Antibody constructs for CLDN18.2 and CD3

Abstract

The present invention relates to an antibody construct comprising a domain which binds to Claudin 18.2 (CLDN18.2) and another domain which binds to CD3. Moreover, the invention provides a polynucleotide encoding the antibody construct, a vector comprising said polynucleotide and a host cell transformed or transfected with said polynucleotide or vector. Furthermore, the invention provides a process for producing the antibody construct of the invention, a medical use of said antibody construct and a kit comprising said antibody construct.

Claims

1. An antibody construct comprising a first domain which binds Claudin 18.2 (CLDN18.2), and a second domain which binds human CD3, wherein the first domain comprises a VH region comprising CDR-H1, CDR-H2 and CDR-H3 and a VL region comprising CDR-L1, CDR-L2 and CDR-L3 selected from the group consisting of: a) CDR-H1 as depicted in SEQ ID NO: 121, CDR-H2 as depicted in SEQ ID NO: 122, and CDR-H3 as depicted in SEQ ID NO: 123, CDR-L1 as depicted in SEQ ID NO: 124, CDR-L2 as depicted in SEQ ID NO: 125 and CDR-L3 as depicted in SEQ ID NO: 126; and b) CDR-H1 as depicted in SEQ ID NO: 133, CDR-H2 as depicted in SEQ ID NO: 134, and CDR-H3 as depicted in SEQ ID NO: 135, and CDR-L1 as depicted in SEQ ID NO: 136, CDR-L2 as depicted in SEQ ID NO: 137 and CDR-L3 as depicted in SEQ ID NO: 138.

2. The antibody construct according to claim 1, wherein the first domain comprises a VH region having an amino acid sequence as depicted in SEQ ID NO: 127 or SEQ ID NO: 139.

3. The antibody construct according to claim 1, wherein the first domain comprises a VL region having an amino acid sequence as depicted in SEQ ID NO: 128 or SEQ ID NO: 140.

4. The antibody construct according to claim 1, wherein the first domain comprises a VH region and a VL region having an amino acid sequence as depicted in SEQ ID NOs: 127 and 128 or SEQ ID NOs: 139 and 140.

5. The antibody construct according to claim 1, wherein the first domain comprises a polypeptide having an amino acid sequence as depicted in SEQ ID NO: 129 or SEQ ID NO: 141.

6. The antibody construct according to claim 1, comprising or consisting of a polypeptide having an amino acid sequence of SEQ ID NO: 131, SEQ ID NO: 132, SEQ ID NO: 143, or SEQ ID NO: 144.

7. A polynucleotide encoding the antibody construct of claim 1.

8. A vector comprising the polynucleotide of claim 7.

9. A host cell transformed or transfected with the polynucleotide of claim 7.

10. A process for producing an antibody construct, said process comprising culturing the host cell of claim 9 under conditions allowing the expression of said antibody construct and recovering the produced antibody construct from the culture.

11. A composition comprising the antibody construct of claim 1 and a carrier, stabilizer, excipient, diluent, solubilizer, surfactant, emulsifier, preservative and/or adjuvant.

12. A method of treating or ameliorating a CLDN18.2-expressing disease or neoplasm, comprising administering an effective amount of the antibody construct of claim 1.

13. The method of claim 12, wherein the disease or neoplasm is gastrointestinal cancer, ovarian cancer or lung cancer.

14. The method of claim 13, wherein the gastrointestinal cancer is gastric cancer, esophageal cancer, gastroesophageal cancer, pancreatic cancer, or colorectal cancer.

15. A kit comprising the antibody construct of claim 1 and a recipient.

16. The antibody construct according to claim 1, wherein the second domain binds to human CD3 epsilon and to Callithrix jacchus or Saimiri sciureus CD3 epsilon.

17. The antibody construct according to claim 1, wherein a) the antibody construct is a single chain antibody construct; b) the first domain is in the format of an scFv; c) the second domain is in the format of an scFv; d) the first and the second domain are connected via a linker; and/or e) the antibody construct further comprises a domain providing an extended serum half-life.

18. The antibody construct according to claim 1, wherein the first domain does not bind or does not significantly bind to CLDN18.1, CLDN1, CLDN2, CLDN3, CLDN4, CLDN6, and/or CLDN9.

Description

(1) The Figures show:

(2) FIG. 1

(3) Alignment of the human CLDN18.1 and CLDN18.2 amino acid sequence. The first and the second extracellular domains (=extracellular loops) are highlighted, as well as the amino acid positions differing between CLDN18.1 and CLDN18.2. Within the extracellular loop 1, CLDN18.2 and CLDN18.1 differ in eight positions. See also Example 1.

(4) FIG. 2

(5) The figure depicts the CLDN18.2 constructs (chimeras/point mutations) that were generated for the epitope mapping analysis of Example 2. It also shows the sequence alignment between human CLDN18.1 loop 1 and human CLDN18.2 loop 1 and highlights the eight positions P1-P8 in which these two molecules differ. See also Example 1.

(6) FIG. 3

(7) Results of epitope mapping analysis, see Example 2. The figure shows FACS analyses of untransfected CHO cells, as well as of CHO cells transfected with hu CLDN18.1, hu CLDN18.2, hu CLDN6 and hu CLDN9 (left-hand side). On the right hand side, the figure shows FACS analyses of CHO cells transfected with three different chimeric hu CLDN18.2 constructs: The amino acid sequence of the entire loop 2 (ECL2, E2) was exchanged for a sequence originating from human CLDN9, and the regions E2A and E2B were exchanged for a counterpart sequence of human CLDN6. All antibodies/antibody constructs were tested at a concentration of 5 μg/ml.

(8) FIG. 4

(9) Results of epitope mapping analysis, see Example 2. The figure shows FACS analyses of untransfected CHO cells, as well as of CHO cells transfected with hu CLDN18.1, hu CLDN18.2, hu CLDN6 and hu CLDN9 Furthermore, the figure shows FACS analyses of CHO cells transfected with eleven different hu CLDN18.2 constructs having one, two or three point mutations at the indicated positions P1-P8 (for further details, see also Example 1), as well as of CHO cells transfected with four different chimeric hu CLDN18.2 constructs: The amino acid sequence of the entire loop 2 (ECL2, E2) was exchanged for a sequence originating from human CLDN9, and the regions E1B, E1D and E1C were exchanged for a counterpart sequence of human CLDN6 (for further details, see also Example 1). All monospecific antibodies were tested at a concentration of 5 μg/ml, while the CD3×CLDN18.2 bispecific antibody constructs were tested at a concentration of 20 μg/ml.

(10) Row 1: in-house α-hu CLDN-18.1

(11) Row 2: in-house α-hu CLDN-18.2

(12) Row 3: α-hu CLDN-6 (R&D; MAB3656)

(13) Row 4: α-hu CLDN-9 (ABIN1720917)

(14) Row 5: CL-1×I2C-scFc

(15) Row 6: CL-2×I2C-scFc

(16) Row 7: CL-4×I2C-scFc

(17) Row 8: CL-3×I2C-scFc

(18) FIG. 5

(19) Results of the FACS assay described in Example 5.

(20) FIG. 6

(21) Results of the FACS assay described in Example 6.

(22) FIG. 7

(23) Results of the FACS-based cytotoxicity assays described in Example 7.4. In all six cell lines tested (five natural expresser cell lines and CHO cells transfected with hu CLDN18.2), the antibody constructs of the invention (CL-1 and CL-2) showed significantly higher EC50 values compared with the control constructs binding to a different CLDN18.2 epitope (CL-3 and CL-4).

(24) FIG. 8

(25) Anti-tumor activity of a CLDN18.2×CD3 antibody construct (mean tumor volume with standard error of mean (SEM)). See Example 13.

(26) FIG. 9

(27) Anti-tumor formation activity of a CLDN18.2×CD3 antibody construct (mean tumor volume with standard error of mean (SEM)). See Example 14.

(28) As used herein, the singular forms “a”, “an”, and “the” include plural references unless the context clearly indicates otherwise. Thus, for example, reference to “a reagent” includes one or more of such different reagents and reference to “the method” includes reference to equivalent steps and methods known to those of ordinary skill in the art that could be modified or substituted for the methods described herein.

(29) Unless otherwise indicated, the term “at least” preceding a series of elements is to be understood to refer to every element in the series. Those skilled in the art will recognize, or be able to ascertain using no more than routine experimentation, many equivalents to the specific embodiments of the invention described herein. Such equivalents are intended to be encompassed by the present invention.

(30) The term “and/or” wherever used herein includes the meaning of “and”, “or” and “all or any other combination of the elements connected by said term”.

(31) The term “about” or “approximately” as used herein means within ±20%, preferably within ±15%, more preferably within ±10%, and most preferably within ±5% of a given value or range. It also includes the concrete value, e.g., “about 50” includes the value “50”.

(32) Throughout this specification and the claims, unless the context requires otherwise, the word “comprise”, and variations such as “comprises” and “comprising”, will be understood to imply the inclusion of a stated integer or step or group of integers or steps but not the exclusion of any other integer or step or group of integer or step. When used herein the term “comprising” can be substituted with the term “containing” or “including” or sometimes when used herein with the term “having”.

(33) When used herein “consisting of” excludes any element, step, or ingredient not specified in the claim element. When used herein, “consisting essentially of” does not exclude materials or steps that do not materially affect the basic and novel characteristics of the claim.

(34) In each instance herein, any of the terms “comprising”, “consisting essentially of” and “consisting of” may be replaced with either of the other two terms.

(35) It should be understood that the above description and the below examples provide exemplary arrangements, but the present invention is not limited to the particular methodologies, techniques, protocols, material, reagents, substances, etc., described herein and as such can vary. The terminology used herein is for the purpose of describing particular embodiments only, and is not intended to limit the scope of the present invention, which is defined solely by the claims. Aspects of the invention are provided in the independent claims. Some optional features of the invention are provided in the dependent claims.

(36) All publications and patents cited throughout the text of this specification (including all patents, patent applications, scientific publications, manufacturer's specifications, instructions, etc.), whether supra or infra, are hereby incorporated by reference in their entirety. Nothing herein is to be construed as an admission that the invention is not entitled to antedate such disclosure by virtue of prior invention. To the extent the material incorporated by reference contradicts or is inconsistent with this specification, the specification will supersede any such material.

(37) A better understanding of the present invention and of its advantages will be obtained from the following examples, offered for illustrative purposes only. The examples are not intended and should not be construed as to limit the scope of the present invention in any way.

Example 1

(38) Generation of CHO Cells Expressing CLDN18.2 Mutations

(39) For the purposes of epitope mapping, E1 (extracellular loop 1; ECL1, SEQ ID NO: 2) of CLDN18.2 was divided into four sub-domains (E1A, E1B, E1C and E1D), and E2 (extracellular loop 2; ECL2, SEQ ID NO: 3) was divided into two sub-domains (E2A and E2B). The amino acid sequence of the respective epitope region (loop/domain or sub-domain) of human CLDN18.2 (E1, E1A, E1B, E1C, E1D, E2, E2A and E2B) was exchanged for a counterpart sequence of human CLDN6, with two exceptions (see FIGS. 1 and 2): E2 was exchanged for a sequence originating from human CLDN9; and in the construct having a mutated E1C (exchange of CLDN18.2 E1C for the counterpart CLDN6 region), E2 of CLDN18.2 was also exchanged for a sequence originating from human CLDN9. The latter served as a means to detect expression of the CLDN18.2/CLDN6-chimera. The expression of all chimeric constructs in CHO cells was verified via FACS analysis (see below). As there was no proof of expression for the E1 and the E1A chimeric constructs, these constructs were not used for the epitope mapping analysis. The following constructs were hence generated and used for epitope mapping: CLDN18.2-E1B (CLDN6).fwdarw.SEQ ID NO: 22 CLDN18.2-E1C (CLDN6)-E2 (CLDN9).fwdarw.SEQ ID NO: 23 CLDN18.2-E1D (CLDN6).fwdarw.SEQ ID NO: 24 CLDN18.2-E2 (CLDN9).fwdarw.SEQ ID NO: 25 CLDN18.2-E2A (CLDN6).fwdarw.SEQ ID NO: 26 CLDN18.2-E2B (CLDN6).fwdarw.SEQ ID NO: 27

(40) The amino acid sequences of human CLDN18.2-ECL2 and human CLDN18.1-ECL2 are identical, but within extracellular loop 1, CLDN18.2 and CLDN18.1 differ in eight positions (29, 37, 42, 45, 47, 56, 65 and 69). Therefore, additional CLDN18.2 mutants were generated and expressed in CHO cells, in which these eight CLDN18.2 positions “P1” to “P8” were exchanged by their respective CLDN18.1 counterparts, either individually or in a group of two or three positions. The following constructs were hence generated: CLDN18.2-P1-CLDN18.1 (Q29M mutation).fwdarw.SEQ ID NO: 11 CLDN18.2-P2-CLDN18.1 (N37D mutation).fwdarw.SEQ ID NO: 12 CLDN18.2-P1/P2-CLDN18.1 (Q29M/N37D mutation).fwdarw.SEQ ID NO: 13 CLDN18.2-P3-CLDN18.1 (A42S mutation).fwdarw.SEQ ID NO: 14 CLDN18.2-P4-CLDN18.1 (N45Q mutation).fwdarw.SEQ ID NO: 15 CLDN18.2-P5-CLDN18.1 (Q47E mutation).fwdarw.SEQ ID NO: 16 CLDN18.2-P3/P4/P5-CLDN18.1 (A42S/N45Q/Q47E mutation).fwdarw.SEQ ID NO: 17 CLDN18.2-P6-CLDN18.1 (E56Q mutation).fwdarw.SEQ ID NO: 18 CLDN18.2-P7-CLDN18.1 (G65P mutation).fwdarw.SEQ ID NO: 19 CLDN18.2-P8-CLDN18.1 (L69I mutation).fwdarw.SEQ ID NO: 20 CLDN18.2-P7/P8-CLDN18.1 (G65P/L69I mutation).fwdarw.SEQ ID NO: 21

(41) For the generation of CHO cells expressing the above constructs, as well as of CHO cells expressing hu-CLDN18.2, hu-CLDN18.1, hu-CLDN6 and hu-CLDN9 (SEQ ID NOs: 1, 4, 9 and 10) as controls, the respective coding sequences were cloned into a plasmid designated pEF-DHFR (pEF-DHFR is described in Raum et al. Cancer Immunol Immunother 50 (2001) 141-150). All cloning procedures were carried out according to standard protocols (Sambrook, Molecular Cloning; A Laboratory Manual, 3rd edition, Cold Spring Harbour Laboratory Press, Cold Spring Harbour, N.Y. (2001)). For each construct, a corresponding plasmid was transfected into DHFR deficient CHO cells for eukaryotic expression, as described by Kaufman R. J. (1990) Methods Enzymol. 185, 537-566.

(42) The expression of the above constructs on CHO cells was verified in a FACS assay using antibodies against CLDN18.2 (in-house monoclonal anti-hu CLDN18.2 antibody), CLDN18.1 (in-house monoclonal anti-hu CLDN18.1 antibody), CLDN6 (R&D mouse anti-human CLDN6 monoclonal antibody MAB3656) and CLDN9 (rat anti-human CLDN9 monoclonal antibody ABIN1720917), respectively, at a concentration of 5 μg/ml. As negative control, cells were incubated with an isotype control antibody (BD 553454/R&D MAB0041/R&D MAB0061) instead of the first antibody. Bound monoclonal antibody was detected with a secondary anti-mouse/anti-rat/anti-human IgG Fc-gamma-PE (Jackson ImmunoResearch 115-116-071/112-116-071/109-116-098). The samples were measured by flow cytometry.

Example 2

(43) Epitope Mapping of Anti-CLDN18.2 Antibody Constructs

(44) CHO cells transfected with the constructs described in Example 1 were stained with purified CLDN18.2×CD3 antibody constructs at a concentration of 20 μg/ml. Bound antibody constructs were detected with an anti-human IgG Fc-gamma-PE (Jackson ImmunoResearch; 1:100). All antibodies were diluted in PBS/2% FCS. As negative control, cells were incubated with PBS/2% FCS followed by the anti-human IgG Fc-gamma-PE. The samples were measured by flow cytometry.

(45) The results of the epitope mapping analysis are shown in FIGS. 3 and 4. When a loss of the FACS signal is observed for cells expressing a certain CLDN18.2 chimera or mutation, the respective CLDN18.2×CD3 antibody construct is assumed to bind to the epitope (loop/domain/sub-domain) or to the specific amino acid that was exchanged in this CLDN18.2 chimeric or mutated molecule. In other words, this epitope region or amino acid is required for the binding of the CLDN18.2×CD3 antibody construct that was analyzed. In addition to the control antibodies which were used to demonstrate proper expression of the respective target, the following CLDN18.2×CD3 antibody constructs were specifically tested in the epitope mapping analysis: CL-1×I2C-scFc (SEQ ID NO: 132) CL-2×I2C-scFc (SEQ ID NO: 144) CL-3×I2C-scFc (SEQ ID NO: 149) CL-4×I2C-scFc (SEQ ID NO: 154)

(46) While CL-1×I2C-scFc and CL-2×I2C-scFc are antibody constructs according to the invention, CL-3×I2C-scFc and CL-4×I2C-scFc have anti-CLDN18.2 VH and VL regions that are disclosed as SEQ ID NOs: 8+15 and SEQ ID NOs: 6+11, respectively, of WO 2014/075788.

(47) As shown in FIG. 4, the CL-3×I2C-scFc antibody construct clearly requires positions P3 (A42), P4 (N45) and P6 (E56) for its specific binding to CLDN18.2. Position P4 is located within the sub-domain denominated E1B, and position P6 is located within the sub-domain denominated E1C. Consequently and likewise, an exchange of these sub-domains with the CLDN6 counterpart sequence leads to a loss of the FACS signal. The observation that these three positions are relevant for the binding of CL-3×I2C-scFc to CLDN18.2 confirms previously published results. The antibody construct denominated CL-4×I2C-scFc has a very similar binding pattern (see FIG. 4).

(48) In contrast, both antibody constructs CL-1×I2C-scFc and CL-2×I2C-scFc clearly require position P6 (E56) for their specific binding to CLDN18.2. However, the exchange of other positions, in particular of P3 and P4, does not appear to have any impact on the binding of CL-1×I2C-scFc or CL-2×I2C-scFc to CLDN18.2. In line with this observation, the exchange of sub-domain E1C (in which position P6 is located) with the CLDN6 counterpart sequence—but not the exchange of E1B or E1D—leads to a loss of the FACS signal. The epitope mapping result depicted in FIG. 4 hence shows that CL-1×I2C-scFc and CL-2×I2C-scFc bind to the same epitope within ECL1 of CLDN18.2, and this epitope differs from the one of CL-3×I2C-scFc and CL-4×I2C-scFc.

Example 3

(49) Biacore-Based Determination of Affinity to Human and Cynomolgus CD3 and FcRn

(50) Biacore analysis experiments were performed using recombinant human/macaque CD3-ECD (ECD=extracellular domain) fusion proteins with chicken albumin to determine target binding of the antibody constructs of the invention.

(51) In detail, CM5 Sensor Chips (GE Healthcare) were immobilized with approximately 600-800 RU of the respective recombinant antigen using acetate buffer pH 4.5 according to the manufacturer's manual. The CLDN18.2×CD3 antibody construct was loaded in a dilution series of the following concentrations: 50 nM, 25 nM, 12.5 nM, 6.25 nM and 3.13 nM diluted in HBS-EP running buffer (GE Healthcare). Flow rate was 30 μl/min for 3 min, then HBS-EP running buffer was applied for 8 min to 20 min again at a flow rate of 30 μl/ml. Regeneration of the chip was performed using 10 mM glycine 10 mM NaCl pH 1.5 solution. Data sets were analyzed using BiaEval Software. In general, two independent experiments were performed.

(52) The CLDN18.2×CD3 antibody constructs according to the invention showed high affinities to human CD3 in the nanomolar range. Binding to macaque CD3 was balanced, also showing affinities in similar ranges. The affinity values as well as the calculated affinity gap are shown in Table 2.

(53) TABLE-US-00002 TABLE 2 Affinities of CLDN18.2 × CD3 antibody constructs to human and macaque CD3 as determined by Biacore analysis, as well as the calculated interspecies affinity gaps. The constructs in rows 3-5 were measured in a different (separate) assay than the constructs in rows 1 and 2. CLDN18.2 × CD3 KD hu CD3 KD cyno CD3 Affinity gap antibody construct [nM] [nM] KD mac/KD hu 1 CL-1 × I2C-scFc 23.5 ± 1.0 19.7 ± 1.1  0.84 2 CL-1 × I2C-6His  4.0 ± 0.2 3.4 ± 0.2 0.85 3 CL-2 × I2C-scFc  8.32 ± 0.64 4.67 ± 0.18 0.56 4 CL-3 × I2C-scFc 14.50 ± 0.07 7.91 ± 0.37 0.55 5 CL-4 × I2C-scFc 15.30 ± 2.55 9.19 ± 0.16 0.60

(54) Likewise, a balanced binding to human and cyno FcRn was confirmed via Biacore assays for the constructs denominated CL-1×I2C-scFc, CL-2×I2C-scFc, CL-3×I2C-scFc, and CL-4×I2C-scFc.

Example 4

(55) Scatchard-Based Analysis of the Affinity to Human and Macaque CLDN18.2 on Target Antigen Positive Cells

(56) The affinities of CLDN18.2×CD3 antibody constructs to CHO cells transfected with human or macaque CLDN18.2 were determined by Scatchard analysis. For this analysis, saturation binding experiments were performed using a monovalent detection system to precisely determine monovalent binding of the CLDN18.2×CD3 antibody constructs to the respective cell line.

(57) 2×10.sup.4 cells of the CHO cell line recombinantly expressing human CLDN18.2 were incubated each with 50 μl of a dilution series (twelve dilutions at 1:2) of the respective antibody construct (until saturation was reached) starting at 100-200 nM followed by 16 h incubation at 4° C. under agitation and one residual washing step. Then, the cells were incubated for another hour with 30 μl of an Alexa Fluor™ 488-conjugated AffiniPure™ Fab fragment goat anti-human IgG (H+L) solution. After one washing step, the cells were resuspended in 150 μl FACS buffer containing 3.5% formaldehyde, incubated for further 15 min, centrifuged, resuspended in FACS buffer and analyzed via FACS software. Data were generated from two independent sets of experiments, each using triplicates. Respective one site-specific binding evaluation was calculated to extrapolate maximal binding (Bmax). The concentrations of the antibody constructs at half-maximal binding were determined reflecting the respective KDs. Values of triplicate measurements were plotted as hyperbolic curves and as S-shaped curves to demonstrate proper concentration ranges from minimal to optimal binding.

(58) Values depicted in Table 3 were derived from two independent experiments per antibody construct. Cell based Scatchard analysis confirmed that the CLDN18.2×CD3 antibody constructs of the invention are nanomolar in affinity to human CLDN18.2 and—due to sequence identity of human and macaque extracellular domains—present with a cyno/human interspecies affinity gap of 1.

(59) TABLE-US-00003 TABLE 3 Affinities (KD) of CLDN18.2 × CD3 antibody constructs to CLDN18.2 as determined in cell based Scatchard analysis. Antibody constructs were measured in two independent experiments, each using a dilution series. CLDN18.2 × CD3 Cell based affinity Affinity gap antibody construct hu CLDN18.2 [nM] KD mac/KD hu* CL-1 × I2C-scFc 56.8 ± 10.7 1 CL-1 × I2C-6His 13.9 ± 1.8  1 *Human and cynomolgus CLDN18.2 share the identical amino acid sequence in the extracellular domains. Therefore the affinity gap equals “1” per definition

(60) In a separate Scatchard assay carried out under the same conditions, a cell-based affinity for human CLDN18.2 of 11.13±2.72 was measured for CL-2×I2C-scFc under the same conditions.

Example 5

(61) Confirmation of Binding to CLDN18.2 and Human/Cyno CD3 Expressing Cells

(62) For confirmation of binding to human CLDN18.2 and CD3 and to cyno CD3, antibody constructs of the invention were tested by flow cytometry using CHO cells transfected with human CLDN18.2, CD3-expressing human T cell leukemia cell line HPB-all (DSMZ, Braunschweig, ACC483), and the cynomolgus CD3-expressing T cell line LnPx 4119

(63) For flow cytometry 200,000 cells of the respective cell lines were incubated for 60 min at 4° C. with the purified antibody construct at a concentration of 5 μg/ml. After washing, bound antibody constructs having an Fc domain were detected with a goat anti-human Fc-gamma-PE (1:100) for 30 min at 4° C. The antibody construct having a his-tag was detected with an in-house mouse antibody specific for the CD3 binding part, followed by a goat anti-mouse Fc-gamme-PE (1:100) for 30 min at 4° C. Samples were measured by flow cytometry. Non-transfected CHO cells were used as negative control.

(64) The results are shown in FIG. 5. The CLDN18.2×CD3 antibody constructs of the invention stained CHO cells transfected with human CLDN18.2. Human and cyno T cell lines expressing CD3 were also recognized by the antibody constructs. There was no staining of non-transfected CHO cells, and no staining of the CLDN18.2-transfected CHO cells by the negative control anti-EGFRvIII×I2C-scFc antibody construct.

Example 6

(65) Confirmation of the Absence of Binding to Human CLDN18.2 Paralogues

(66) Human CLDN18.2 paralogues CLDN18.1, CLDN1, CLDN2, CLDN3, CLDN4, CLDN6 and CLDN9 were stably transfected into CHO cells. The sequences of the paralogues as used in the present example are depicted in SEQ ID Nos: 4-10. Protein expression was confirmed in FACS analyses with antibodies specific for the respective paralogues: CLDN1: rat-anti-human (5 μg/ml final) stock: 100 μg/ml R&D, MAB4618 CLDN2: in-house mouse Ab (1:100-200) CLDN3: mouse-anti-human (5 μg/ml final) stock: 500 μg/ml R&D, MAB4620 CLDN4: mouse-anti-human (5 μg/ml final) stock: 500 μg/ml R&D, MAB4219 CLDN6: mouse-anti human (5 μg/ml final) stock: 500 μg/ml R&D, MAB3656 CLDN9: rat-anti-human (5 μg/ml final) stock: 2 mg/ml, antibodies-online.com, ABIN1720917 CLDN18.1: in-house monoclonal mouse Ab (5 μg/ml final)

(67) The transfected CHO cells were incubated for 60 min at 4° C. with the respective antibody (see above) at a concentration of 5 μg/ml, followed by the respective PE conjugated antibody goat-anti-mouse IgG, Fc-gamma fragment PE conjugated (1:100), Jackson 115-116-071 or goat-anti-rat IgG, Fc-gamma fragment PE conjugated (1:100) Jackson, 112-116-071 for 30 min at 4° C.

(68) The flow cytometry assay was carried out as described in Example 5. The results are shown in FIG. 6. The analysis confirmed that the CLDN18.2×CD3 antibody construct of the invention that was tested in this assay does not cross-react with the human CLDN18.2 paralogues.

Example 7

(69) Cytotoxic Activity

(70) The potency of CLDN18.2×CD3 antibody constructs of the invention in redirecting effector T cells against CLDN18.2-expressing target cells was analyzed in different in vitro cytotoxicity assays: The potency of CLDN18.2×CD3 antibody constructs in redirecting stimulated human CD8+ effector T cells against human CLDN18.2-transfected CHO cells was measured in a 48 hour FACS-based cytotoxicity assay. The potency of CLDN18.2×CD3 antibody constructs in redirecting a macaque T cell line against human CLDN18.2-transfected CHO cells was measured in a 48 hour FACS-based cytotoxicity assay. The potency of CLDN18.2×CD3 antibody constructs in redirecting the T cells in unstimulated human PBMC against human CLDN18.2-transfected CHO cells (along with a negative control using human CLDN18.1-transfected CHO cells) was measured in a 48 hour FACS-based cytotoxicity assay. The potency of CLDN18.2×CD3 antibody constructs in redirecting the T cells in unstimulated human PBMC against a CLDN18.2 positive human gastric cancer line such as SNU-601 or SNU-620 was measured in a 48 hour FACS-based cytotoxicity assay.

Example 7.1

(71) Chromium Release Assay with Stimulated Human T Cells

(72) Stimulated T cells enriched for CD8.sup.+ T cells are obtained as follows: A petri dish (145 mm diameter) is coated with a commercially available anti-CD3 specific antibody (OKT3, Orthoclone) in a final concentration of 1 μg/ml for 1 hour at 37° C. Unbound protein is removed by one washing step with PBS. 3-5×10.sup.7 human PBMC are added to the precoated petri dish in 120 ml of RPMI 1640 with stabilized glutamine/10% FCS/20 U/ml IL-2 and stimulated for 2 days. On the third day, the cells are collected and washed once with RPMI 1640. IL-2 is added to a final concentration of 20 U/ml, and the cells are cultured again for one day in the same cell culture medium as above. CD8.sup.+ cytotoxic T lymphocytes (CTLs) are enriched by depletion of CD4.sup.+ T cells and CD56.sup.+ NK cells using Dynal-Beads according to the manufacturer's protocol.

(73) Human CLDN18.2-transfected CHO target cells are washed twice with PBS and labeled with 11.1 MBq .sup.51Cr in a final volume of 100 μl RPMI with 50% FCS for 60 minutes at 37° C. Subsequently, the labeled target cells are washed 3 times with 5 ml RPMI and then used in the cytotoxicity assay. The assay is performed in a 96-well plate in a total volume of 200 μl supplemented RPMI with an E:T ratio of 10:1. A starting concentration of 0.01-1 μg/ml of purified antibody construct and threefold dilutions thereof are used. Incubation time for the assay is 18 hours. Cytotoxicity is determined as relative values of released chromium in the supernatant relative to the difference of maximum lysis (addition of Triton-X) and spontaneous lysis (without effector cells). All measurements are carried out in quadruplicates. Measurement of chromium activity in the supernatants is performed in a gamma counter. Analysis of the results is carried out with appropriate software. EC50 values calculated by the analysis program from the sigmoidal dose response curves are used for comparison of cytotoxic activity.

Example 7.2

(74) FACS-Based Cytotoxicity Assay with Unstimulated Human PBMC

(75) Isolation of Effector Cells

(76) Human peripheral blood mononuclear cells (PBMC) were prepared by Ficoll density gradient centrifugation from enriched lymphocyte preparations (buffy coats). PBMC were prepared on the day of blood collection. After Ficoll density centrifugation and extensive washes with Dulbecco's PBS, remaining erythrocytes were removed from PBMC via incubation with erythrocyte lysis buffer (155 mM NH.sub.4Cl, 10 mM KHCO.sub.3, 100 μM EDTA). Platelets were removed via the supernatant upon centrifugation of PBMC at 100× g. Remaining lymphocytes mainly encompass B lymphocytes, T lymphocytes, NK cells and monocytes. PBMC were kept in culture at 37° C./5% CO.sub.2 in RPMI medium with 10% FCS.

(77) Depletion of CD14.sup.+ and CD56.sup.+ Cells

(78) For depletion of CD14.sup.+ cells, human CD14 MicroBeads (Miltenyi Biotec, MACS, #130-050-201) were used. For depletion of NK cells, human CD56 MicroBeads (Miltenyi Biotec, MACS, #130-050-401) were used. PBMC were counted and centrifuged for 10 min at room temperature with 300×g. The supernatant was discarded and the cell pellet resuspended in MACS isolation buffer (80 μl/10.sup.7 cells; PBS, 0.5% (v/v) FBS, 2 mM EDTA). CD14 MicroBeads and CD56 MicroBeads (20 μl/10.sup.7 cells) were added and incubated for 15 min at 4-8° C. The cells were washed with MACS isolation buffer (1-2 ml/10.sup.7 cells). After centrifugation (see above), supernatant was discarded and cells resuspended in MACS isolation buffer (500 μl/10.sup.8 cells). CD14/CD56 negative cells were then isolated using LS Columns (Miltenyi Biotec, #130-042-401). PBMC w/o CD14+/CD56+ cells were cultured in RPMI complete medium (i.e. RPMI1640 supplemented with 10% FBS, 1× non-essential amino acids, 10 mM Hepes buffer, 1 mM sodium pyruvate and 100 U/ml penicillin/streptomycin) at 37° C. in an incubator until needed.

(79) Target Cell Labeling

(80) For the analysis of cell lysis in flow cytometry assays, the fluorescent membrane dye DiOC.sub.18 (DiO) (Molecular Probes) was used to label target cells (such as human CLDN18.2-transfected CHO cells) and distinguish them from effector cells. Briefly, cells were harvested, washed once with PBS and adjusted to 10.sup.6 cells/ml in PBS containing 2% (v/v) FBS and the membrane dye DiO (5 μl/10.sup.6 cells). After incubation for 3 min at 37° C., cells were washed twice in complete RPMI medium, and the cell number was adjusted to 1.25×10.sup.5 cells/ml. The vitality of cells was determined using 0.5% (v/v) isotonic EosinG solution.

(81) Flow Cytometry Based Analysis

(82) This assay was designed to quantify the lysis of target cells (such as human CLDN18.2-transfected CHO cells) in the presence of serial dilutions of CLDN18.2×CD3 antibody constructs. Equal volumes of DiO-labeled target cells and effector cells (i.e., PBMC w/o CD14.sup.+ cells) were mixed, resulting in an E:T cell ratio of 10:1. 160 μl of this suspension were transferred to each well of a 96-well plate. 40 μl of serial dilutions of the CLDN18.2×CD3 antibody constructs to be analyzed (and possibly a negative control antibody construct such as an anti-CD3 (I2C)-based bispecific antibody construct recognizing an irrelevant target antigen) or RPMI complete medium as an additional negative control were added. The antibody construct-mediated cytotoxic reaction proceeded for 48 hours in a 7% CO.sub.2 humidified incubator. Then cells were transferred to a new 96-well plate and loss of target cell membrane integrity was monitored by adding propidium iodide (PI) at a final concentration of 1 μg/ml. PI is a membrane impermeable dye that normally is excluded from viable cells, whereas dead cells take it up and become identifiable by fluorescent emission.

(83) Samples were measured by flow cytometry on a FACSCanto II instrument and analyzed by FACSDiva software (both from Becton Dickinson). Target cells were identified as DiO-positive cells. PI-negative target cells were classified as living target cells. Percentage of cytotoxicity was calculated according to the following formula:

(84) $\mathrm{Cytotoxicity}\left[\%\right]=\frac{n_{\mathrm{dead}\mathrm{target}\mathrm{cells}}}{n_{\mathrm{target}\mathrm{cells}}}\times 100$$n=\mathrm{number}\mathrm{of}\mathrm{events}$

(85) Using the Prism software (GraphPad Software Inc.), the percentage of cytotoxicity was plotted against the corresponding antibody construct concentrations. Dose response curves were analyzed with the four parametric logistic regression models for evaluation of sigmoid dose response curves with fixed hill slope and EC50 values were calculated.

Example 7.3

(86) Potency of Redirecting Unstimulated Human PBMC Against Target Cells

(87) The cytotoxic activity of CLDN18.2×CD3 antibody constructs was analyzed in a FACS-based 48h cytotoxicity assay using unstimulated human PBMC (CD14 neg./CD56 neg.) as effector cells and using as target cells in an E:T ratio of 10:1: (1) CHO cells transfected with human CLDN18.2, (2) CHO cells transfected with human CLDN18.1, and (3) natural expresser cell lines SNU-601 and SNU-620.

(88) The assay was carried out as described in Example 7.2 above. The results of the cytotoxicity assays are shown in Table 4.

(89) TABLE-US-00004 TABLE 4 EC50 values [pM] of CLDN18.2 × CD3 antibody constructs as measured in a 48-hour FACS-based cytotoxicity assay with unstimulated human PBMC as effector cells. CLDN18.2 × CD3 Target cells antibody construct EC50 [pM] (1) CL-1 × I2C-scFc 5.2 (1) CL-1 × I2C-6His 0.5 (2) CL-1 × I2C-scFc 99396 (2) CL-1 × I2C-6His 1985 (3) CL-1 × I2C-scFc 104 (3) CL-1 × I2C-6His 2.3 (4) CL-1 × I2C-scFc 185 (4) CL-1 × I2C-6His 12 Target cells: (1) CHO cells transfected with hu CLDN18.2; (2) CHO cells transfected with hu CLDN18.1; (3) SNU-601; (4) SNU-620.

(90) The assay demonstrates that the antibody constructs of the invention do not exhibit significant undesired cytotoxic activity against CHO cells transfected with the CLDN18.2 paralogue CLDN18.1 (rows highlighted in grey). In the case of the construct denominated “CL-1×I2C-6His”, the factor between the EC.sub.50 value for CLDN18.2-CHO and the EC50 value for CLDN18.1-CHO is almost 4.000. Moreover, in the case of the construct denominated “CL-1×I2C-scFc”, the factor between the EC.sub.50 value for CLDN18.2-CHO and the EC50 value for CLDN18.1-CHO is almost 20.000.

(91) Usually EC.sub.50 values are expected to be lower when using target cells that express higher levels of CLDN18.2 on the cell surface compared with target cells having a lower target expression rate. Therefore, it is usually observed—and demonstrated in the present assay—that the use of CHO cells transfected with CLDN18.2 has a tendency to result in lower EC.sub.50 values compared with the use of natural expressers.

Example 7.4

(92) Potency of Redirecting Human PBMC Against Natural Expresser Target Cells

(93) The cytotoxic activity of CLDN18.2×CD3 antibody constructs of the invention was compared against other CLDN18.2×CD3 antibody constructs which bind to a different epitope within the CLDN18.2 target (CL-3 and CL-4, see Example 2). Cancer cell lines stably expressing endogenous levels of CLDN18.2 (GSU, NUGC, IM95, SNU620, SNU601) and the CLDN18.2 negative control cell line AGS were stably labeled with luciferase (Luc). Furthermore, CHO cells were stably transfected to overexpress CLDN18.2. Two human T cell donors were used as a source for effector cells. The E:T ratio of the cytotoxicity assay was 10:1. The 1:3 serial titration of the antibody constructs started at a concentration of 30 nM. The assays were incubated for 48 h at 37° C. Cytotoxicity readout was performed by the luciferase assay system ONE-Glo™ (Promega) for Luc labeled cells and the luminescent cell viability assay CellTiter-Glo™ (Promega) for CLDN18.2 transfected CHO cells.

(94) With the exception of the sequences for the anti-CLDN18.2 VH and VL regions, the antibody constructs CL-3 and CL-4 were identical to the antibody constructs of the invention analyzed in the cytotoxicity assays of the present Example. The VH and VL regions of the two antibody constructs CL-3 and CL-4 are disclosed as SEQ ID NOs: 8+15 and SEQ ID NOs: 6+11, respectively, of WO 2014/075788. See also SEQ ID NOs: 145+146 and SEQ ID NOs: 150+151 as disclosed herein.

(95) Results are shown in Table 5 below and in FIG. 7. Irrespective of the target cells, the antibody constructs of the present invention (here: CL-1, CL-2) were shown to have a significantly higher cytotoxic potency than the antibody constructs binding to a different CLDN18.2 epitope (CL-3, CL-4). While the antibody constructs of the present invention display EC.sub.50 values in the two-digit picomolar range, the comparative constructs display EC.sub.50 values in the three-digit up to the five-digit picomolar range. None of the constructs showed any activity against target-negative cell line (data not shown).

(96) TABLE-US-00005 TABLE 5 EC.sub.50 values [pM] of CLDN18.2 × CD3 antibody constructs as measured in a 48-hour cytotoxicity assay. Target cell CL-1 × I2C-scFc CL-2 × I2C-scFc CL-3 × I2C-scFc CL-4 × I2C-scFc GSU  7.4 ± 0.3 12.1 ± 0.7 360.3 ± 47.2 1655.5 ± 71.5  IM95 16.0 ± 2.7 21.4 ± 4.2 623.1 ± 93.3 5900.0 ± 514.0 NUGC4 117.1 ± 46.2  97.1 ± 35.1 1196.5 ± 426.4 >30000 SNU-601 61.9 ± 6.9 64.7 ± 6.7 13797.5 ± 4169.5 >30000 SNU-620 29.0 ± 3.6 31.0 ± 5.3 557.1 ± 98.4 11093.5 ± 1996.5 CHO-CLDN18.2 13.6 ± 0.1 13.2 ± 0.5  440.2 ± 112.5 3181.0 ± 119.0

(97) In order to exclude that these observations were due to a significantly higher affinity of the antibody constructs of the present invention (here: CL-1, CL-2) as compared to the control constructs (here: CL-3), a cell-based affinity assay was carried out with CHO cells transfected with hu Cldn18.2. It was shown that the affinities of the three tested constructs CL-1, CL-2 and CL-3 were in a very similar range. This means that the favorable epitope/activity relationship demonstrated for the antibody constructs of the present invention is not due to the control constructs merely having a lower affinity and hence exhibiting a lower cytotoxic activity. Instead, the potency seems to be due to the particular epitope within CLDN18.2 that is recognized by the present antibody constructs.

Example 7.5

(98) Potency of Redirecting Macaque T Cells Against CLDN18.2-Expressing CHO Cells

(99) The cytotoxic activity of CLDN18.2×CD3 antibody constructs was analyzed in a 48 h FACS-based cytotoxicity assay using CHO cells transfected with human CLDN18.2 as target cells, and the macaque T cell line 4119LnPx (Knappe et al. Blood 95:3256-61 (2000)) as source of effector cells at an E:T ratio of 10:1. Note that human and cynomolgus CLDN18.2 share the identical amino acid sequence in the extracellular domains. Target cell labeling of transfected CHO cells and flow cytometry based analysis of cytotoxic activity was performed as described in Example 7.2 above.

(100) Results are shown in Table 6. Macaque T cells from cell line 4119LnPx were induced to efficiently kill CLDN18.2-transfected CHO cells by CLDN18.2×CD3 antibody constructs according to the invention. The antibody constructs presented potently with 1-digit to 2-digit picomolar EC50-values in this assay, confirming that they are very active in the macaque system.

(101) TABLE-US-00006 TABLE 6 EC50 values [pM] of CLDN18.2 × CD3 antibody constructs as measured in a 48-hour FACS-based cytotoxicity assay with macaque T cell line 4119LnPx as effector cells and CHO cells transfected with human CLDN18.2 as target cells. CLDN18.2 × CD3 antibody construct EC50 [pM] CL-1 × I2C-scFc 38 CL-1 × I2C-6His 7.2

Example 8

(102) Monomer to Dimer Conversion after (i) Three Freeze/Thaw Cycles and (ii) 7 Days of Incubation at 37° C.

(103) CLDN18.2×CD3 monomeric antibody constructs were subjected to different stress conditions followed by high performance SEC to determine the percentage of initially monomeric antibody construct which had been converted into dimeric antibody construct.

(104) (i) 25 μg of monomeric antibody construct were adjusted to a concentration of 250 μg/ml with generic formulation buffer and then frozen at −80° C. for 30 min followed by thawing for 30 min at room temperature. After three freeze/thaw cycles the dimer content was determined by HP-SEC.

(105) (ii) 25 μg of monomeric antibody construct were adjusted to a concentration of 250 μg/ml with generic formulation buffer followed by incubation at 37° C. for 7 days. The dimer content was determined by HP-SEC.

(106) A high-performance (HP) silica-based size exclusion liquid chromatography (SEC) column was connected to an FPLC equipped with an autosampler. Column equilibration and running buffer consisted of 100 mM KH.sub.2PO.sub.4-200 mM Na.sub.2SO.sub.4 adjusted to pH 6.6. The antibody construct solution (25 μg protein) was applied to the equilibrated column and elution was carried out at a flow rate of 0.75 ml/min at a maximum pressure of 7 MPa. The whole run was monitored at 280, 254 and 210 nm optical absorbance. Analysis was done by peak integration of the 210 nm signal recorded in the software run evaluation sheet. Dimer content was calculated by dividing the area of the dimer peak by the total area of monomer plus dimer peak.

(107) The results are shown in Table 7 below. The analyzed CLDN18.2×CD3 antibody constructs presented with dimer percentages of 0.0% after three freeze/thaw cycles, and with dimer percentages of ≤2% after 7 days of incubation at 37° C.

(108) TABLE-US-00007 TABLE 7 Percentage of monomeric versus dimeric CLDN18.2 × CD3 antibody constructs as determined by High Performance Size Exclusion Chromatography (HP-SEC). Percentage of dimer Percentage of dimer CLDN18.2 × CD3 after three freeze/ after 7 days of antibody construct thaw cycles incubation at 37° C. CL-1 × I2C-scFc 0.0 0.0 CL-1 × I2C-6His 0.72 0.0

Example 9

(109) Thermostability

(110) Antibody aggregation temperature was determined as follows: 40 μl of antibody construct solution at a concentration of 250 μg/ml are transferred into a single use cuvette and placed in a dynamic light scattering device. The sample was heated from 40° C. to 70° C. at a heating rate of 0.5° C./min with constant acquisition of the measured radius. Increase of radius indicating melting of the protein and aggregation was used by the software package delivered with the DLS device to calculate the aggregation temperature of the antibody construct.

(111) The antibody construct CL-1×I2C-scFc was shown to have a beneficial aggregation temperature of ≥51° C., more specifically, of 51.7° C. In a separate assay, the aggregation temperature of the molecule denominated CL-1×I2C-6His was shown to have a value of ≥45° C., more specifically, of 49.7° C.

Example 10

(112) Turbidity at a Concentration of the Monomeric Antibody Construct of 2.5 mg/ml

(113) 1 ml of purified antibody construct solution of a concentration of 250 μg/ml was concentrated by spin concentration units to 2500 μg/ml. After 16 h storage at 5° C. the turbidity of the solution was determined by OD340 nm optical absorption measurement against the generic formulation buffer.

(114) The antibody construct CL-1×I2C-scFc was shown to have a very favourable turbidity of ≤0.021, while CL-1×I2C-6His was shown to have a very favourable turbidity of ≤0.029. A similar measurement of turbidity after three freeze/thaw cycles resulted in a turbidity of 0.021 for CL-1×I2C-scFc.

Example 11

(115) Protein Homogeneity by High Resolution Cation Exchange Chromatography

(116) The protein homogeneity of the antibody constructs of the invention was analyzed by high resolution cation exchange chromatography (CIEX).

(117) In one assay, 50 μg of antibody construct monomer were diluted with 50 ml binding buffer A (20 mM sodium dihydrogen phosphate, 30 mM NaCl, 0.01% sodium octanate, pH 5.5), and 40 ml of this solution were applied to a 1 ml BioPro SP-F column (YMC, Germany) connected to an FPLC device. After sample binding, a wash step with further binding buffer was carried out. For protein elution, a linear increasing salt gradient using buffer B (20 mM sodium dihydrogen phosphate, 1000 mM NaCl, 0.01% sodium octanate, pH 5.5) up to 50% percent buffer B was applied over 10 column volumes. The whole run was monitored at 280, 254 and 210 nm optical absorbance. Analysis was done by peak integration of the 280 nm signal recorded in the software run evaluation sheet. In this assay, the homogeneity of the molecule denominated “CL-1×I2C-6His” (SEQ ID NO: 131) was shown to have a value of 82.5%.

(118) In another assay, a Bio SCX analytical CIEX column (Agilent, Frankfurt, Germany) was connected to a UPLC device (Waters, Eschborn, Germany) and equilibrated with buffer A (binding buffer) consisting of 50 mM MES, pH 5.6, 0.05% sodium azide. 10 μg of antibody construct monomer was applied to the column and bound to the column matrix. After sample binding, a wash step with further binding buffer B was carried out. For protein elution, a linear increasing salt gradient using buffer B consisting of 50 mM MES, 1000 mM sodium chloride, pH 5.6, 0.05% sodium azide up to 50% percent buffer B was applied. The whole run was monitored at 280 nm optical absorbance. Analysis was done by peak integration of the 280 nm signal recorded in the software run evaluation sheet. In this assay, the molecule denominated “CL-1×I2C-scFc”(SEQ ID NO: 132) was shown to have a homogeneity of 97.6% (area under the curve (=AUC) of the main peak).

Example 12

(119) Surface Hydrophobicity as Measured by HIC Butyl

(120) The surface hydrophobicity of antibody constructs CL-1×I2C-scFc and CL-1×I2C-6His was tested in Hydrophobic Interaction Chromatography HIC in flow-through mode.

(121) 50 μg of monomeric antibody construct were diluted with generic formulation buffer to a final volume of 500 μl (10 mM citric acid, 75 mM lysine HCl, 4% trehalose, pH 7.0) and applied to a 1 ml Butyl Sepharose FF column connected to a Akta Purifier FPLC system. The whole run was monitored at 280, 254 and 210 nm optical absorbance. Analysis was done by peak integration of the 280 nm signal recorded in the Akta Unicorn software run evaluation sheet. Elution behavior was evaluated by comparing area and velocity of rise and decline of protein signal thereby indicating the strength of interaction of the antibody construct with the matrix.

(122) The antibody construct had a good elution behaviour, which was rapid and complete.

Example 13

(123) Efficacy Evaluation of a CLDN18.2×CD3 Antibody Construct in a GSU-Luc Advanced Stage Gastric Cancer Model in NOD/SCID Mice

(124) The anti-tumor activity of the CLDN18.2×CD3 antibody construct having SEQ ID NO: 132 (CL-1×I2C-scFc) was tested in a model of female NOD/SCID mice which were subcutaneously injected on day 1 with 5×10.sup.6 GSU-luc (luciferase) cells. 2×10.sup.7 effector cells (in vitro expanded and activated human CD3.sup.+ T cells) were injected intraperitoneally on day 8. Treatment occurred on days 12, 19 and 26 (Q7Dx3). Two control groups, one w/o T cells (group 1), another one with T cells (group 2) were treated with 0.1 ml/admin of vehicle (25 mM L-lysine, 0.002% (w/v) polysorbate 80 in 0.9% (w/v) sodium chloride pH 7.0) by intravenous bolus injections. The antibody construct was administered at a concentration of 1 mg/kg/admin by intravenous bolus injections in a final volume of 0.1 ml (group 3). The number of mice per group was 5 (group 1), 10 (group 2) and 10 (group 3).

(125) Tumors were measured by caliper during the study and progress evaluated by intergroup comparison of tumor volumes (TV). The tumor growth inhibition T/C [%] on day x is determined by calculating the tumor volume as T/C (%)=100×(median TV of analyzed group)/(median TV of control group), and the calculated values are depicted in Table 8.

(126) TABLE-US-00008 TABLE 8 T/C [%] values of the GSU-luc advanced stage gastric cancer model in NOD/SCID mice. Group 2 Group 3 Day of Study median n T/C [%] median n T/C [%] 11 208 10 100 208 10 100 13 275 10 100 316 10 115 15 430 10 100 369 10 86 18 695 10 100 185 10 27 20 891 10 100 123 10 14 22 896 4 100 79 10 9 25 1373 2 100 52 9 4 “Median” = median tumor volume of analyzed group.

(127) The results are furthermore shown in FIG. 8. Significant tumor growth inhibition was shown in the NOD/SCID mouse model in the present advanced stage tumor analysis. Furthermore, treatment with the CLDN18.2×CD3 antibody construct had no effect on the body weight of the mice (data not shown).

Example 14

(128) Evaluation of the Anti-Tumor Activity of a CLDN18.2×CD3 Antibody Construct in a SNU620-Luc Tumor Formation Xenograft Model in Female Athymic Nude Mice

(129) The anti-tumor activity of the CLDN18.2×CD3 antibody construct having SEQ ID NO: 132 (CL-1×I2C-scFc) was tested in a model of female athymic nude mice which were subcutaneously injected on day 1 with SNU620-luc (5×10.sup.6 target cells)/PBMC (2.5×10.sup.6 effector cells) mix in 50% matrigel. Treatment of groups 1-4 occurred on days 3, 10 and 17 (Q7D). The control group which received tumor/PBMC mix (group 1) was treated with vehicle (25 mM L-lysine, 0.002% (w/v) polysorbate 80 in 0.9% (w/v) sodium chloride pH 7.0) at 0.1 ml/admin by intravenous bolus injections. The antibody construct was administered at a concentration of 1 mg/kg/admin (group 2), 0.1 mg/kg/admin (group 3) and 0.01 mg/kg/admin (group 4) in intravenous bolus injections at a volume of 0.1 ml. The number of mice per group was 10.

(130) Tumors were measured by caliper during the study and progress evaluated by intergroup comparison of tumor volumes (TV). The tumor growth inhibition T/C [%] on day x is determined by calculating the tumor volume as T/C (%)=100×(median TV of analyzed group)/(median TV of control group), and the calculated values are depicted in Table 9.

(131) TABLE-US-00009 TABLE 9 T/C [%] values of the SNU620-luc tumor formation xenograft model in female athymic nude mice. Group 1 Group 2 Group 3 Group 4 Group 5 Day median T/C [%] median T/C [%] median T/C [%] median T/C [%] median T/C [%] 5 142 100 97 68 107 76 148 104 156 110 8 196 100 53 27 96 49 110 56 180 91 10 231 100 34 15 51 22 83 36 217 94 12 274 100 21 8 29 11 46 17 168 61 15 321 100 13 4 19 6 27 8 70 22 17 353 100 10 3 13 4 21 6 44 12 19 448 100 8 2 11 2 16 4 53 12 22 552 100 7 1 10 2 15 3 64 12 “Median” = median tumor volume of analyzed group.

(132) The results are furthermore shown in FIG. 9. Significant tumor growth inhibition was shown in the athymic nude mouse model with all tested concentrations of the CLDN18.2×CD3 antibody construct. Furthermore, treatment with the CLDN18.2×CD3 antibody construct had no effect on the body weight of the mice (data not shown).

Example 15

(133) Cyno Exploratory Toxicology Study

(134) An exploratory NHP tolerability study was carried out with a CLDN18.2×CD3 antibody construct CL-1×I2C-6His (SEQ ID NO: 131) by intravenous administration. Systemic exposures achieved after the starting dose of 25 μg/kg/day (˜20× the human EC90) were well tolerated. Dose escalation to 125 μg/kg/day (˜120× the human EC90) resulted in clinical as well as histopathological effects expected from a T cell engaging antibody construct targeting CLDN18.2

Example 16

(135) In Vitro Combination Therapy Studies

(136) Treatment with CL-1×I2C-scFc activates human T cells, leading to upregulation of PD-1 on T cells (data not shown). The treatment can also lead to upregulation of PD-L1 on tumor cells. GSU and NUG-C4 gastric cancer cell lines which are CLDN18.2 positive were engineered to overexpress PD-L1. Cells were incubated with CL-1×I2C-scFc, and with activated human T cells in the absence or presence of an anti-PD-1 antibody having a heavy chain amino acid sequence as depicted in SEQ ID NO: 360 and a light chain amino acid sequence as depicted in SEQ ID NO: 361. Cytotoxicity was assessed after a 24 h incubation. The results are shown in Table 10 below.

(137) TABLE-US-00010 TABLE 10 EC.sub.50 and EC.sub.90 values [pM] of two separate cytotoxicity assays for two different cell lines each GSU NUG-C4 EC.sub.50 EC.sub.90 EC.sub.50 EC.sub.90 CL-1 × I2C-scFc 52.8 ± 9.4  168.0 ± 77.7  168.0 ± 85.1  .sup. 712 ± 91.7 alone CL-1 × I2C-scFc + 28.1 ± 10.7 83.4 ± 45.0 63.9 ± 57.0 219.9 ± 77.2 anti-PD-1 ab

(138) This assay demonstrates that the addition of an anti-PD-1 antibody increases the efficacy of the CD3×CLDN18.2 antibody constructs of the invention.

Example 17

(139) Combination Therapy Studies in Mice

(140) Using a human/murine chimeric CD3 epsilon knock-in mouse model, therapeutic combinations of antibodies/antibody constructs were evaluated to potentially enhance efficacy of bispecific antibody constructs as described herein. A bispecific single-chain anti-human CD3 (“I2C” scFv)×anti-mouse CLDN18.2 (scFv)-scFc surrogate antibody construct was generated. This molecule demonstrated potent activity against mouse CLDN18.2-expressing cells in vitro. The molecule was then tested in a genetically modified immune-competent mouse model (human/murine chimeric CD3 epsilon knock-in) for its anti-tumor activity in combination with anti-mouse PD-1, anti-mouse CTLA4 and anti-mouse 4-1BB (CD137) antibodies against CLDN18.2-positive sub-cutaneously implanted tumors (B16F10 muCLDN18.2 syngeneic model). According to the study design, the mice were randomized on day 10, having a tumor volume of about 50-100 mm.sup.3. Ten mice were included in each group. The CD3×CLDN18.2 antibody construct (150 μg/kg) or a control (anti-CD3×anti-EGFRvIII, 150 μg/kg) bispecific antibody construct were administered (i.v.) on days 11 and 18, and the antibodies (anti-PD-1 100 μg; anti-4-1BB 150 μg; anti-CTLA4 300 μg; antibody idotype as control) were dosed on days 11, 14, 17 and 20. Tumor volumes were measured on days 10, 13, 17, 20 and 25 (terminal harvest, all groups).

(141) All three antibodies enhanced efficacy of the bispecific construct. The agonistic anti-4-1BB monoclonal antibody demonstrated no single-agent activity against the CLDN18.2 positive tumors. Specifically, tumor growth inhibition (TGI) on day 25 was determined as follows: Control bispecific antibody construct+antibody isotype.fwdarw.0% (standard) Control bispecific antibody construct+anti-4-1BB ab.fwdarw.not significant Control bispecific antibody construct+anti-CTLA-4 ab.fwdarw.21% CD3×CLDN18.2 antibody construct+antibody isotype.fwdarw.36% Control bispecific antibody construct+anti-PD-1 ab.fwdarw.57% CD3×CLDN18.2 antibody construct+anti-CTLA-4 ab.fwdarw.76% CD3×CLDN18.2 antibody construct+anti-4-1BB ab.fwdarw.77% CD3×CLDN18.2 antibody construct+anti-PD-1 ab.fwdarw.79%