV of water and 2 CV of histidine (10 mM), glycine (130 mM), sucrose (5%), pH=5.5 buffer. Conjugate is eluted with histidine (10 mM), glycine (130 mM), sucrose (5%), pH=5.5 buffer, and monomeric conjugate fractions are pooled and filtered on 0.22 m filter.

(157) 10.3 ml of huMAb2-2-SPDB-DM4 conjugate (c=1.35 mg/ml) was thus obtained as a colorless clear solution. The conjugate is then analyzed for final drug load and monomeric purity: DAR (UV)=3.7; DAR (SEC)=3.6; RT=17.29 min; monomeric purity=97.9%.

(158) Result of HRMS analysis is shown on FIG. 12.

(159) huMAb2-1-SPDB-DM4

(160) Analytical Data: MW(Ab)=147563 g/mol; MW(DM4)=780.38 g/mol .sub.280nm(Ab)=201400; .sub.252nm(Ab)=69669 .sub.280nm(DM4)=5180; .sub.252nm(DM4)=26159

(161) Under stirring, at RT, 3.8 ml of a solution of huMAb2-1 (C=5.08 mg/ml in PBS pH=7.4 buffer) are introduced in a vessel, followed by 0.341 ml of DMA and 0.0392 ml of nitro-SPDB linker solution (4.5 Eq-15 mM solution in DMA). Solution is vortexed for 30 sec and then slowly stirred at RT for 3 hours. An extra volume of 0.0087 ml of nitro-SPDB linker solution (1.0 Eq-15 mM solution in DMA) is added. After 2 hours at RT under magnetic stirring, 2.62 ml of PBS pH7.5 buffer, 0.254 ml of DMA and 0.076 ml of DM4 solution (15 mM solution in DMA) were successively added. After 1 hour at RT, crude reaction mixture is filtered on 0.45 m filter and purified on HiPrep 26/10 desalting column (Sephadex G25, GE Healthcare), pre-conditioned with 1 CV of NaOH 1 M, 2 CV of water and 2 CV of histidine (10 mM), glycine (130 mM), sucrose (5%), pH=5.5 buffer. Conjugate is eluted with histidine (10 mM), glycine (130 mM), sucrose (5%), pH=5.5 buffer, and monomeric conjugate fractions are pooled and filtered on 0.22 m filter.

(162) 9.5 ml of huMAb2-1-SPDB-DM4 conjugate (c=1.35 mg/ml) was thus obtained as a colorless clear solution. The conjugate is then analyzed for final drug load and monomeric purity: DAR (UV)=4.1; DAR (SEC)=4.0; RT=17.39 min; monomeric purity=96.7%.

(163) Result of HRMS analysis is shown on FIG. 13.

(164) huMAb2-3-SPDB-DM4

(165) Analytical Data: MW(Ab)=147417 g/mol; MW(DM4)=780.38 g/mol .sub.280nm(Ab)=201400; .sub.252nm(Ab)=71451 .sub.280nm(DM4)=5180; .sub.252nm(DM4)=26159

(166) Under stirring, at RT, 3.8 ml of a solution of huMAb2-3 (C=5.09 mg/ml in PBS pH=7.4 buffer) are introduced in a vessel, followed by 0.336 ml of DMA and 0.0437 ml of nitro-SPDB linker solution (5 Eq-15 mM solution in DMA). Solution is vortexed for 30 sec and then slowly stirred at RT for 3 hours. An extra volume of 0.0035 ml of nitro-SPDB linker solution (0.4 Eq-15 mM solution in DMA) is added. After 1 hour at RT under magnetic stirring, 2.60 ml of PBS pH7.5 buffer, 0.248 ml of DMA and 0.074 ml of DM4 solution (15 mM solution in DMA) were successively added. After 1 hour at RT, crude reaction mixture is filtered on 0.45 m filter and purified on HiPrep 26/10 desalting column (Sephadex G25, GE Healthcare), pre-conditioned with 1CV of NaOH 1M, 2 CV of water and 2 CV of histidine (10 mM), glycine (130 mM), sucrose (5%), pH=5.5 buffer. Conjugate is eluted with histidine (10 mM), glycine (130 mM), sucrose (5%), pH=5.5 buffer, and monomeric conjugate fractions are pooled and filtered on 0.22 m filter.

(167) 11 ml of huMAb2-3-SPDB-DM4 conjugate (c=1.08 mg/ml) was thus obtained as a colorless clear solution. The conjugate is then analyzed for final drug load and monomeric purity: DAR (UV)=3.9; DAR (SEC)=3.8; RT=17.44 min; monomeric purity=98.4%.

(168) Result of HRMS analysis is shown on FIG. 14.

(169) huMAb2-4-SPDB-DM4

(170) Analytical Data: MW(Ab)=147628 g/mol; MW(DM4)=780.38 g/mol .sub.280nm(Ab)=201400; .sub.252nm(Ab)=70628 .sub.280nm(DM4)=5180; .sub.252nm(DM4)=26159

(171) Under stirring, at RT, 3.8 ml of a solution of huMAb2-4 (C=5.09 mg/ml in PBS pH=7.4 buffer) are introduced in a vessel, followed by 0.345 ml of DMA and 0.0448 ml of nitro-SPDB linker solution (5 Eq-15 mM solution in DMA). Solution is vortexed for 30 sec and then slowly stirred at RT for 3 hours. An extra volume of 0.0027 ml of nitro-SPDB linker solution (0.3 Eq-15 mM solution in DMA) is added. After 1 hour at RT under magnetic stirring, 2.70 ml of PBS pH7.5 buffer, 0.263 ml of DMA and 0.075 ml of DM4 solution (15 mM solution in DMA) were successively added. After 1 hour at RT, crude reaction mixture is filtered on 0.45 m filter and purified on HiPrep 26/10 desalting column (Sephadex G25, GE Healthcare), pre-conditioned with 1CV of NaOH 1M, 2 CV of water and 2 CV of histidine (10 mM), glycine (130 mM), sucrose (5%), pH=5.5 buffer. Conjugate is eluted with histidine (10 mM), glycine (130 mM), sucrose (5%), pH=5.5 buffer, and monomeric conjugate fractions are pooled and filtered on 0.22 m filter.

(172) 11 ml of huMAb2-4-SPDB-DM4 conjugate (c=1.23 mg/ml) was thus obtained as a colorless clear solution. The conjugate is then analyzed for final drug load and monomeric purity: DAR (UV)=3.8; DAR (SEC)=3.8; RT=17.53 min; monomeric purity=99.3%.

(173) Result of HRMS analysis is shown on FIG. 15.

Example 7.2

In Vitro Cytotoxicity

(174) Material and Methods:

(175) The effect of the anti-CEACAM5 maytansinoid conjugates on tumor cell viability was assessed as described in example 3.4.

(176) Results:

(177) TABLE-US-00021 TABLE 19 Cytotoxic activities of the CEACAM5-specific humanized ADCs on CEACAM5 + MKN45 cell line Cytotoxic activity ADC IC.sub.50 (nM) StD chMAb2-SPDB-DM4 0.24 0.02 huMAb2-1-SPDB-DM4 0.18 0.01 huMAb2-2-SPDB-DM4 0.23 0.02 huMAb2-3-SPDB-DM4 0.16 0.01 Irrelevant ADC 8.52 2.07

(178) These chMAb2-SPDB-DM4, huMAb2-1-SPDB-DM4, huMAb2-2-SPDB-DM4, and huMAb2-3-SPDB-DM4 conjugates and the DM4 irrelevant conjugate showed in vitro cytotoxic activities on MKN45 cells in culture with an 1050 of 0.24, 0.18, 0.23, 0.16, and 8.52 nM respectively. The cytotoxic activities of the anti-CEACAM5 conjugates was 53 to 35 fold lower than the measured activity of the irrelevant DM4 conjugate indicating CEACAM5-mediated cytotoxic activities of the anti-CEACAM5 conjugates.

Example 7.3

In Vivo Efficacy Against Primary Colon CR-IGR-034P Tumors Implanted s.c. In Female CD-1 Nude Mice

(179) Material and Method

(180) Two humanized sequences as conjugates huMAb2-3-SPDB-DM4 and huMAb2-4-SPDB-DM4 were evaluated at 4-dose levels compared to the chMAb2-SPDB-DM4, against measurable primary colon CR-IGR-034P tumors implanted s.c. in female CD-1 nude mice. Control groups were left untreated. The doses conjugates were given in mg/kg. They were administered at 10, 5, 2.5 and 1.25 mg/kg by an intravenous (IV) bolus injection, on day 19 after tumor implantation.

(181) Toxicity and efficacy evaluation were performed as reported in example 5.

(182) Results:

(183) The results are presented in FIG. 5 and Table 20 (below).

(184) Using a single administration schedule at 1.25, 2.5, 5 and 10 mg/kg, all conjugates tested in this study did not induce toxicity.

(185) huMAb2-4-SPDB-DM4 and chMAb2-SPDB-DM4 were highly active at 10 mg/kg with T/C of 4% (p<0.0001 vs control) and tumor regression of 21 and 19%, respectively, active at 5 mg/kg with T/C of 12 (p=0.0105 vs control) and 17% (p=0.0417 vs control), respectively and marginally active at 2.5 mg/kg with T/C of 36 and 37% (ns vs control), respectively, and inactive at 1.25 mg/kg. huMAb2-3-SPDB-DM4 was highly active at 10 mg/kg with T/C of 6% (p<0.0001 vs control) and tumor regression of 31%, very active at 5 mg/kg with T/C of 4% (p<0.0001 vs control), active at 2.5 mg/kg with T/C of 12 (p=0.0322 vs control) and marginally active at 1.25 mg/kg T/C of 34% (ns vs control).

(186) From these results, both humanized sequences huMAb2-3-SPDB-DM4 and huMAb2-4-SPDB-DM4 were usable to develop a therapeutic ADC. huMAb2-3-SPDB-DM4 was the best of the both sequence.

(187) TABLE-US-00022 TABLE 20 Evaluation of the anti-tumor activity of huMAb2-3-SPDB-DM4 and huMAb2-4-SPDB-DM4 and chMAb2-SPDB-DM4 conjugates against primary human colon adenocarcinoma CR-IGR-034P in CD-1 female mice. Average body weight change Median % Route/ Dosage in Drug in % per mouse Median of Dosage in mg/kg per Schedule death at nadir T/C in regression Regressions Biostatistic Agent.sup.1 mL/kg injection in days (Day) (day of nadir) % (day) (day) Partial Complete p value.sup.2 Comments chMAb2- IV 10 19 0/6 7.3 (D20) 4 (D32) 19 (D32) 2/6 0/6 <0.0001 Highly SPDB- (10 mL/Kg) active DM4 5 19 0/6 4.5 (D45) 12 (D32) 0/6 0/6 =0.0105 Active 2.5 19 0/6 4.2 (D20) 36 (D32) 0/6 0/6 ns Marginally active 1.25 19 0/6 4.1 (D20) 42 (D32) 0/6 0/6 ns Inactive huMAb2-3- IV 10 19 0/6 4.3 (D27) 6 (D35) 31 (D35) 2/6 0/6 <0.0001 Highly SPDB-DM4 (10 mL/Kg) active 5 19 0/6 3.3 (D20) 4 (D38) 0/6 0/6 <0.0001 Very active 2.5 19 0/6 5.4 (D45) 12 (D38) 0/6 0/6 =0.0322 Active 1.25 19 0/6 3.0 (D27) 34 (D38) 0/6 0/6 ns Marginally active huMAb2-4- IV 10 19 0/6 3.7 (D22) 4 (D32) 21 (D32) 2/6 0/6 <0.0001 Very SPDB-DM4 (10 mL/Kg) active 5 19 0/6 3.2 (D27) 17 (D32) 0/6 0/6 =0.0417 Very active 2.5 19 0/6 3.4 (D20) 37 (D32) 0/6 0/6 ns Marginally active 1.25 19 0/6 2.8 (D27) 50 (D32) 0/6 0/6 ns Inactive Control 19 3.9 (D24) .sup.1drug formulation: HGS (10 mM Histidine, 130 mM Glycine, 5% v/v Sucrose, 0.01% Tween80) pH 7.4; .sup.2p-value: Dunnett's test versus control after 2-way Anova with repeated measures on rank transformed changes of tumour volume torn baseline; ns: no significant

Example 7.4

In Vivo Efficacy Against Primary Stomach STO-IND-006 Tumors Implanted s.c. In Female SCID Mice

(188) Material and Method

(189) The humanized conjugate huMAb2-3-SPDB-DM4 was evaluated at 3-dose levels against measurable primary stomach STO-IND-006 tumors implanted s.c. in female SCID mice. Control groups were left untreated. The doses conjugates were given in mg/kg. They were administered at 10, 5 and 2.5 mg/kg by an intravenous (IV) bolus injection, on day 27 after tumor implantation.

(190) Toxicity and efficacy evaluation were performed as reported in example 5.

(191) Results:

(192) Using a single administration schedule at 2.5, 5 and 10 mg/kg, huMAb2-3-SPDB-DM4 did not induce toxicity.

(193) As shown on FIG. 6 and in Table 21, huMAb2-3-SPDB-DM4 was very active at 10 mg/kg with T/C of 7% (p<0.0001 vs control), active at 5 mg/kg with T/C of 36% (p=0.0281 vs control) and inactive at 2.5 mg/kg.

(194) TABLE-US-00023 TABLE 21 Evaluation of the anti-tumor activity of huMAb2-3-SPDB-DM4 conjugates against primary human stomach adenocarcinoma STO-IND-006 in SCID female mice Average body weight change Median % Route/ Dosage in Drug in % per mouse Median of Dosage in mg/kg per Schedule death at nadir T/C in regression Regressions Biostatistic Agent.sup.1 mL/kg injection in days (Day) (day of nadir) % (day) (day) Partial Complete p value.sup.2 Comments huMAb2-3- IV 10 27 0/6 10.5 (D45) 7 (D34) 0/6 0/6 <0.0001 Very active SPDB- (10 mL/Kg) 5 27 0/6 8.4 (D45) 36 (D45) 0/6 0/6 =0.0281 Active DM4 2.5 27 0/6 5.8 (D45) 50 (D38) 0/6 0/6 ns Inactive Control 27 2.5 (D38) .sup.1drug formulation: HGS (10 mM Histidine, 130 mM Glycine, 5% v/v Sucrose, 0.01% Tween80) pH 7.4; .sup.2p-value: Dunnett's test versus control after 2-way Anova with repeated measures on rank transformed changes of tumour volume from baseline; ns: no significant

Example 7.5

In Vivo Efficacy Against Primary Lung LUN-NIC-0014 Tumors Implanted s.c. In Female SCID Mice

(195) Material and Method

(196) The humanized conjugate huMAb2-3-SPDB-DM4 was evaluated at 3-dose levels against measurable primary lung LUN-NIC-0014 tumors implanted s.c. in female SCID mice. Control groups were left untreated. The doses conjugates were given in mg/kg. It was administered at 10, 5 and 2.5 mg/kg by an intravenous (IV) bolus injection, on day 29 after tumor implantation.

(197) Toxicity and efficacy evaluation were performed as reported in example 5.

(198) Results

(199) Using a single administration schedule at 2.5, 5 and 10 mg/kg, huMAb2-3-SPDB-DM4 did not induce toxicity.

(200) As shown on FIG. 18 and in Table 22, huMAb2-3-SPDB-DM4 was highly active at 10 and 5 mg/kg with T/C inferior to 0 (p<0.0001 vs control) and tumor regression of 67 and 57% respectively and active at 2.5 mg/kg with T/C of 12% (p=0.0363 vs control).

(201) TABLE-US-00024 TABLE 22 Evaluation of the anti-tumor activity of huMAb2-3-SPDB-DM4 conjugate against primary human lung adenocarcinoma LUN-NIC-0014 in SCID female mice Average body weight change in Median % Route/ Dosage in Drug % per mouse at Median of Dosage mg/kg per Schedule death nadir (day T/C in regression Regressions Biostatistic Agent.sup.1 in mL/kg injection in days (Day) of nadir) % (day 42) (day 42) Partial Complete p value.sup.2 Comments huMAb2-3- IV 10 29 0/6 +1.7 (D32) <0 67 5/6 1/6 <0.0001 Highly SPDB-DM4 (10 mL/ active kg) 5 29 0/6 1.1 (D36) <0 57 4/6 0/6 <0.0001 Highly active 2.5 29 0/6 +0.5 (D32) 12 (D39) 0/6 0/6 0.0363 Active (D39) Control +0.1 (D34) .sup.1drug formulation: HGS (10 mM Histidine, 130 mM Glycine, 5% v/v Sucrose, 0.01% Tween80) pH 7.4; .sup.2p-value: Dunnett's test versus control after 2-way Anova with repeated measures on rank transformed changes of tumour volume from baseline.

Example 7.6

In Vivo Efficacy Against Primary Colon CR-IGR-034P Tumors Implanted s.c. In Female SCID Mice

(202) Material and Method

(203) Three conjugates, constituted by the humanized huMAb2-3 conjugated to the DM4 through two different linkers (SPDB and sulfo-SPDB), were evaluated at 2-dose levels against measurable primary colon CR-IGR-034P tumors implanted s.c. in female SCID mice. Control groups were left untreated. The doses conjugates were given in mg/kg. They were administered at 5 and 2.5 mg/kg by an intravenous (IV) bolus injection, on day 19 after tumor implantation.

(204) Toxicity and efficacy evaluation were performed as reported in example 5.

(205) Results

(206) Using a single administration schedule at 2.5 and 5 mg/kg, huMAb2-3-SPDB-DM4 and huMAb2-3-sulfo-SPDB-DM4 did not induce toxicity.

(207) As shown on FIG. 19 and in Table 23, huMAb2-3-SPDB-DM4 was active at 5 and 2.5 mg/kg with T/C of 12% and 40%, respectively (p<0.0001 vs control), huMAb2-3-sulfo-SPDB-DM4 was highly active at 5 mg/kg with T/C inferior to 0% (p<0.0001 vs control) and a tumor regression of 12% and active at 2.5 mg/kg with T/C of 1 (p<0.0001 vs control).

(208) TABLE-US-00025 TABLE 23 Evaluation of the anti-tumor activity of huMAb2-3-SPDB-DM4 and huMAb2-3-sulfo-SPDB-DM4 conjugates against primary human colon adenocarcinoma CR-IGR-034P in SCID female mice Average body weight change in % per Median Median % Route/ Dosage in Schedule Drug mouse at T/C in of Dosage in mg/kg per in death nadir (day of % regression Regressions Biostatistic Agent.sup.1 mL/kg injection days (Day) nadir) (day 34) (day 34) Partial Complete p value.sup.2 Comments huMAb2-3- IV 5 19 0/6 +1.6 (D20) 12 0/6 0/6 <0.0001 Active SPDB-DM4 (10 mL/kg) 2.5 19 0/6 1.5 (D38) 40 0/6 0/6 <0.0001 Marginally active huMAb2-3- IV 5 19 0/6 +0.1 (D20) <0 12 0/6 0/6 <0.0001 Highly sulfo-SPDB- (10 mL/kg) active DM4 2.5 19 0/6 +0.7 (D20) 11 0/6 0/6 <0.0001 Active 2.5 19 0/6 +2.5 (D20) 66 0/6 0/6 0.0306 Inactive Control +0.5 (D34) .sup.1drug formulation: HGS (10 mM Histidine, 130 mM Glycine, 5% v/v Sucrose, 0.01% Tween80) pH 7.4; .sup.2p-value: Dunnett's test versus control after 2-way Anova with repeated measures on rank transformed changes of tumour volume from baseline.

Example 8

Development of an Immunohistochemistry (IHC) Protocol Dedicated to the Detection of Human and Monkey CEACAM5 Protein in Formalin-Fixed and Paraffin Embedded (FFPE) Tissues

(209) Materials and Methods

(210) Tissues

(211) FFPE tissue microarrays (TMA, Table 24) were used as source of human (tumor and non tumor) as well as cynomolgus monkey (normal) tissues.

(212) TABLE-US-00026 TABLE 24 formalin-fixed and paraffin embedded tissue micro-arrays used as issue sources Reference Provider Description ASM221 Pantomics Cyno monkey, 22 organs, 22 samples CyFDA US Biomax Cyno monkey normal tissue microarray, 33 organs, taken from 6 normal individual (99 cases) COC1501 Pantomics Colon cancer tissue array, 150 cores from normal/benign (5 cases) and cancer (70 cases) tissues COC1502 Pantomics Colon cancer tissue array, 150 cores from normal/benign (5 cases) and cancer (70 cases) tissues COC1503 Pantomics Colon cancer tissue array, 150 cores from normal/benign (5 cases) and cancer (70 cases) tissues MTU951 Pantomics 40 types of tumors covering benign, malignant and metastatic entities of 27 anatomic sites LUC1501 Pantomics Lung cancer tissue array, 150 cores from normal/benign (5 cases) and cancer (70 cases ) tissues LUC1502 Pantomics Lung cancer tissue array, 150 cores from normal/benign (5 cases) and cancer (70 cases ) tissues LUC1503 Pantomics Lung cancer tissue array, 150 cores from normal/benign (5 cases) and cancer (70 cases ) tissues MN0961 Pantomics 35 types of normal tissues based on the FDA recommendation for antibody cross-reactivity testing. MN0661 Pantomics 33 types of normal tissues based on the FDA recommendation for antibody cross-reactivity testing. MN0341 Pantomics 33 types of normal tissues based on the FDA recommendation for antibody cross-reactivity testing. PAC481 Pantomics Pancreatic cancer tissue array contains 20 cases of cancers and 4 cases of normal and non-malignant pancreatic tissues CC4 Superbiochips 59 cores array including 59 cases of lung cancer A218(III) Accumax Esophagus cancer tissue array contains 40 cases of tumors and 8 non-neoplastic A219(II) Accumax Head&Neck cancer tissue array contains 45 cases of tumors and 8 non-neoplastic A213(II) Accumax Ovary cancer tissue array contains 43 cases of tumors and 8 non-neoplastic A301(IV) Accumax Various cancer tissues array with corresponding normal tissues (30 cancer cases, 30 non-neoplastic cases) A103(V) Accumax Various normal tissues array in duplicate (45 cases) MAN2 Superbiochips 59 cores array including 9 or 10 normal cases of stomach, esophagus, lung, colorectal, thyroid and kidney MA2 Superbiochip 59 cores array including 9 or 10 cases of stomach, sesophagus, lung, colorectal, thyroid and kidney cancers MBN4 Superbiochips 59 cores array including 9 or 10 normal cases of breast, liver, urinary bladder, ovary, pancreas, prostate MB4 Superbiochips 59 cores array including 9 or 10 cases of breast, liver, urinary bladder, ovary, pancreas, prostate cancers MCN4 Superbiochips 59 cores array including 9 or 10 normal cases of endometrium, gallbladder, larynx, uterine cervix, skin MC4 Superbiochips 59 cores array including 9 or 10 cases of endometrium, gallbladder, larynx, cervix, lymphoma, melanoma cancers CJ1 Superbiochips 59 cores array including 59 cases of ovary cancer CDN3 Superbiochips 59 cores array including 59 cases of normal colon and rectum CCN2 Superbiochips 59 cores array including 59 cases of normal lung (matching CC4) BB5 Superbiochips 60 cores, 30 human various cancer types AA9 Superbiochips 59 cores array including 59 cases of normal organs TMAhu3a Asterand Various cancer tissues array (76 cases) STC1501 Pantomics Stomach cancer tissue array, 150 cores including 75 cases of normal, reactive and cancerous tissues of the stomach STC1502 Pantomics Stomach cancer tissue array, 150 cores including 75 cases of normal, reactive and cancerous tissues of the stomach STC1503 Pantomics Stomach cancer tissue array, 150 cores including 75 cases of normal, reactive and cancerous tissues of the stomach STC481 Pantomics Stomach cancer tissue array, 16 cases, 48 cores, one normal paired with two tumor tissue cores from each patient

(213) Antibodies

(214) MAb2 was used as primary mouse anti-human CEACAM5 monoclonal antibody. A biotin-conjugated goat anti-mouse IgG1 (1 chain specific) (reference 1070-08, batch L4309-X761, Southern Biotech, USA) was used as secondary antibody.

(215) Immunostaining

(216) Antigen retrieval procedure was applied with Cell Conditioning 1 (CC1) buffer at 95 C. for 8 min and then at 100 C. for 28 min. After endogen biotins blocking step, slides were incubated with the primary anti-antibody diluted in phosphate buffer saline (PBS) at 5 g/mL for 2 hours at 24 C. The secondary antibody biotin-conjugated goat anti-mouse was incubated at 24 C. for 32 minutes at 0.5 g/mL. Immunostaining was done with 3,3-diaminobenzidine tetrahydrochloride (DAB) from DABMap chromogenic detection kit (760-124, Ventana Medical Systems, Inc, USA) according to manufacturer's recommendations. A couterstaining step was done with hematoxylin (760-2037, Ventana Medical Systems, Inc, USA) and bluing reagent was applied (760-2037, Ventana Medical Systems, Inc, USA). Stained slides were dehydrated and coverslipped with cytoseal XYL (8312-4, Richard-Allan Scientific, USA).

(217) IHC Scoring

(218) Immunostained slides were scanned using the ScanScope XT system (Aperio Technologies, Vista, Calif.). Digitized images were captured using the ImageScope software (version 10.2.2.2319, Aperio Technologies) at 20 magnification.

(219) Staining evaluation included the histologic site of reactivity, main type of reactive cell, staining intensity and cell staining frequency. The negative samples were scored as 0+. The positive samples were scored with a scale of intensity from 1+ to 4+. Ranges of intensities were described as weak ]0; 2+[, moderate [2+; 3+[and strong [3+; 4+]. Cell frequency was the percentage of immunostained cells and was estimated by the histologist observation as a median by sample. The cell frequency was ordered in 5 categories of proportion score: 1 (0-5%), 2 (6-25%), 3 (26-50%), 4 (51-75%) and 5 (76-100%).

(220) For tumors, a global expression score was adapted from the Allred score (AS) (Mohsin S, Weiss H, Havighurst T, Clark G M, Berardo M, Roanh L D, et al. Progesterone receptor by immunohistochemistry and clinical outcome in breast cancer: a validation study. Mod. Pathol. 2004; 17:1545-1554). This AS was obtained by adding the intensity and the proportion scores to obtain a total score that ranged from 0-9. The AS was reported as a percent of the maximum global score and ranged in 5 categories: very low (0-25%), weak (26-50%), moderate (51-75%), and high (76-100%). The prevalence was defined as the percent of positive cases for the indication

(221) Descriptive Statistical Analysis

(222) Descriptive statistics were calculated with Microsoft Excel 2003 software. For each indication, number of cases, positive cases number, prevalence, intensity score median, frequency median, Allred score mean, intensity range, frequency range and Allred score range were described.

Example 8.1

Use of an Anti-CEACAM5 Monoclonal Antibody for Evaluation of CEACAM5 Protein in FFPE Human Tumors by Immunohistochemistry (IHC)

(223) Large panel of human tumors were studied using commercial tissue array slides (FFPE format). Expression of CEACAM5 protein was located in membrane+/cytoplasm of tumor cells (FIG. 1C, D). Some membrane staining was polarized at apical pole of cells in the more differentiated tumors. CEACAM5 protein was found to be expressed in: 89% of colon adenocarcinoma cases (194/219, intensity 2-2.5+, frequency 53-59%, AS 60-66%) 49% of stomach adenocarcinoma cases (95/195, intensity 2.5+, frequency 53%, AS 62%) 41% of lung adenocarcinoma cases (24/58, intensity 1.8-2+, frequency 50-53%, AS 54-58%) 79% of uterus cervix squamous carcinoma cases (11/14, intensity 2+, frequency 22%, AS 46%) 53% of pancreas adenocarcinoma cases (18/34, intensity 2+, frequency 23%, AS 42%) 37% of esophagus squamous cell carcinoma cases (23/62, intensity 2+, frequency 16%, AS 38%) 4% of ovary carcinoma cases (3/77, intensity 2+, frequency 43%, AS 54%) 11% of thyroid carcinoma cases (2/18, intensity 1.5+, frequency 63%, AS 56%) 25% of bladder carcinoma cases (5/20, intensity 1.5+, frequency 61%, AS 56%) 7% of endometrium adenocarcinoma cases (1/14, intensity 2+, frequency 50%, AS 56%) 11% of breast ductal carcinoma cases (2/18, intensity 1.5+, frequency 53%, AS 50%) 53% of cholangiocarcinoma cases (2/6, intensity 1.5+, frequency 75%, AS 50%) 53% of lung squamous cell carcinoma cases (31/148, intensity 1.5+, frequency 22%, AS 39%) 8% of prostate adenocarcinoma cases (1/13, intensity 2+, frequency 50%, AS 44%) 25% of skin squamous carcinoma cases (2/8, intensity 1.5+, frequency 23%, AS 39%)

Example 8.2

Tissue Cross-Reactivity of an Anti-CEACAM5 Monoclonal Antibody in Cynomolgus Monkey (Macaca fascicularis) and Comparison with Human Expression Pattern

(224) The extracellular protein domain of CEACAM5 from human (h) or cynomolgus monkey (c) origin have been prepared by transient expression in human embryonic kidney HEK293 cells with CEACAM5 cDNA plasmid (example 1, Table 1). Cell pellets were fixed in 10% formalin (Sigma Aldrich, USA) for 16 hours, and embedded in paraffin as a piece of tissue according to standard histological procedure.

(225) Commercial TMA were used as human and monkey normal tissues source (Table 21).

(226) Crossreactivity of Mab2 was shown by immunostaining in both human and monkey CEACAM5 transfected cells (membrane and cytoplasm localization).

(227) In cynomolgus normal tissues, CEACAM5 protein expression was found in columnar absorptive epithelial cells (2/3 positive cases, median intensity 1.5+, mean frequency 55%).

(228) In human non tumor tissues, CEACAM5 expression was also observed in columnar absorptive epithelial cells (62/64 positive cases, median intensity 2+, mean frequency 90%). In human tissues, CEACAM5 expression was observed in less extent in esophagus epithelial cells, head&neck epithelial cells, stomach gastric pit epithelial cells and uterus cervix epithelial cells.

Example 9

Antibody Drug Conjugate (Variant)

(229) AntiCEACAM5 huMAb2-3-sulfoSPDB-DM4

(230) Analytical Data: MW(Ab)=147417 g/mol; MW(DM4)=780.38 g/mol .sub.280nm(Ab)=201400; .sub.252nm(Ab)=71451 .sub.280nm(DM4)=5180; .sub.252nm(DM4)=26159

(231) Under stirring, at RT, 7.0 ml of a solution of antiCEACAM5 huMAb2-3 (C=5.32 mg/ml in PBS pH=7.4 buffer) are introduced in a vessel, followed by 1.6 ml of DMA and 168.4 l of nitro-sulfoSPDB linker (described in WO2009134977) solution (10 Eq-15 mM solution in DMA). Solution is slowly stirred at RT for 3 hours. An extra volume of 3.4 l of nitro-sulfoSPDB linker solution (2.0 Eq-15 mM solution in DMA) is added. After 2 hours at RT under magnetic stirring, 2.90 ml of PBS pH 7.4 buffer, 0.407 ml of DMA and 0.322 ml of DM4 solution (15 mM solution in DMA) were sucessively added. After 1 hour at RT, and 16 hours at 5 C., crude reaction mixture is purified on HiPrep 26/10 desalting column (Sephadex G25, GE Healthcare), pre-conditioned with 1CV of NaOH 1M, 2 CV of water and 2 CV of histidine (10 mM), glycine (130 mM), sucrose (5%), pH=5.5 buffer. Conjugate is eluted with histidine (10 mM), glycine (130 mM), sucrose (5%), pH=5.5 buffer, and monomeric conjugate fractions are pooled and filtered on 0.22 m filter.

(232) 19 ml of antiCEACAM5 huMAb2-3-sulfoSPDB-DM4 conjugate (c=1.51 mg/ml) was thus obtained as a colorless clear solution. The conjugate is then analyzed for final drug load and monomeric purity: DAR (UV)=3.4; DAR (SEC)=3.3; monomeric purity=99.8%; HRMS data: see FIG. 20.

(233) AntiCEACAM5 huMAb2-3-SMCC-DM1

(234) Analytical Data: MW(Ab)=147417 g/mol; MW(DM1)=738 g/mol .sub.280nm(Ab)=201400; .sub.252nm(Ab)=71451 .sub.280nm(DM4)=5180; .sub.252nm(DM4)=26159

(235) Under stirring, at RT, 11.3 ml of a solution of antiCEACAM5 huMAb2-3 (C=3.47 mg/ml in buffer A pH=6.5) are introduced in a vessel, followed by 0.387 ml of DMA and 178 l of SMCC linker solution (10 Eq-15 mM solution in DMA). Solution is slowly stirred at RT for 2 hours. Crude reaction mixture is buffer exchanged on HiPrep 26/10 desalting column (Sephadex G25, GE Healthcare), pre-conditioned with 2CV of NaOH 0.2M, 5 CV of water and 5 CV of citrate buffer (pH 5.5). Conjugate is eluted with citrate buffer (pH 5.5) and monomeric conjugate fractions are pooled and filtered on 0.22 m filter. To this solution are sucessively added, under stirring, at RT, 0.476 ml of DMA and 0.124 ml of DM1 solution (15 mM solution in DMA). After 2 hours at RT, crude reaction mixture is purified twice on HiPrep 26/10 desalting column (Sephadex G25, GE Healthcare), pre-conditioned with 2CV of NaOH 0.2M, 5 CV of water and 5 CV of histidine (10 mM), glycine (130 mM), sucrose (5%), pH=5.5 buffer. Conjugate is eluted with histidine (10 mM), glycine (130 mM), sucrose (5%), pH=5.5 buffer, and monomeric conjugate fractions are pooled, filtered on 0.22 m filter.

(236) 9.5 ml of antiCEACAM5 huMAb2-3-SMCC-DM1 (c=1.73 mg/ml) was thus obtained as a colorless clear solution. The conjugate is then analyzed for final drug load and monomeric purity: DAR (UV)=2.7; DAR (SEC)=2.9; monomeric purity=99.6%; HRMS data: see FIG. 21.

Example 10

Characterization of the Epitope and of the Paratope of

(237) CEACAM5-A3B3 in complex with MAb2_VH1aVL1c Fab using hydrogen-deuterium exchange associated with mass spectrometry (HDX MS)

Example 10.1

Principle of HDX MS

(238) Amide hydrogen-deuterium exchange (HDX) associated with mass spectrometry (MS) enables identification of regions of proteins implied in conformational changes or interactions. This technique enables more specifically to identify the regions of an antigen showing, after incubation in a deuterated buffer and proteolysis, a decrease of deuterium incorporation in its form bound to an antibody compared to its free form.

(239) The epitope belongs to these regions, the exchange of which is slowed down by the binding to the antibody. A recent article describes in detail the different steps to characterize epitopes using this approach (Zhang, Q., Willison, L. N., Tripathi, P., Sathe, S. K., Roux, K. H., Emmett, M. R., Blakney, G. T., Zhang, H. M. & Marshall, A. G. (2011). Analytical Chemistry 83, 7129-7136.).

Example 10.2

Materials

(240) The variable domain coding sequences of MAb2_VH1aVL1c (SEQ ID NO:5 and SEQ ID NO:29) were cloned into a mammalian expression vector in fusion with the coding sequences of human CH1 domain (as found in papain cleaved IgG1 derived Fabs), followed by an hexa-Histidine tag or with the human Ckappa constant domain, respectively. A batch of MAb2_VH1aVL1c Fab was produced in suspension-cultivated HEK293-FS cells by transient transfection of two expression plasmids, encoding the two chains, complexed with 293Fectin (Invitrogen). Culture supernatant containing the secreted protein was harvested seven days post-transfection, centrifuged and filtered on 0.22 m membrane. The Fab was purified by affinity chromatography on IMAC (HisTrap, GE Healthcare) using imidazole gradient in PBS. Then, the pool of fractions containing the Fab was purified by size exclusion chromatography (Superdex 200, GE Healthcare) equilibrated with PBS.

(241) His-tagged hCEACAM5-A3B3 domain (SEQ ID NO:67) was produced with HEK293-FS cells cultivated in flask by transient transfection of expression plasmid. Kifunensine (inhibitor of trimming glycosylation process) was added each day. Culture supernatant containing the secreted protein was harvested seven days post-transfection, centrifuged and filtered on 0.22 m membrane. EndoH was added into supernatant up to 625 u/ml then incubated 3 h at 37 C. Deglycosylated hCEACAM5-A3B3 was purified by affinity chromatography on IMAC (HisTrap, GE Healthcare) using imidazole gradient in PBS. Then, the pool of fractions containing deglycosylated hCEACAM5-A3B3 was purified by size exclusion chromatography (Superdex 200, GE Healthcare) equilibrated with PBS. Mass spectrometry analysis of deglycosylated hCEACAM5-A3B3 showed two species (22 485 and 22 278 Da), indicating that the protein carries 7 or 8 N-acetylglucosamine residues (GlcNAc).

(242) To build a complex, both proteins, were pooled with an excess of 1.5 moles of deglycosylated hCEACAM5-A3B3 for one mole of Fab. This excess was removed by size exclusion chromatography on superdex 200 equilibrated with phosphate buffer saline. Fractions corresponding to complex Fab with the antigen were used for deuterium exchange study.

Example 10.3

Methods

(243) Hydrogen/deuterium exchange (HDX) experiments were fully automated using a PAL autosampler (CTC Analytics). It enabled exchange start and quench, control of proteolysis temperature (4 C.), injection of the deuterated peptides, management of the injection and washing valves and triggering acquisition of the mass spectrometer and HPLC pumps. A Peltier-cooled box (4 C.) contained two Rheodyne automated valves (6-port for injection and 10-port for washing), a desalting cartridge (peptide Micro Trap from Bruker-Michrom) and a HPLC column (Poroshell 120 EC-C18, 150 mm, 2.7 M from Agilent Technologies). Deuteration was initiated by a 5-fold dilution of CEACAM5, mAb or complex with PBS in D.sub.2O. 2M GndHCl, 0.8 M TCEP, 1 M glycine was used to quench back-exchange and reduce the disulfide bridges for 2 min at 4 C.

(244) The proteins were digested with pepsin and nepenthesin proteases and the peptides were desalted using an Agilent Technologies HPLC pump with TFA 0.03% in water at 100 L/min. The peptides were then separated using another Agilent Technologies HPLC pump with a 15-100% B gradient in 20 min (A: TFA 0.03% in water; B: acetonitrile 90%, TFA 0.03% in water). The peptides masses were measured using an electrospray-TOF mass spectrometer (Agilent 6210).

(245) The peptides were identified by tandem MS (MSMS), using a Bruker APEX-Q FTMS (9.4 T) and a Bruker 12 T SolariX.

(246) Data Analysis (Bruker) and Mass Hunter (Agilent Technologies) softwares were used for data acquisitions. Data Analysis and Mascot (Matrix Science) were used to process the MSMS data. Mass Hunter and HD Examiner (Sierra Analytics) softwares were used for HDX data processing.

(247) The HDX experiments were repeated at least three times.

Example 10.4

Results

(248) Identification and Selection of the Peptides

(249) The disulfide bridges remained intact during the deuteration to keep the structural information related to them. To favor proteolysis and peptides identification the bridges were reduced with TCEP after the quench step at low pH and low temperature. Using MSMS after digestion of the CEACAM5-Fab complex it was possible to identify a large number of peptides arising from the three protein chains. After the HDX experiments only the ones giving good quality signals were selected: 25, 30 and 20 peptides from the CEACAM5-A3-B3 antigen, MAb2_VH1aVL1c Fab heavy chain and MAb2_VH1aVL1c light chain, respectively. These peptides cover 89%, 77% and 68% of CEACAM5-A3-B3 antigen, MAb2_VH1aVL1c Fab heavy chain and MAb2_VH1aVL1c light chain sequences, respectively (Table 25). The uncovered regions of the Fab chains are mainly in their C-terminal parts.

(250) TABLE-US-00027 TABLE 25 sequence coverage with deuterated peptides Peptides Sequence coverage CEACAM5-A3-B3 1-18; 1-22; 1-23; 1-19; 23-35; 36-51; 35-49; 50-70; 36-43; 44-51; 36-51; 36-49; 50-67; 37-49; 44-49; 59-67; 71-89; 93-107; 108-115; 128-143; 128-142; 143-157; 130-143; 130-142; 140-143; 163-186 MAb2_VH1aVL1c 1-6; 1-20; 1-19;1-17; 1-18; 4-18; 5-20; Fab heavy chain 5-18; 24-29; 27-32; 27-29; 34-46; 47-68; 48-68; 50-68; 69-86; 84-93; 88-98; 92-104; 100-109; 110-115; 116-136; 111-128; 149-158; 151-158; 159-177; 162-177; 167-177; 187-206 MAb2_VH1aVL1c 1-11; 5-11; 22-46; 47-54; 55-70; 55-71; light chain 72-82; 87-104; 105-115; 117-132; 124-131; 127-145; 133-144; 136-145; 136-143; 136-144; 143-161; 144-151; 146-151

(251) All the 8 asparagine residues that are potential sites of glycosylations were identified within several peptides with an GlcNAc remaining from the endo H deglycosylation. In particular, N114 was found in peptide 108-115. In first experiments (not used for HDX), N166 was found in both forms (with and without GlcNAc). It might explain the heterogeneity observed in the mass spectrum of CEACAM5-A3B3 after deglycosylation, corresponding to 7 and 8 GlcNAc.

(252) Epitope and Paratope Identification

(253) The free antigen, the free Fab and their complex were deuterated during 2 min or 20 min at 4 C. or 20 min at room temperature (26 C.). Considering the exchange kinetics of amide hydrogens with temperature (about 3-fold exchange increase with 10 C.) the last condition is equivalent to 200 min deuteration at 4 C.

(254) Epitope

(255) The kinetics of deuterium incorporation for the 25 selected peptides of CEACAM5-A3B3 were compared when the antigen was deuterated in the free form and when it was in complex with the Fab. Several peptides did not show any significant HDX difference (HDX) between both states. In contrast some of them (108-115 and 128-143), showed significant HDX. The second region was covered with 5 different peptides: 128-142, 128-143, 130-142, 130-143 and 140-143 showing 13-152% (up to 1.60.2 D) HDX after 2 min deuteration.

(256) Comparing 128-142 with 130-142 and 128-143 with 130-143, we did not measure any significant HDX change in each case (1.3-1.4 D for the first two peptides and 1.6 for the last two, after 2 min deuteration), meaning that the amides W129 and R130 are likely not involved in the epitope. In contrast, comparing 128-142 with 128-143 and 130-142 with 130-143, we measured a small HDX change (about 0.2 D), meaning that the amide F143 is involved. The HDX in peptide 140-143 (about 0.3 D) indicates that amides V141 or L142 might be also involved. Within the 9 amides from 1131 to Q140, several of them are involved in the epitope (about 1 HDX shared on average).

(257) These differences of deuterium incorporation indicate that the epitope belongs in particular to regions (amides), i.e. peptides of sequences SGANLNL (SEQ ID NO: 76) and INGIPQQHTQVLF (SEQ ID NO: 77).

(258) Paratope

(259) The kinetics of deuterium incorporation for the 30 selected peptides of the Fab heavy chain were compared when the Fab was deuterated in the free form and when it was in complex with the antigen. Almost all peptides did not show any significant HDX between both states. Only one peptide (100-109) presented a HDX after 200 min deuteration: 112% (0.70.2 D). The region (amides) 101-109 of MAb2_VH1aVL1c Fab heavy chain is implied in the paratope.

(260) The kinetics of deuterium incorporation for the 20 selected peptides of the Fab light chain were compared when the Fab was deuterated in the free form and when it was in complex with the antigen. Almost all peptides did not show any significant HDX between both states. Only two peptides (47-54 and 87-104) presented a difference. After 20 min deuteration, it was 102% (0.60.2 D) for the first one and 52% (0.90.2 D) for the second one, respectively. The regions 48-54 and 88-104 of MAb2_VH1aVL1c light chain are involved in the paratope.