EA040594B1 - OPTIMIZED BISPECIFIC ANTI-CD3 ANTIBODIES AND THEIR APPLICATIONS - Google Patents

Description

Link to sequence listing

This document incorporates by reference a sequence listing provided in machine-readable form as file 10151WO01_ST25.txt, created on September 22, 2016, and containing

264418 bytes.

Technical field

The invention relates to bispecific antibodies targeting an effector antigen, such as the CD3 antigen and a tumor-associated antigen, and methods for killing tumors. The present invention provides bispecific antibodies comprising an effector arm that binds to an effector antigen with little or no detectable binding affinity, such as an anti-CD3 antigen binding arm that binds to CD3 with a KD greater than about 500 nM, in an affinity binding assay in vitro.

State of the art

The potential of therapeutic bispecific antibodies (bsAbs), particularly in cancer immunotherapies, is to combine multiple antigenic targets to elicit a more robust innate immune response against unwanted target cells or organisms.

It is now well known that to mediate redirected lysis, a bsAb must group the target cell directly with the initiator molecule on an effector cell, such as a T cell. There are many factors in the construction of bsAbs, for example size and composition will affect in vivo distribution and stability (Segal, D.M., Weiner, G.J., and Weiner, L.M.. Current Opinion in Immunol. 1999, 11:558-562; Chames, P. and Baty, D. MAbs 2009, 1(6):539-547). Differential outcomes are difficult to predict depending on the subset of T cells that fire to respond as well as the state of the stimulated T cell. It is well known that bsAbs do not provide consistent results (Manzke O., et al. Cancer Immunol. Immunother. 1997, 45:198-202). For example, without adequate cytokine production, CD3 cross-linking can induce an apoptotic response in the T cell (Noel P.J., Boise L.H., Thompson C.B.: Regulation of T cell activation by CD28 and CTLA4. Adv. Exp. Med. Biol. 1996, 406:209 -217). The subset of T cells and the state of differentiation of such recruited T cells, e.g., naive T cells, is important for efficiency because naive T cells cannot lyse target cells without prior activation (such as cross-linking to TCR in the presence of IL-2) .

Some bispecific therapies have been successful, but as with many cancer treatments, they are not without consequences. Toxicity is a major cause of failure among cancer therapies. It is well known that the toxicity of so-called chemotherapeutic drugs is the main cause of patient side effects and secondary diseases. The act of cellular destruction itself is associated with trouble for the patient. A variety of cytotoxic responses can be elicited by activating effector cells, such as T cells and cancer target cells, but which type of response is most useful in tumor immunotherapy remains to be seen. The preferred method would be to identify anti-CD3 antibodies for use in bispecific therapy having reduced side effects while maintaining efficacy and desirable pharmacokinetic (PK) properties.

Techniques such as affinity maturation have been described which use structure/activity ratio (SAR) mutagenesis to optimize antibodies with increased and improved binding specificity or affinity for the target antigen compared to the parent antibody (see e.g. WO 2011/056997 published May 12, 2011). Modified OKT3 antibodies have been described that can bind and interact with CD3 with varying degrees of affinity while exhibiting moderate to high T cell activation (US 7820166). However, methods for reducing the binding affinity of antibody molecules near or beyond the detectable level of binding have not been described or demonstrated to be effective in reducing or suppressing a tumor.

Thus, there is a need for alternative bispecific antigen-binding molecules with controlled cytotoxicity and better PK properties. Such cancer therapy would be highly beneficial in therapeutic settings.

Brief summary of the invention

In a first aspect, the invention provides antibodies and antigen-binding fragments thereof that bind human CD3 having little or no detectable affinity for human and/or cynomolgus monkey CD3. The antibodies of this aspect of the invention are particularly useful for targeting CD3-expressing T cells and for stimulating T cell activation, for example, under conditions where T cell-mediated lysis is beneficial or desirable. Ahtu-CD3 antibodies of the invention, or antigen-binding portions thereof, can be incorporated as part of a bispecific antibody that directs CD3-mediated T-cell activation to specific cell types, such as tumor cells or infectious agents.

Typical anti-CD3 antibodies according to this invention are shown in table. 2 and 3 in this document. In table. 2 shows the identifiers of the amino acid sequences of the variable regions

- 1 040594 heavy chain (HCVR), as well as regions for determining the complementarity of the heavy chain (HCDR1, HCDR2 and HCDR3). In table. 3 shows the sequence identifiers of nucleic acid molecules encoding the HCVR, HCDR1, HCDR2, and HCDR3 regions of exemplary anti-CD3 antibodies. In table. 4 and shows the light chain variable regions (LCVR) as well as the complementarity detection regions (LCDR1, LCDR2 and LCDR3) of exemplary anti-CD3 antibodies.

This invention relates to antibodies or antigennegative fragments containing HCVR containing an amino acid sequence selected from any of the HCVR amino acid sequences shown in table. 2 or substantially similar sequences having at least 90%, at least 95%, at least 98%, or at least 99% sequence identity.

This invention also relates to antibodies or their antigennegative fragments containing a pair of amino acid sequences HCVR and LCVR (HCVR/LCVR), containing any of the amino acid sequences of HCVR shown in table. 2, paired with any of the LCVR amino acid sequences shown in table. 4, or a conventional light chain derived from a cognate light chain, an anti-TAA heavy chain, or a variable region derived from a known or publicly available light chain derived light chain domain exhibiting heterogeneity or the ability to pair with a wide variety of non-cognate heavy chains, i.e. . universal or conventional light chain. In accordance with some variants of implementation of this invention relates to antibodies or their antigennegative fragments containing the pair of amino acid sequences HCVR/LCVR contained in any of the illustrative anti-CD3 antibodies shown in table. 2, paired with the exemplary light chain variable regions shown in Table 2. 4. In some embodiments, the HCVR/LCVR amino acid sequence pair is selected from the group consisting of SEQ ID NO: 10/162 (eg, CD3-VH-G2); 18/162 (eg CD3-VH-G3); 26/162 (eg CD3-VH-G4); 34/162 (eg CD3-VH-G5); 42/162 (eg CD3-VH-G8); 50/162 (eg CD3-VH-G9); 58/162 (eg CD3-VH-G10); 66/162 (eg CD3-VH-G11); 74/162 (eg CD3-VH-G12); 82/162 (eg CD3-VH-G13); 90/162 (eg CD3-VH-G14); 98/162 (eg CD3-VH-G15); 106/162 (eg CD3-VH-G16); 114/162 (eg CD3-VH-G17); 122/162 (eg CD3-VH-G18); 130/162 (eg CD3-VH-G19); 138/162 (eg CD3-VH-G20) and 146/162 (eg CD3-VH-G21).

This invention also relates to antibodies or antigennegative fragments thereof, containing a heavy chain CDR1 (HCDR1) containing an amino acid sequence selected from any of the HCDR1 amino acid sequences shown in table. 2 or substantially similar sequences having at least 95%, at least 98%, or at least 99% sequence identity. This invention also relates to antibodies or antigen-binding fragments thereof, containing a heavy chain CDR1 (HCDR1) containing the amino acid sequence shown in SEQ ID NO: 178.

This invention also relates to antibodies or antigennegative fragments thereof, containing a heavy chain CDR2 (HCDR2) containing an amino acid sequence selected from any of the HCDR2 amino acid sequences shown in table. 2 or substantially similar sequences having at least 95%, at least 98%, or at least 99% sequence identity. This invention also relates to antibodies or antigen-binding fragments thereof, containing a heavy chain CDR2 (HCDR2) containing the amino acid sequence shown in SEQ ID NO: 179.

This invention also relates to antibodies or antigennegative fragments thereof, containing a heavy chain CDR3 (HCDR3) containing an amino acid sequence selected from any of the HCDR3 amino acid sequences shown in table. 2 or substantially similar sequences having at least 95%, at least 98%, or at least 99% sequence identity. This invention also relates to antibodies or antigen-binding fragments thereof, containing a heavy chain CDR3 (HCDR3) containing the amino acid sequence shown in SEQ ID NO: 180.

This invention also relates to antibodies or antigennegative fragments thereof, containing a light chain CDR1 (LCDR1) containing an amino acid sequence selected from any of the LCDR1 amino acid sequences shown in table. 4 or substantially similar sequences having at least 95%, at least 98%, or at least 99% sequence identity. The invention also relates to antibodies or antigen-binding fragments thereof, comprising a light chain CDR1 (LCDR1) derived from a cognate light chain, an anti-TAA heavy chain, or derived from a light chain exhibiting heterogeneity or the ability to pair with a wide variety of non-natural heavy chains, i.e. e. universal or conventional light chain.

This invention also relates to antibodies or antigennegative fragments thereof, containing a light chain CDR2 (LCDR2) containing an amino acid sequence selected from any of the LCDR2 amino acid sequences shown in table. 4 or substantially similar sequences having at least 95%, at least 98%, or at least 99% sequence identity. The invention also relates to antibodies or antigen-binding fragments thereof comprising a light chain CDR2 (LCDR2) derived from a cognate light chain, an anti-TAA heavy chain, or derived from a light chain exhibiting heterogeneity or the ability to pair with a wide variety of non-natural heavy chains, i.e. e. universal or conventional light chain.

This invention also relates to antibodies or antigennegative fragments thereof, containing a light chain CDR3 (LCDR3) containing an amino acid sequence selected from any of the LCDR3 amino acid sequences shown in table. 4 or substantially similar sequences having at least 95%, at least 98%, or at least 99% sequence identity. The invention also relates to antibodies or antigen-binding fragments thereof, comprising a light chain CDR3 (LCDR3) derived from a cognate light chain, an anti-TAA heavy chain, or derived from a light chain exhibiting heterogeneity or the ability to pair with a wide variety of non-natural heavy chains, i.e. e. universal or conventional light chain.

This invention also relates to antibodies or antigennegative fragments containing HCDR3 and a pair of LCDR3 amino acid sequences (HCDR3/LCDR3) containing any of the HCDR3 amino acid sequences shown in table. 2, paired with any of the LCDR3 amino acid sequences shown in table. 4. In accordance with some variants of implementation of this invention relates to antibodies or their antigennegative fragments containing a pair of amino acid sequences HCDR3/LCDR3 contained in any of the illustrative anti-CD3 antibodies shown in table. 2. In some embodiments, the HCDR3/LCDR3 amino acid sequence pair is selected from the group consisting of SEQ ID NO: 16/168 (eg, CD3-VH-G2); 24/168 (eg CD3-VH-G3); 32/168 (for example,

CD3-VH-G4); 40/168 (eg CD3-VH-G5); 48/168 (eg CD3-VH-G8); 56/168 (for example,

CD3-VH-G9); 64/168 (eg CD3-VH-G10); 72/168 (eg CD3-VH-G11); 80/168 (for example,

CD3-VH-G12); 88/168 (eg CD3-VH-G13); 96/168 (eg CD3-VH-G14); 104/168 (for example,

CD3-VH-G15); 112/168 (eg CD3-VH-G16); 120/168 (eg CD3-VH-G17); 128/168 (eg CD3-VH-G18); 136/168 (eg CD3-VH-G19); 144/168 (eg CD3-VH-G20) and 152/168 (eg CD3-VH-G21).

This invention also relates to antibodies or antigennegative fragments thereof, containing a set of six CDRs (ie HCDR1-HCDR2-HCDR3-LCDR1-LCDR2-LCDR3) contained in any of the illustrative anti-CD3 antibodies shown in table. 2 and 4. In some embodiments, the set of amino acid sequences HCDR1-HCDR2-HCDR3-LCDR1-LCDR2-LCDR3 is selected from the group consisting of SEQ ID NO: 12-14-16-164-166-168 (e.g., CD3-VH- G2); 20-22-24-164-166-168 (eg CD3-VH-G3); 28-30-32-164-166-168 (eg CD3-VH-G4);

36-38-40-164-166-168 (eg CD3-VH-G5); 44-46-48-164-166-168 (eg CD3-VH-G8);

52-54-56-164-166-168 (eg CD3-VH-G9); 60-62-64-164-166-168 (eg CD3-VH-G10);

68-70-72-164-166-168 (eg CD3-VH-G11); 76-78-80-164-166-168 (eg CD3-VH-G12);

84-86-88-164-166-168 (for example, CD3-VH-G13); 92-94-96-164-166-168 (eg CD3-VH-G14);

100-102-104-164-166-168 (eg CD3-VH-G15); 108-110-112-164-166-168 (eg CD3-VH-G16); 116-118-120-164-166-168 (eg CD3-VH-G17); 124-126-128-164-166-168 (eg CD3-VH-G18); 132-134-136-164-166-168 (eg CD3-VH-G19); 140-142-144-164-166-168 (eg CD3-VH-G20) and 148-150-152-164-166-168 (eg CD3-VH-G21).

In a related embodiment, the invention relates to antibodies, or antigen-binding fragments thereof, comprising a set of six CDRs (i.e., HCDR1-HCDR2-HCDR3-LCDR1LCDR2-LCDR3) contained in an HCVR/LCVR amino acid sequence pair, as defined by any of the illustrative anti -CD3 antibodies are given in table. 2 and 4. For example, the invention includes antibodies or antigen-binding fragments thereof comprising the amino acid sequence set HCDR1-HCDR2-HCDR3-LCDR1-LCDR2-LCDR3 contained in the HCVR/LCVR amino acid sequence pair selected from the group consisting of SEQ ID NO: 10/162 (eg CD3-VH-G2); 18/162 (eg CD3-VH-G3); 26/162 (eg CD3-VH-G4); 34/162 (eg CD3-VH-G5); 42/162 (eg CD3-VH-G8); 50/162 (eg CD3-VH-G9); 58/162 (eg CD3-VH-G10); 66/162 (eg CD3-VH-G11); 74/162 (eg CD3-VH-G12); 82/162 (eg CD3-VH-G13); 90/162 (eg CD3-VH-G14); 98/162 (eg CD3-VH-G15); 106/162 (eg CD3-VH-G16); 114/162 (eg CD3-VH-G17); 122/162 (eg CD3-VH-G18); 130/162 (eg CD3-VH-G19); 138/162 (eg CD3-VH-G20) and 146/162 (eg CD3-VH-G21).

Methods and techniques for identifying CDRs in HCVR and LCVR amino acid sequences are well known in the art and can be used to identify CDRs in the HCVR and/or LCVR amino acid sequences disclosed herein. Typical rules that can be used to identify CDR boundaries include, for example, Kabat definition, Chothia definition, and AbM definition. In general, the definition

- 3 040594

Kabat is based on sequence variability, the Chothia definition is based on the location of structural loop regions, and the AbM definition is a compromise between the Kabat and Chothia approaches. See, for example, Kabat, Sequences of Proteins of Immunological Interest, National Institutes of Health, Bethesda, Md. (1991); Al-Lazikani et al., J. Mol. Biol. 273:927-948 (1997); and Martin et al., Proc. Natl. Acad. sci. USA, 86:9268-9272 (1989). Public databases are also available to identify CDR sequences within an antibody.

This invention also relates to nucleic acid molecules encoding anti-CD3 antibodies or parts thereof. For example, this invention relates to nucleic acid molecules encoding any of the HCVR amino acid sequences shown in table. 3; in some embodiments, the implementation of the nucleic acid molecule contains a polynucleotide sequence selected from any of the HCVR nucleic acid sequences shown in table. 3, or a substantially similar sequence having at least 90%, at least 95%, at least 98%, or at least 99% sequence identity.

This invention also relates to nucleic acid molecules encoding any of the LCVR amino acid sequences shown in table. 4; or an LCVR derived from a cognate light chain, an anti-TAA heavy chain, or derived from a light chain that exhibits heterogeneity or the ability to pair with a wide variety of non-natural heavy chains, ie. universal or conventional light chain. In some embodiments, the implementation of the nucleic acid molecule contains a polynucleotide sequence selected from any of the LCVR nucleic acid sequences shown in table. 5, or a substantially similar sequence having at least 90%, at least 95%, at least 98%, or at least 99% sequence identity.

This invention also relates to nucleic acid molecules encoding any of the HCDR1 amino acid sequences shown in table. 2; in some embodiments, the nucleic acid molecule contains a polynucleotide sequence selected from any of the HCDR1 nucleic acid sequences shown in Table. 3, or a substantially similar sequence having at least 90%, at least 95%, at least 98%, or at least 99% sequence identity.

This invention also relates to nucleic acid molecules encoding any of the HCDR2 amino acid sequences shown in table. 2; in some embodiments, the implementation of the nucleic acid molecule contains a polynucleotide sequence selected from any of the HCDR2 nucleic acid sequences shown in table. 3, or a substantially similar sequence having at least 90%, at least 95%, at least 98%, or at least 99% sequence identity.

This invention also relates to nucleic acid molecules encoding any of the HCDR3 amino acid sequences shown in table. 2; in some embodiments, the nucleic acid molecule contains a polynucleotide sequence selected from any of the HCDR3 nucleic acid sequences shown in Table. 3, or a substantially similar sequence having at least 90%, at least 95%, at least 98%, or at least 99% sequence identity.

This invention also relates to nucleic acid molecules encoding any of the LCDR1 amino acid sequences shown in table. 4; or an LCDR1 derived from a cognate light chain, an anti-TAA heavy chain, or derived from a light chain that exhibits heterogeneity or the ability to pair with a wide variety of non-natural heavy chains, ie. universal or conventional light chain. In some embodiments, the nucleic acid molecule contains a polynucleotide sequence selected from any of the LCDR1 nucleic acid sequences shown in Table 1. 5, or a substantially similar sequence having at least 90%, at least 95%, at least 98%, or at least 99% sequence identity.

This invention also relates to nucleic acid molecules encoding any of the LCDR2 amino acid sequences shown in table. 4; or an LCDR2 derived from a cognate light chain, an anti-TAA heavy chain, or derived from a light chain that exhibits heterogeneity or the ability to pair with a wide variety of non-natural heavy chains, ie. universal or conventional light chain. In some embodiments, the nucleic acid molecule contains a polynucleotide sequence selected from any of the LCDR2 nucleic acid sequences shown in Table 1. 5, or a substantially similar sequence having at least 90%, at least 95%, at least 98%, or at least 99% sequence identity.

This invention also relates to nucleic acid molecules encoding any of the LCDR3 amino acid sequences shown in table. 4; or an LCDR3 derived from an anti-TAA heavy chain cognate light chain, or derived from a light chain that exhibits heterogeneity or the ability to pair with a wide variety of non-natural heavy chains, ie. 4 040594 universal or conventional light chain. In some embodiments, the nucleic acid molecule contains a polynucleotide sequence selected from any of the LCDR3 nucleic acid sequences shown in Table 1. 5, or a substantially similar sequence having at least 90%, at least 95%, at least 98%, or at least 99% sequence identity.

This invention also relates to nucleic acid molecules encoding HCVR, and HCVR contains a set of three CDRs (i.e. HCDR1-HCDR2-HCDR3), while the set of amino acid sequences HCDR1-HCDR2-HCDR3 is as defined by any illustrative anti- CD3 antibody shown in table. 2.

The invention also provides nucleic acid molecules encoding an LCVR, wherein the LCVR contains a set of three CDRs (i.e. LCDR1-LCDR2-LCDR3), wherein the set of amino acid sequences LCDR1-LCDR2-LCDR3 is as defined by any of the examples of the universal light chain antibodies shown in table. 4; or an LCDR1-LCDR2-LCDR3 derived from a cognate anti-TAA heavy chain light chain, or derived from a light chain exhibiting heterogeneity or the ability to pair with a wide variety of non-natural heavy chains, i. e. universal or conventional light chain.

This invention also relates to nucleic acid molecules encoding both HCVR and LCVR, and HCVR contains the amino acid sequence of any of the HCVR amino acid sequences shown in table. 2, and while LCVR contains the amino acid sequence of any of the amino acid sequences of the LCVR shown in table. 4; or LCVR is derived from a cognate anti-TAA heavy chain light chain, or derived from a light chain that exhibits heterogeneity or the ability to pair with a wide variety of non-natural heavy chains, i. e. universal or conventional light chain. In some embodiments, the implementation of the nucleic acid molecule contains a polynucleotide sequence selected from any of the HCVR nucleic acid sequences shown in table. 2, or a sequence substantially similar thereto, having at least 90%, at least 95%, at least 98%, or at least 99% sequence identity and a polynucleotide sequence selected from any of the LCVR nucleic acid sequences shown in tab. 5, or a substantially similar sequence having at least 90%, at least 95%, at least 98%, or at least 99% sequence identity. In some embodiments, nucleic acid molecules encoding both HCVR and LCVR are fully human sequences or are derived from human germline immunoglobulin sequences.

The invention also relates to recombinant expression vectors capable of expressing a polypeptide containing the heavy or light chain variable region of an anti-CD3 antibody. For example, this invention includes recombinant expression vectors containing any of the nucleic acid molecules mentioned above, ie. nucleic acid molecules encoding any of the sequences HCVR, LCVR and/or CDR, as indicated in table. 2 or 4. Also included within the scope of this invention are host cells into which such vectors have been introduced, as well as methods for producing antibodies or parts thereof by culturing host cells under conditions that allow the production of antibodies or antibody fragments, and recovering antibodies and antibody fragments. obtained in this way.

The present invention includes anti-CD3 antibodies and/or anti-TAA antibodies, as well as anti-CD3/anti-TAA bispecific antibodies having a modified glycosylation profile. In some embodiments, a modification to remove unwanted glycosylation sites may be useful, or an antibody lacking the fucose moiety present in the oligosaccharide chain, for example, to increase the function of antibody-dependent cellular cytotoxicity (ADCC) (see Shield et al. (2002) JBC 277:26733 ). In other applications, galactosylation can be modified to modify complement dependent cytotoxicity (CDC).

In one aspect, the invention provides a cytotoxic composition comprising a bispecific antigen-binding molecule that i) is unable to specifically bind to an effector cell, and ii) specifically binds a target tumor cell, the specific binding being measured in an in vitro FACS binding assay or an in vitro binding assay based on plasmon resonance. In some embodiments, the invention provides a cytotoxic composition comprising a bispecific antigen-binding molecule that exhibits no detectable binding to an effector cell and that specifically binds to a tumor target cell with a measurable binding affinity, wherein the binding affinity value is measured in an in vitro FACS binding assay or in vitro binding assay based on plasmon resonance.

In other embodiments, the invention provides a cytotoxic composition comprising a bispecific antigen-binding molecule comprising i) a first antigen-binding fragment (Fab1) that does not bind to CD3, and ii) a second antigen-binding fragment (Fab2) that specifically binds to a tumor target cell with a measurable binding affinity, wherein the binding affinity value is measured in an in vitro FACS binding assay or an in vitro plasmon resonance binding assay. In some instances, the binding affinity is monovalent binding affinity (eg, in a bispecific antibody construct).

In another aspect, the invention relates to a cytotoxic composition comprising a bispecific antigen-binding molecule that specifically binds an effector cell with a low binding affinity, e.g., manifested by an _EC50 value of about or greater than about 100 nM, and that specifically binds a target tumor cell with an appreciable EC value. ₅₀ or a high EC ₅₀ value with high affinity, such as less than 50 nM, wherein the EC ₅₀ binding affinity value is measured in an in vitro FACS binding assay. In some aspects, the invention relates to a cytotoxic composition comprising a bispecific antigen binding molecule that specifically binds an effector cell with an EC50 value greater than about 500 nM and that specifically binds a target tumor cell with a marked _EC50 value or a high _EC50 value with high affinity, such as less than 50 nm, and the value of the binding affinity EC ₅₀ is measured in an in vitro FACS binding assay.

In some examples, the bispecific antigen binding molecule comprises a Fab1 that specifically binds human CD3 with an EC ₅₀ value of greater than about 40 nM, or greater than about 100 nM, greater than about 200 nM, greater than about 300 nM, greater than about 400 nM, greater than about 500 nm or more than about 1 μm (for example, in the context of monovalent binding). In some embodiments, the bispecific antigen binding molecule comprises a Fab2 derived from a second antibody that specifically binds the target tumor cell with high affinity, e.g., an EC50 value of less than about 50 nM, less than about 40 nM, less than about 20 nM, less than about 10 nM or less than about 6 nM (eg, in the context of monovalent binding). In some instances, Fab1 specifically binds each of human CD3 and cynomolgus CD3 with an EC ₅₀ value of greater than about 40 nM, or greater than about 100 nM, greater than about 200 nM, or greater than about 1 μM. In some instances, Fab1 specifically binds each of human CD3 and cynomolgus monkey CD3 with little or no measurable affinity.

In some embodiments, the target tumor cell is a human tumor cell. In some embodiments, Fab1 (or a bispecific antigen-binding molecule) induces T cell mediated tumor cell lysis with an EC ₅₀ value of less than about 1.3 nM as measured in an in vitro T cell mediated tumor cell lysis assay.

In some applications, Fab1 or a bispecific antigen-binding molecule specifically binds human CD3 with a KD value of greater than about 11 nM, as measured in an in vitro binding assay based on surface plasmon resonance. In other instances, Fab1 or a bispecific antigen-binding molecule specifically binds each of human CD3 and cynomolgus CD3 with a KD value of greater than about 15 nM, or greater than about 30 nM, greater than about 60 nM, greater than about 120 nM, or greater than about 300 nM. , as measured in an in vitro binding assay based on surface plasmon resonance. In some other applications, the Fab1 or bispecific antigen-binding molecule i) exhibits no detectable binding to human CD3 as measured in each of the in vitro binding assays based on surface plasmon resonance and FACS binding assays, and ii) induces T-cell mediated tumor cell lysis , as measured in an in vitro T-cell mediated tumor cell lysis assay.

In some applications, the bispecific antigen-binding molecule comprises a first heavy chain containing an HCDR1 region containing the amino acid sequence shown in SEQ ID NO: 12 or 20. In some embodiments, the first heavy chain contains an HCDR2 region containing the amino acid sequence shown in SEQ ID NO: 14 or 54. In other embodiments, the first heavy chain comprises an HCDR3 region comprising the amino acid sequence shown in SEQ ID NO: 16, SEQ ID NO: 24, SEQ ID NO: 32, SEQ ID NO: 40, SEQ ID NO: 48 , SEQ ID NO: 56, SEQ ID NO: 64, SEQ ID NO: 72, SEQ ID NO: 80, SEQ ID NO: 88, SEQ ID NO: 96, SEQ ID NO: 104, SEQ ID NO: 112, SEQ ID NO: 120, SEQ ID NO: 128, SEQ ID NO: 136, SEQ ID NO: 144, or SEQ ID NO: 152. In other uses, the first heavy chain contains an HCVR containing HCDR1-HCDR2-HCDR3 having the amino acid sequences of SEQ ID NO: 178-179-180. In other embodiments, the first heavy chain comprises a CDR1 containing amino acid residues 1-7 of SEQ ID NO: 178, a CDR2 containing amino acid residues 1-7 of SEQ ID NO: 179, a CDR3 containing amino acid residues 4-11 of SEQ ID NO: 180.

In other embodiments, the first heavy chain comprises variable domain framework regions having an amino acid sequence selected from FR1 (SEQ ID NO: 174), FR2 (SEQ ID NO: 175), FR3 (SEQ ID NO: 176), and FR4 (SEQ ID NO: 176). NO: 177).

This invention relates to bispecific antigen-binding molecules containing

- 6 040594 pair of amino acid sequences HCVR and LCVR Fabl (HCVR/LCVR) selected from the group consisting of SEQ ID NO: 10/162; 18/162; 26/162; 34/162; 42/162; 50/162; 58/162; 66/162; 74/162; 82/162;

90/162; 98/162; 106/162; 114/162; 122/162; 130/162; 138/162; 146/162.

Antibodies and antigen-binding fragments and their bispecific antibodies were obtained by replacing the amino acid residues of the parent antibody in steps based on differences between the germline sequence and the sequence of the parent antibody. The present invention relates to a method for preparing a cytotoxic composition comprising (a) identifying the amino acid sequence of a first heavy chain derived from a first antibody that specifically binds to CD3 with high affinity, eg has a binding affinity EC ₅₀ value of less than about 40 nM; (b) modifying selected amino acid residues in the heavy chain variable region of the first antibody to obtain a modified antibody; (c) pairing the modified antibody with a second heavy chain derived from a second antibody that specifically binds the target tumor antigen to produce a bispecific antibody; (d) testing the bispecific antibody in a binding affinity assay, and if the binding affinity for CD3 has an EC ₅₀ value of greater than about 40 nM, or greater than about 100 nM, or greater than about 300 nM, or greater than about 500 nM, or no detectable binding then (e) obtaining a composition containing the bispecific antibody and a pharmaceutically acceptable carrier or diluent. In addition to modifying the heavy chain variable region of selected antibodies to design antigen binding arms having little or no affinity but specifically targeting the effector cell, the present invention provides methods for modifying the heavy chain constant region (e.g., CH3 domain) of each binding arm, a binder to prepare and isolation of bispecific antibodies.

An exemplary method provides a method for producing a cytotoxic bispecific antibody, comprising (a) identifying a first human antibody, or antigen-binding fragment thereof, that interacts with an effector cellular antigen from multiple species; (b) identifying the amino acid residues of the germline of the heavy chain variable region (HCVR) of the human antibody; (c) comparing the HCVR amino acid sequence of the first human antibody with the amino acid sequence of the corresponding HCVR germline; (d) identifying amino acids in the modified HCVR region of the first antibody, wherein the modified region in the first antibody exhibits at least one amino acid modification by substitution, deletion, or addition of one amino acid residue compared to the same region in the HCVR germline; (e) producing a plurality of modified antibodies, each containing at least one modified HCVR region; (f) screening each of the plurality of modified antibodies for monovalent affinity for an effector cell antigen; (g) selecting those modified antibodies that exhibit weaker binding affinity or no binding affinity for the effector cell antigen compared to the first antibody; and (h) pairing the selected first antibody with a second antibody that interacts with the tumor associated antigen to produce a cytotoxic bispecific antibody.

In another aspect, the invention provides a pharmaceutical composition comprising a recombinant human bispecific antibody or fragment thereof that specifically binds CD3 and a pharmaceutically acceptable carrier. In a related aspect, the invention relates to a composition that is a combination of an anti-CD3 antibody and a second therapeutic agent. In one embodiment, the second therapeutic agent is any agent that preferentially combines with an anti-CD3 antibody. Representative agents that may be advantageously combined with an anti-CD3 antibody include, without limitation, other agents that bind and/or activate CD3 signaling (including other antibodies or antigen-binding fragments thereof, etc.) and/or agents that do not directly bind CD3, but nevertheless activate or stimulate the activation of immune cells. Additional combination therapies and co-formulations comprising the anti-CD3 antibodies of this invention are disclosed elsewhere herein.

In yet another aspect, the invention provides therapeutic methods for stimulating T cell activation using an anti-CD3 antibody or an antigen-binding portion of an antibody of the invention, the therapeutic methods comprising administering a therapeutically effective amount of a pharmaceutical composition comprising the bispecific antibody of the invention or an antigen-binding fragment thereof to a subject in need of this. A treated disorder is any disease or condition that is ameliorated, alleviated, inhibited, or prevented by cytotoxic therapy targeting a tumor-associated antigen, such as cancer.

According to another aspect, the present invention relates to bispecific antigen-binding molecules that bind CD3 and a target antigen, especially tumor associated antigen (TAA).

- 7 040594

The invention also includes the use of the aHTu-CD3/aHTu-TAA bispecific antigen-binding molecule of the invention in the manufacture of a medicament for the treatment of a disease or disorder associated with or caused by TAA expression. The present invention also relates to the use of an anti-CD3/anti-TAA bispecific antigen-binding molecule exhibiting poor affinity for CD3-expressing effector cells and reduced clearance in the manufacture of a medicament for the treatment of a disease or disorder associated with or caused by TAA expression compared to anti -CD3/anti-TAA bispecific antigen-binding molecule showing high affinity for CD3-expressing effector cells.

Other embodiments will become apparent from a review of the following detailed description.

Brief description of the figures

In FIG. 1 shows the amino acid alignment of the following antibody heavy chain variable region (HCVR) sequences: hIgHV germline (SEQ ID NO: 181); CD3-VH-P (SEQ ID NO: 154); CD3-VH-G (SEQ ID NO: 2); CD3-VH-G2 (SEQ ID NO: 10); CD3-VH-G3 (SEQ ID NO: 18); CD3-VH-G4 (SEQ ID NO: 26); CD3-VH-G5 (SEQ ID NO: 34); CD3-VH-G8 (SEQ ID NO: 42); CD3-VH-G9 (SEQ ID NO: 50); CD3-VH-G10 (SEQ ID NO: 58); CD3-VH-G11 (SEQ ID NO: 66); CD3-VH-G12 (SEQ ID NO: 74); CD3-VH-G13 (SEQ ID NO: 82); CD3-VH-G14 (SEQ ID NO: 90); CD3-VH-G15 (SEQ ID NO: 98); CD3-VH-G16 (SEQ ID NO: 106); CD3-VH-G17 (SEQ ID NO: 114); CD3-VH-G18 (SEQ ID NO: 122); CD3-VH-G19 (SEQ ID NO: 130); CD3-VH-G20 (SEQ ID NO: 138) and CD3-VH-G21 (SEQ ID NO: 146). Each HCVR derivative is compared to the parent antibody and germline amino acid residues, with rectangular boxes representing mutations in the CDR.

In FIG. 2A-2C are mean serum total IgG concentrations after a single intraperitoneal injection of 0.4 mg/kg BSMUC16/CD3-001, BSMUC16/CD3-005, and isotype control antibodies in WT (FIG. 2A), CD3 humanized mice (FIG. .2B) and MUC16xCD3 humanized mice (Fig. 2C).

Detailed description

Before the present invention is described, it should be understood that the present invention is not limited to the particular methods and experimental conditions described, as such methods and conditions may vary. It should also be understood that the terminology used herein is only intended to describe specific embodiments and is not intended to be limiting, as the scope of this invention will only be limited by the appended claims.

Unless otherwise indicated, all technical and scientific terms used in this document have the same meaning as generally understood by a person skilled in the art to which this invention pertains. Used in this document, the term about when used in relation to a specific specified numerical value means that the value may differ from the specified value by no more than 1%. For example, the expression about 100 as used herein includes 99 and 101 and all values in between (eg, 99.1, 99.2, 99.3, 99.4, etc.).

While any methods and materials similar or equivalent to those described herein may be used in the practice or testing of this invention, the following describes the preferred methods and materials.

Definitions

CD3 expression refers to an antigen that is expressed on T cells as part of the multimolecular T cell receptor (TCR) and that consists of a homodimer or heterodimer formed from the association of two of the four receptor chains: CD3 epsilon, CD3 delta, CD3 zeta and CD3 gamma.

Human CD3 epsilon (hCD3ε) contains the amino acid sequence shown in SEQ ID NO: 169 (UniProtKB/Swiss-Prot: P07766.2). Human CD3 delta (hCD3δ)) contains the amino acid sequence shown in SEQ ID NO: 170 (UniProtKB/Swiss-Prot: P04234.1). All references to proteins, polypeptides, and protein fragments herein are intended to refer to the human version of the corresponding protein, polypeptide, or protein fragment, unless explicitly stated to be of a non-human species. Thus, the expression CD3 means human CD3 unless indicated to refer to a non-human species, eg mouse CD3, monkey CD3, etc.

The phrase antibody that binds to CD3 or anti-CD3 antibody includes antibodies and their antigen-binding fragments that specifically recognize and bind to one subunit of CD3 (e.g., epsilon, delta, gamma, or zeta), as well as antibodies and their antigen-binding fragments that specifically recognize and bind to a dimeric complex of two CD3 subunits (eg, epsilon/delta, epsilon/gamma, and zeta/zeta CD3 dimers). Antibodies and antigen bonds

- 8 040594 fragments according to the invention can bind soluble CD3, bound CD3 and/or cell surface expressing CD3. Soluble CD3 includes native CD3 proteins as well as variants of the recombinant CD3 protein, such as, for example, CD3 monomeric and dimeric constructs that lack a transmembrane domain or are not otherwise associated with the cell membrane. The present invention relates to antibodies that bind and activate human and cynomolgus monkey CD3 with weak or undetectable binding affinity. Binding to CD3 with a non-detectable binding affinity means that the interaction of an antibody and/or antigen-binding fragment with a CD3 target may not be measurable or detectable by a known detection assay, such as the FACS binding assay (cell-based) described herein, or the surface plasmon resonance binding assay described herein and well known in the art. Other binding assays are well known in the art. The antibody and/or antigen-binding fragment can recognize the CD3 target through a very weak protein-protein biochemical interaction, but determination of a specific KD or _EC50 value cannot be measured because the interaction is beyond the detection of the assay, e.g., cannot be measured. Otherwise, no detectable binding affinity is determined if the affinity of the antibody corresponding to the KD value is equal to or less than 10 times lower than a non-specific antigen such as BSA, casein, or the like. Binding to CD3 with low binding affinity includes interactions in which the measurement of binding affinity is at or slightly above the limit of detection of the assay, or equivalent to the binding affinity for a non-specific antigen.

Cell surface expressed CD3 expression means one or more CD3 proteins that are expressed on the cell surface in vitro or in vivo such that at least a portion of the CD3 protein resides on the extracellular side of the cell membrane and is accessible to the antigen-binding portion of the antibody. CD3 expressed on the cell surface includes CD3 proteins contained in the context of a functional T cell receptor in the cell membrane. Cell surface expression of CD3 includes CD3 protein expressed as part of a cell surface homodimer or heterodimer (eg, CD3 delta/epsilon, gamma/epsilon, and zeta/zeta dimers). Expression of CD3 expressed on the cell surface also includes a CD3 chain (eg CD3 delta, CD3 epsilon or CD3 gamma) that is expressed by itself without other types of CD3 chain on the cell surface. CD3 expressed on the cell surface may contain or consist of the CD3 protein expressed on the cell surface, which normally expresses the CD3 protein. Alternatively, cell surface expressed CD3 may comprise or consist of a CD3 protein expressed on the surface of a cell that does not normally express human CD3 on its surface but is artificially engineered to express CD3 on its surface.

Effector cells include effector T cells (T lymphocytes), such as CD4+ T cells, CD8+ T cells, Th1, Th2, and regulatory T cells (Tregs). Effector cells may also include natural killer (NK) cells, macrophages, granulocytes, plasma cells, or B cells (lymphocytes). It is understood that therapy can mediate a variety of cellular immune responses or effector functions through Ig interaction with effector surface receptors such as CD3 (T cell surface receptor), CD28 (T cells), Fcy receptors (FcyRs) (NK cells activated macrophages, etc.). As a result of these interactions, effector functions such as cell lysis, complement activation, phagocytosis, and opsonization are activated. Binding to the tumor effector cell and tumor target cell provides a valuable and effective immunotherapy design that extends to tumor cell lysis and induces endogenous immune functions to fight the tumor or cancer.

The expression of an anti-CD3 antibody includes both monovalent antibodies with one specificity and bispecific antibodies containing a first arm that binds CD3 and a second arm that binds a second (target) antigen, in which the anti-CD3 arm contains any of the HCVR/LCVR sequences or CDR presented in table. 2-4 and/or 5 in this document. Examples of bispecific anti-CD3 antibodies are described elsewhere in this document. The term antigen-binding molecule includes antibodies and antigen-binding fragments of antibodies, including, for example, bispecific antibodies.

The term antibody includes any antigen-binding molecule or molecular complex containing at least one complementarity determining region (CDR) that specifically binds to or interacts with a particular antigen (eg, CD3). The term antibody includes immunoglobulin molecules containing four polypeptide chains, two heavy (H) chains and two light (L) chains linked by disulfide bonds, as well as their multimers (eg, IgM). Each heavy chain includes a heavy chain variable region (abbreviated herein as HCVR or VH) and a heavy chain constant region. The heavy chain constant region consists of three domains, CH1, C _H 2 and C _H 3. Each light chain contains a light chain variable region (abbreviated as LCVR or VL) and a light chain constant region.

- 9 040594 chain. The light chain constant region contains one domain (C _L 1). The VH and V _L regions can be further subdivided into regions of hypervariability called complementarity determining regions (CDRs) interspersed with more conserved regions called framework regions (FRs). Each VH and V _L consists of three CDRs and four FRs arranged from amino to carboxyl in the following order: FR1, CDR1, FR2, CDR2, FR3, CDR3, FR4. In various embodiments, the FR of an anti-CD3 antibody (or antigen-binding portion thereof) may be identical to human germline sequences, or may be naturally or artificially modified. The consensus amino acid sequence can be determined based on a parallel analysis of two or more CDRs.

The term antibody also includes antigen-binding fragments of complete antibody molecules. The terms antigen-binding portion of an antibody, antigen-binding fragment of an antibody, and the like, as used herein, include any naturally occurring, enzymatically available synthetic or genetically modified polypeptide or glycoprotein that specifically binds an antigen to form a complex. Antigen-binding antibody fragments can be generated, for example, from whole antibody molecules using any suitable standard techniques such as proteolytic cleavage or recombinant genetic engineering techniques involving the manipulation and expression of DNA encoding the variable and optionally constant domains of an antibody. Such DNA is known and/or readily available, for example from commercial sources, DNA libraries (including, for example, phage antibody libraries), or can be synthesized. The DNA can be sequenced and processed chemically or using molecular biology techniques, for example, to organize one or more variable and/or constant domains into a suitable configuration, or to introduce codons, create cysteine residues, modify, add or remove amino acids, etc.

Non-limiting examples of antigen-binding fragments include (i) Fab fragments; (ii) F(ab') ₂ fragments; (iii) Fd fragments; (iv) Fv fragments; (v) single chain Fv molecules (scFv); (vi) dAb fragments; and (vii) minimal recognition units consisting of amino acid residues that mimic an antibody hypervariable region (e.g., a distinguished complementarity determining region (CDR) such as a CDR3 peptide) or a FR3-CDR3- FR4. Other engineered molecules such as domain specific antibodies, single domain antibodies, domain deleted antibodies, chimeric antibodies, CDR grafted antibodies, diabodies, tribodies, tetrabodies, minibodies, nanoparticles (e.g., monovalent nanoparticles, divalent nanoparticles, etc.) .), small modular immunopharmaceuticals (SMIPs), and shark IgNAR variable domains are also included in the expression antigen-binding fragment used herein.

An antigen binding fragment of an antibody typically contains at least one variable domain. The variable domain may be of any size or amino acid composition and will typically contain at least one CDR that is adjacent to or in association with one or more backbone sequences. In antigen-binding fragments having a V _H domain associated with a V _L domain, the V _H and V _L domains may be located relative to each other in any suitable arrangement. For example, the variable region may be dimeric and contain V _H -V _H , V _H -V _L or V _L -V _L dimers. Alternatively, the antigen-binding fragment of an antibody may comprise a V _H or V _L monomeric domain.

In some embodiments, an antigen-binding fragment of an antibody may comprise at least one variable domain covalently linked to at least one constant domain. Non-limiting exemplary variable and constant domain configurations that may be found in an antigen-binding fragment of an antibody of the present invention include (i) Vh-Ch1; (ii) Vh-Ch2; (iii) Vh-Ch3; (iv) Vh-Ch1-Ch; (v) Vh-Ch1-Ch2-Ch3; (vi) Vh-Ch2-Ch3; (vii) Vh-Cl; (viii) Vl-Ch1; (ix) Vl-Ch2; (x) Vl-Ch3; (xi) Vl-Ch1-Ch2; (xii) Vl-Ch1-Ch2-Ch3; (xiii) Vl-Ch2C _h 3 and (xiv) V _l -C _l . In any configuration of the variable and constant domains, including any of the above illustrative configurations, the variable and constant domains may either be directly linked to each other or may be linked through a complete or partial hinge or linker region. The hinge region may be at least 2 (eg, 5, 10, 15, 20, 40, 60 or more) amino acids that result in a flexible or semi-flexible bond between adjacent variable and/or constant domains in a single polypeptide molecule. In addition, an antigen-binding fragment of an antibody of the present invention may comprise a homodimer or heterodimer (or other multimer) of any of the variable and constant domain configurations listed above in non-covalent association with each other and/or with one or more VH or V _l monomeric domains ( for example, via disulfide bond(s)).

As with full antibody molecules, antigen binding fragments can be monospecific or multispecific (eg, bispecific).

A multispecific antigen-binding fragment of an antibody will typically contain at least two different variable domains, each variable domain being capable of specifically binding to a separate antigen or to a different epitope on the same antigen. Any multispecific antibody format, including the exemplary bispecific antibody formats described herein, can be adapted for use in the context of an antigen-binding fragment of an antibody of this invention using conventional techniques available in the art.

The antibodies of this invention can function through complement dependent cytotoxicity (CDC) or antibody dependent cell mediated cytotoxicity (ADCC). Complement dependent cytotoxicity (CDC) refers to the lysis of antigen expressing cells by an antibody of the invention in the presence of complement. Antibody-dependent cell-mediated cytotoxicity (ADCC) refers to a cell-mediated response in which non-specific cytotoxic cells that express Fc receptors (FcRs) (e.g., natural killer (NK) cells, neutrophils, and macrophages) recognize the bound antibody on the cell- target and thus leads to lysis of the target cell. CDC and ADCC can be measured using assays that are well known and available in the art. (See, for example, US Pat. Nos. 5,500,362 and 5,821,337 and Clynes et al. (1998) Proc. Natl. Acad. Sci. (USA) 95:652-656). The constant region of an antibody is important in the ability of an antibody to fix complement and mediate cell-dependent cytotoxicity. Thus, the antibody isotype can be selected based on whether it is desirable for the antibody to mediate cytotoxicity.

In some embodiments, the anti-CD3 antibodies of the invention (monospecific or bispecific) are human antibodies.

As used herein, the term human antibody is intended to include antibodies having variable and constant regions derived from human germline immunoglobulin sequences. Human antibodies of the invention may include amino acid residues not encoded by germline human immunoglobulin sequences (e.g., mutations introduced by in vitro random or site-directed mutagenesis or in vivo somatic mutation), for example, in CDRs and in particular CDR3. However, the term human antibody as used herein is not intended to include antibodies in which CDR sequences derived from the germline of other mammalian species, such as the mouse, have been grafted onto human scaffold sequences.

The term recombinant human antibody is intended to include all human antibodies that are produced, expressed, generated, or isolated by recombinant means, such as antibodies expressed using a recombinant expression vector transfected into a host cell (described below), antibodies isolated from a recombinant combinatorial library human antibodies (described below), antibodies isolated from an animal (e.g., mouse) that is transgenic for human immunoglobulin genes (see, e.g., Taylor et al. (1992), Nucl. Acids Res. 20:6287-6295) or antibodies produced, expressed, made or isolated by any other method which involves splicing human immunoglobulin gene sequences into other DNA sequences. Such recombinant human antibodies have variable and constant regions derived from human germline immunoglobulin sequences. However, in some embodiments, such human recombinant antibodies undergo in vitro mutagenesis (or, when a transgenic animal for human Ig sequences is used, in vivo somatic mutagenesis), and therefore the amino acid sequences of the VH and V _L regions of the recombinant antibody are sequences that , being derived from and related to the human germline VH and V _L sequences, do not necessarily naturally exist in the human germline antibody repertoire in vivo.

Human antibodies can exist in two forms that are associated with hinge heterogeneity. In one form, the immunoglobulin molecule contains a stable four-strand construct containing approximately 150-160 kDa, in which the dimers are held together by an interstrand disulfide bond. In the second form, the dimers are not linked through interchain disulfide bonds, and a molecule of about 75-80 kDa is formed, consisting of a covalently linked light and heavy chain (half-antibody). These forms are extremely difficult to separate even after affinity purification.

The frequency of appearance of the second form in various intact IgG isotypes is due to, but not limited to, structural differences associated with the isotype of the hinge region of the antibody. A single amino acid substitution in the hinge region of the human IgG4 hinge can significantly reduce the appearance of the second form (Angal et al. (1993), Molecular Immunology, 30:105) to levels normally seen with the human IgG1 hinge. This invention encompasses antibodies having one or more mutations in the hinge region, CH2 or CH3, which may be desirable, for example in production, to improve the yield of the desired antibody form.

The antibodies of the invention may be isolated antibodies. An isolated antibody as used herein means an antibody that has been identified and isolated and/or reconstituted from at least one component of its natural environment. For example, an antibody that has been isolated or removed from at least one component of an organism, or from a tissue or cell in which the antibody naturally exists or is naturally produced, is an isolated antibody for the purposes of this invention. An isolated antibody also includes an antibody in situ in a recombinant cell. Isolated antibodies are antibodies that have been subjected to at least one purification or isolation step. In accordance with some embodiments, the isolated antibody may be substantially free of other cellular material and/or chemicals.

The invention also includes single-armed antibodies that bind CD3. The phrase single-armed antibody means an antigen-binding molecule comprising one antibody heavy chain and one antibody light chain. Single-armed antibodies of the invention may contain any of the HCVR/LCVR or CDR amino acid sequences shown in Table 1. 2 in this document.

The term epitope refers to an antigenic determinant that interacts with a specific antigen binding site in the variable region of an antibody molecule known as a paratope. One antigen may have more than one epitope. Thus, different antibodies may bind to different regions on an antigen and may have different biological effects. Epitopes can be either conformational or linear. A conformational epitope is produced by spatially aligned amino acids from different segments of the linear polypeptide chain. A linear epitope is one obtained by adjacent amino acid residues in a polypeptide chain. In certain circumstances, an epitope may contain fragments of saccharides, phosphoryl groups, or sulfonyl groups on the antigen.

The term substantial identity or substantially identical, when referring to a nucleic acid or fragment thereof, indicates that, when optimally aligned with the corresponding nucleotide insertions or deletions with another nucleic acid (or its complementary strand), there is a nucleotide sequence identity of at least about 95% and more preferably at least about 96, 97, 98, or 99% nucleotide bases, as measured by any known sequence identity algorithm such as FASTA, BLAST, or Gap, as discussed below. A nucleic acid molecule that has a significant identity with a reference nucleic acid molecule may, in some cases, encode a polypeptide having the same or substantially similar amino acid sequence as the polypeptide encoded by the reference nucleic acid molecule.

When applied to polypeptides, the term significant similarity or substantially similar means that two peptide sequences when optimally aligned, such as the GAP or BESTFIT programs using default difference masses, have at least 95% sequence identity, even more preferably at least 98 or 99% sequence identity. Preferably, residue positions that are not identical differ by conservative amino acid substitutions. A conservative amino acid substitution is one in which an amino acid residue is replaced by another amino acid residue having a side chain (R group) with similar chemical properties (eg, charge or hydrophobicity). In general, a conservative amino acid substitution will not significantly alter the functional properties of a protein. In cases where two or more amino acid sequences differ from each other by conservative substitutions, the percent sequence identity or degree of similarity may be adjusted upward to correct for the conservative nature of the substitution. Means for effecting this adjustment are well known to those skilled in the art. See, for example, Pearson (1994), Methods Mol. Biol. 24:307-331. Examples of groups of amino acids having side chains with similar chemical properties include (1) aliphatic side chains: glycine, alanine, valine, leucine, and isoleucine; (2) aliphatic hydroxyl side chains: serine and threonine; (3) amide-containing side chains: asparagine and glutamine; (4) aromatic side chains: phenylalanine, tyrosine and tryptophan; (5) main side chains: lysine, arginine and histidine; (6) acidic side chains: aspartate and glutamate; and (7) sulfur-containing side chains are cysteine and methionine. Preferred conservative amino acid substituents are valine-leucine isoleucine, phenylalanine-tyrosine, lysine-arginine, alanine-valine, glutamate-aspartate, and asparaginglutamine. Alternatively, a conservative substitution is any change that has a positive value in the PAM250 log-likelihood matrix disclosed in Gonnet et al. (1992), Science, 256:1443-1445. A moderately conservative replacement is any change that has a non-negative value in the PAM250 log-likelihood matrix.

Sequence similarity for polypeptides, also referred to as sequence identity, is commonly measured using sequence analysis software. Protein analysis software matches similar sequences using similarity measures assigned to various substitutions, deletions, and other modifications, including conservative amino acid substitutions. For example, the GCG software includes programs such as Gap and Bestfit that can be used with default settings.

- 12 040594 to determine sequence homology or sequence identity between closely related polypeptides, such as homologous polypeptides from different species, or between a wild-type protein and its mutein. See, for example, GCG version 6.1. Polypeptide sequences can also be compared using FASTA using the standard or recommended parameters program in GCG version 6.1. FASTA (eg, FASTA2 and FASTA3) provides alignment and percent sequence identity of regions of best overlap between search sequences (Pearson (2000) supra). Another preferred algorithm for comparing a sequence of the invention with a database containing a large number of sequences from different organisms is the BLAST computer program, especially BLASTP or TBLASTN, using default parameters. See, for example, Altschul et al. (1990), J. Mol. Biol. 215:403-410 and Altschul et al. (1997), Nucleic Acids Res. 25:3389-402.

Germline mutations.

The anti-CD3 antibodies provided herein contain one or more amino acid substitutions, insertions and/or deletions in the framework and/or CDR regions of the heavy chain variable domains compared to the corresponding germline sequences from which the antibodies were derived.

The invention also includes antibodies and antigen-binding fragments thereof that are derived from any of the amino acid sequences disclosed herein, wherein one or more amino acids in one or more framework and/or CDR regions are mutated in the corresponding residue(s) of the germline sequence. from which the antibody was derived, or in the corresponding residue(s) of another human germline sequence, or in a conservative amino acid substitution of the corresponding germline residue(s) (such sequence changes are referred to herein collectively as germline mutations) and with little or no detectable binding to the CD3 antigen. Several such exemplary antibodies that recognize CD3 are described in Table 1. 2 in this document.

In addition, the antibodies of this invention may contain any combination of two or more germline mutations within the framework and/or CDR regions, for example, where certain individual residues mutate in the corresponding residue of a particular germline sequence, while some other residues, that differ from the original germline sequence are retained or mutated with the corresponding residue of another germline sequence. Once generated, antibodies and antigen-binding fragments that contain one or more germline mutations can be tested for one or more desired properties such as improved binding specificity, weak or reduced binding affinity, improved or increased pharmacokinetic properties, decreased immunogenicity, etc. . Antibodies and antigen-binding fragments obtained in this general form, subject to the guidance of this invention are covered in this invention.

The invention also includes anti-CD3 antibodies containing variants of any of the HCVR, LCVR and/or CDR amino acid sequences described herein having one or more conservative substitutions. For example, the invention includes anti-CD3 antibodies having the amino acid sequences of HCVR, LCVR, and/or CDR, e.g., 10 or less, 8 or less, 6 or less, 4 or less, etc. conservative amino acid substitutions compared to any of the amino acid sequences of HCVR, LCVR and/or CDR presented in table. 2 in this document. The antibodies and bispecific antigen-binding molecules of this invention comprise one or more amino acid substitutions, insertions and/or deletions in the framework and/or CDR regions of the heavy and light chain variable domains, as compared to the corresponding germline sequences from which the individual antigen-binding domains were derived, while maintaining or improving the desired detectable binding to the CD3 antigen. A conservative amino acid substitution is one in which an amino acid residue is replaced by another amino acid residue having a side chain (R group) with similar chemical properties (eg, charge or hydrophobicity). As a rule, conservative amino acid substitution will not significantly change the functional properties of the protein, i.e. the amino acid substitution maintains or improves the desired low and undetectable binding affinity for anti-CD3 binding molecules. Examples of groups of amino acids having side chains with similar chemical properties include (1) aliphatic side chains: glycine, alanine, valine, leucine, and isoleucine; (2) aliphatic hydroxyl side chains: serine and threonine; (3) amide-containing side chains: asparagine and glutamine; (4) aromatic side chains: phenylalanine, tyrosine and tryptophan; (5) main side chains: lysine, arginine and histidine; (6) acidic side chains: aspartate and glutamate; and (7) sulfur-containing side chains are cysteine and methionine. Preferred conservative amino acid substituents are valine-leucine isoleucine, phenylalanine-tyrosine, lysine-arginine, alanine-valine, glutamate-aspartate, and asparaginglutamine. Alternatively, a conservative substitution is any change that has a positive value in the PAM250 log-likelihood matrix disclosed in Gonnet et al.

- 13 040594 (1992), Science, 256:1443-1445. A moderately conservative replacement is any change that has a non-negative value in the PAM250 log-likelihood matrix.

The invention also includes antigen-binding molecules comprising an antigen-binding domain with an HCVR and/or CDR amino acid sequence that is substantially identical to any of the HCVR and/or CDR amino acid sequences disclosed herein, while maintaining or improving the desired low affinity for the CD3 antigen. The term substantial identity, or substantially identical when referring to an amino acid sequence, means that two amino acid sequences when optimally aligned, such as the GAP or BESTFIT programs using the default difference masses, have at least 95% sequence identity, even more preferably at least 98 or 99% sequence identity. Preferably, residue positions that are not identical differ by conservative amino acid substitutions. In cases where two or more amino acid sequences differ from each other by conservative substitutions, the percent sequence identity or degree of similarity may be adjusted upward to correct for the conservative nature of the substitution. Means for effecting this adjustment are well known to those skilled in the art. See, for example, Pearson (1994), Methods Mol. Biol. 24:307-331.

Sequence similarity for polypeptides, also referred to as sequence identity, is commonly measured using sequence analysis software. Protein analysis software matches similar sequences using similarity measures assigned to various substitutions, deletions, and other modifications, including conservative amino acid substitutions. For example, the GCG software includes programs such as Gap and Bestfit that can be used with default parameters to determine sequence homology or sequence identity between closely related polypeptides, such as homologous polypeptides from different species, or between a wild-type protein and its mutein. See, for example, GCG version 6.1. Polypeptide sequences can also be compared using FASTA using standard or recommended parameters, the program in GCG version 6.1. FASTA (eg, FASTA2 and FASTA3) provides alignment and percent sequence identity of regions of best overlap between search sequences (Pearson (2000) supra). Another preferred algorithm for comparing a sequence of the invention with a database containing a large number of sequences from different organisms is the BLAST computer program, especially BLASTP or TBLASTN, using default parameters. See, for example, Altschul et al. (1990), J. Mol. Biol. 215:403-410 and Altschul et al. (1997), Nucleic Acids Res. 25:3389-402.

Once generated, antigen binding domains that contain one or more germline mutations were tested for decreased binding affinity using one or more in vitro assays. Although antibodies that recognize a particular antigen are usually screened for by testing for high (ie, strong) binding affinity for the antigen, the antibodies of this invention show little or no detectable binding. Bispecific antigen-binding molecules containing one or more antigen-binding domains obtained in this general method are also within the scope of this invention and have been found to be preferred as avidity-induced tumor therapy.

Unexpected benefits, such as improved pharmacokinetic properties and low toxicity to the patient, can be realized from the methods described herein.

Binding properties of antibodies.

As used herein, the term binding in the context of the binding of an antibody, immunoglobulin, antibody-binding fragment, or Fc-containing protein to any, for example, a given, antigen, such as a cell surface protein or fragment thereof, generally refers to an interaction or association between at least two entities or molecular structures such as antibody-antigen interaction.

For example, the binding affinity typically corresponds to a K _D value of about 10 ^-7 M or less, for example, about 10 ^-8 M or less, for example, about 10 ^-9 M or less, when determined, for example, by surface plasmon resonance (SPR) on an instrument. Biacore 3000 using antigen as ligand and antibody, Ig, antibody-binding fragment, or Fc-containing protein as analyte (or anti-ligand). Cell-based binding strategies such as flow cytometry (FACS) binding assays are also widely used, and FACS data correlate well with other methods such as radioligand binding and SPR (Benedict, CA, J. Immunol. Methods. 1997, 201(2):223-31; Geuijen, CA, et al J. Immunol Methods 2005, 302(1-2):68-77).

Accordingly, an antibody or antigen-binding protein of the invention can bind to a predetermined antigen or cell surface molecule (a receptor such as CD3) having an affinity corresponding to a K _D value that is at least 10 times lower than its binding affinity for a non-specific antigen (e.g., BSA, casein). In accordance with this invention, if the affinity of the antibody corresponding to the K _D value is equal to or less than 10 times lower than

- 14 040594 non-specific antigen, this can be considered as non-detectable binding, however, such an antibody can be paired with a second antigen-binding arm to obtain a bispecific antibody of the invention.

The term K _D (M) refers to the dissociation equilibrium constant of a particular antibody-antigen interaction or the dissociation equilibrium constant of an antibody or antibody-binding fragment that binds to an antigen. There is an inverse relationship between K _D and binding affinity, so the smaller the value of K _D , the higher, i.e. stronger, affinity. Thus, the terms higher affinity or stronger affinity refer to a higher ability to form an interaction and therefore a lower K _D value, and conversely, the terms low affinity or weaker affinity refer to a lower ability to form an interaction and therefore larger value of K _D . In some cases, the higher binding affinity (or K _D ) of a particular molecule (for example, an antibody) for its interactive partner molecule (for example, antigen X) compared to the binding affinity of the molecule (for example, antibody) for another interactive partner molecule (for example, antigen Y) can be expressed as a binding ratio, determined by dividing the larger K _D value (lower or weaker affinity) by the smaller K _D (higher or stronger affinity), for example, expressed as 5-fold or 10-fold greater binding depending on from circumstances. For example, low affinity refers to less strong binding. In some embodiments, low binding affinity corresponds to greater than about 1 nM K _D greater than about 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17 , 18, 19, 20, 25, 30, 35, or greater than about 40 nM K _D , such K _D binding affinities being measured in an in vitro plasmon resonance binding assay or a biomolecular interaction sensitivity assay. In some embodiments, low binding affinity corresponds to greater than about 10 nM EC ₅₀ , greater than about 15 nM EC ₅₀ , 20 nM EC ₅₀ , greater than about 25 nM EC ₅₀ , 30 nM EC ₅₀ , greater than about 35 nM EC ₅₀ , or greater than about 40 nM EC ₅₀ , such EC ₅₀ binding affinity value being measured in an in vitro FACS binding assay or equivalent cell-based binding assay. Weak affinity refers to a weak binding interaction. In some embodiments, weak binding affinity corresponds to greater than about 100 nM K _D or EC ₅₀ , greater than about 200, 300, or greater than about 500 nM K _D or EC ₅₀ , such K _D binding affinity value being measured in an in vitro assay for based on surface plasmon resonance binding or an equivalent biomolecular interaction sensitivity assay, and such an EC ₅₀ binding affinity value is measured in an in vitro FACS binding assay or an equivalent cell-based interaction detection assay to detect monovalent binding. Undetectable binding means that the affinity between two biomolecules, for example especially between the binding arm of a monovalent antibody and its target antigen, exceeds the limits of detection of the assay used.

The term kd (c ^-1 or 1/s) refers to the dissociation rate constant of a particular antibody-antigen interaction or the dissociation rate constant of an antibody or binding antibody fragment. This value is also referred to as the kf value.

The term _ka (M' ¹ xc' ¹ or 1/M) refers to the association rate constant of a particular antibody-antigen interaction or the association rate constant of an antibody or binding antibody fragment.

The term _ka (M ^-1 or 1/M) refers to the equilibrium association constant of a particular antibody-antigen interaction or the equilibrium association constant of an antibody or binding antibody fragment. The association equilibrium constant is obtained by dividing k _a by k _d .

The term EC ₅₀ refers to half the maximum effective concentration, which includes the concentration of an antibody that elicits a response halfway between baseline and maximum after a certain exposure time. EC ₅₀ essentially represents the concentration of an antibody at which 50% of its maximum effect is observed. In some embodiments, the EC ₅₀ value is equal to the concentration of an antibody of the invention that provides half-maximal binding to cells expressing CD3 or tumor associated antigen, as determined, for example, in a FACS binding assay.

Thus, reduced or weaker binding is observed with an increased EC ₅₀ value or half the maximum effective concentration value, such that 500 nM EC ₅₀ indicates weaker binding affinity than 50 nM EC ₅₀ .

In one embodiment, the decrease in binding can be defined as an increased concentration of EC ₅₀ antibodies, which allows you to contact half the maximum number of target cells.

In other experimental measurements, the EC ₅₀ value is the concentration of an antibody of the invention that causes half-maximal depletion of target cells via T cell cytotoxic activity. Thus, increased cytotoxic activity (eg, tumor cell lysis mediated by T cells) is observed with a decrease in the EC ₅₀ value or half of the maximum effective concentration.

- 15 040594

Bispecific antigen-binding molecules.

The antibodies of this invention may be bispecific or multispecific. Multispecific antibodies may be specific for a single effector molecule, such as CD3, in combination with different epitopes of the same target polypeptide, or may contain antigen-binding domains specific for more than one target polypeptide. See, for example, Tutt et al., 1991, J. Immunol. 147:60-69; Kufer et al., 2004, Trends Biotechnol. 22:238-244. The Ahtu-CD3 antibodies of this invention may be linked to or co-expressed with another functional molecule, such as another peptide or protein. For example, an antibody or antibody fragment may be operably linked (e.g., by chemical bonding, genetic synthesis, non-covalent association, or otherwise) to one or more other molecular entities, such as another antibody or antibody fragment, to produce a bispecific or multispecific antibody with a second binding specificity.

As used herein, the term anti-CD3 antibody is intended to include both monospecific anti-CD3 antibodies and bispecific antibodies comprising a CD3-binding arm and a second arm that binds the target antigen. Thus, the invention includes bispecific antibodies in which one arm of the immunoglobulin binds to human CD3 and the other arm of the immunoglobulin is specific for the target antigen. The target antigen that binds to the other arm of the bispecific anti-CD3 antibody can be any antigen expressed on or near a cell, tissue, organ, microorganism, or virus against which a targeted immune response is desired. The CD3 binding arm may contain any of the HCVR or CDR amino acid sequences listed in Table 1. 2 in this document. In some embodiments, the CD3 binding arm binds weakly to human CD3 and induces the activation of human T cells. In other embodiments, the CD3-binding arm binds weakly to human CD3 and induces tumor-associated antigen-expressing cell lysis in the context of a bispecific or multispecific antibody. In other embodiments, the CD3 binding arm binds or associates weakly with human and cynomolgus monkey (monkey) CD3, but no binding interaction is detected using in vitro assays known in the art. In some embodiments, the CD3-binding arm does not bind or associate with human and cynomolgus monkey (monkey) CD3, but the bispecific molecule still causes tumor-associated cell lysis.

In the context of the bispecific antibodies of the invention, in which one arm of the antibody binds CD3 and the other arm binds the target antigen, the target antigen may be a tumor associated antigen (TAA). Non-limiting examples of specific tumor associated antigens include, for example, AFP, ALK, BAGE proteins, BIRC5 (survivin), BIRC7, β-catenin, brc-abl, BRCA1, BORIS, CA9, carbonic anhydrase IX, caspase-8, CALR, CCR5, CD19, CD20 (MS4A1), CD22, CD30, CD40, CDK4, CEA, CTLA4, cyclin B1, CYP1B1, EGFR, EGFRvIII, ErbB2/Her2, ErbB3, ErbB4, ETV6-AML, EpCAM, EphA2, Fra-1, FOLR1, proteins GAGE (eg, GAGE-1, -2), GD2, GD3, GloboH, glypican-3, GM3, gp100, Her2, HLA/B-raf HLA/k-ras, HLA/MAGE-A3, hTERT, LMP2, proteins MAGE (eg, MAGE-1, -2, -3, -4, -6 and -12), MART-1, mesothelin, ML-IAP, Muc1, Muc2, Muc3, Muc4, Muc5, Muc16 (CA-125) , MUM1, NA17, NY-BR1, NY-BR62, NY-BR85, NY-ESO1, OX40, p15, p53, PAP, PAX3, PAX5, PCTA-1, PLAC1, PRLR, PRAME, PSMA (FOLH1), RAGE proteins , Ras, RGS5, Rho, SART-1, SART-3, STEAP1, STEAP2, TAG-72, TGF-β, TMPRSS2, Thompson-nouvelle antigen (Tn), TRP-1, TRP-2, tyrosinase and uroplakin-3 .

The inventors contemplate that the invention includes numerous examples of bispecific antibodies having a weak anti-CD3 binding arm made in accordance with the invention.

In some exemplary embodiments, the invention includes bispecific antigen-binding molecules that specifically bind CD3 and PSMA. Such molecules may be referred to herein as, for example, anti-CD3/anti-PSMA or antu-CD3xPSMA or CD3xPSMA bispecific molecules, etc. As used herein, the term PSMA refers to the human PSMA protein unless indicated to refer to a non-human species (eg, mouse PSMA, monkey PSMA, etc.).

The term PSMA refers to the prostate-specific membrane antigen, also known as folate hydrolase 1 (FOLH1) (UniProtKB/Swiss-Prot. No. Q04609; SEQ ID NO: 171). PSMA is an integral, non-shedd membrane glycoprotein that is highly expressed in prostate epithelial cells and is a cell surface marker for prostate cancer.

In other exemplary embodiments, the invention includes bispecific antigen-binding molecules that specifically bind CD3 and EGFRvIII. Such molecules may be referred to herein as, for example, anti-CD3/anti-EGFRvIII or anti-CD3xEGFRvIII or CD3xEGFRvIII bispecific molecules, etc. The term EGFRvIII refers to

- 16 040594 human EGFRvIII protein, unless stated to be of a non-human species (e.g.,

mouse EGFRvIII, monkey EGFRvIII, etc.).

The term EGFRvIII refers to a class III variant of the epidermal growth factor receptor (EGFRvIII, SEQ ID NO: 172), which is the most frequently occurring EGFR variant in glioblastoma (Bigner et al., 1990, Cancer Res. 50:8017-8022; Humphrey et al ., 1990, Proc Natl Acad Sci USA 87:4207-4211 Yamazaki et al. 1990 Jap J Cancer Res 81:773-779 Ekstrand et al 1992 Proc Natl Acad. Sci. USA, 89:4309-4313; Wikstrand et al., 1995, Cancer Res. 55:3140-3148 and Frederick et al., 2000, Cancer Res. 60:1383-1387). EGFRvIII is characterized by a deletion of exons 2-7 of the EGFR gene, resulting in a deletion within the 801 base pair reading frame of the coding region, i. deletion of 6-273 amino acid residues (depending on the number of mature EGFR subtractions, see UniProtKB/Swiss-Prot. No. P00533), as well as the generation of new glycine at the fusion junction (Humphrey et al., 1988, Cancer Res. 48:2231 -2238; Yamazaki et al., 1990, supra). EGFRvIII has been shown to have ligand-independent, weak but constitutively active kinase activity, as well as increased carcinogenicity (Nishikawa et al., 1994, Proc. Natl. Acad. Sci. USA, 91:7727-7731 and Batra et al. , 1995, Cell Growth and Differentiation, 6:1251-1259). In addition to gliomas, EGFRvIII has been found in ductal and intradermal breast carcinomas (Wikstrand et al., 1995, Cancer Res. 55:3140-3148), non-small cell lung carcinomas (Garcia de Palazzo et al., 1993, Cancer Res. 53: 3217-3220), ovarian carcinomas (Moscatello et al., 1995, Cancer Res. 55:5536-5539), prostate cancer (Olapade-Olaopa et al., 2000, British J. Cancer, 82:186-194) and squamous head and neck carcinoma (Tinhofer et al., 2011, Clin. Cancer Res. 17(15):5197-5204).

In still other exemplary embodiments, the invention includes bispecific antigen-binding molecules that specifically bind CD3 and MUC16. Such molecules may be referred to herein as, for example, anti-CD3/anti-MUC16 or anti-CD3xMUC16 or CD3xMUC16 bispecific molecules, etc. The term MUC16 refers to the human MUC16 protein unless indicated to be of a non-human species (eg, mouse MUC16, monkey MUC16, etc.).

Mucin 16 (MUC16; NCBI Reference Sequence: NP_078966.2, SEQ ID NO: 173), otherwise known as cancer antigen 125 (CA-125), is a mucin encoded by the MUC16 gene in humans. The mucin family of proteins is known to protect the body from infection by binding the pathogen to oligosaccharides in the extracellular domain, preventing the formation of the pathogenic form on the cell surface. For many years, MUC16/CA125 overexpression has been used as a prognostic and diagnostic marker for ovarian cancer (Yin and Lloyd, 2001, J. Biol. Chem. 276(29), 27371-27375; O'Brien, T.J., et al., 2001 , Tumor Biol 22(6), 348-366; Leggieri, C. et al., 2014, Eur J. Gynaecol Oncol 35(4), 438-441). MUC16 has been shown to protect tumor cells from the immune system with its highly glycosylated tandem repeat domain that can bind to galectin-1 (an immunosuppressive protein) (Seelenmeyer, C., et al., 2003, J. Cell. Sci. 116 ( Pt 7): 130518 O'Brien, T. J., et al., 2002, Tumor Biol 23(3), 154-169. Natural killer cells and monocytes are unable to attack tumor cells expressing high levels of MUC16. In its normal physiological role, the MUC16-galactin interaction serves as a barrier to bacterial and viral infection, however, MUC16 is believed to be immunoprotective in the context of tumor cells, thereby preventing cytolysis of cancer cells (Felder, M. et al., 2014, Molecular Cancer, 13 :129). Thus, MUC16 is a desirable target for immunotherapeutic bispecific antibody molecules administered to treat ovarian cancer by activating immune effector cells.

In still other exemplary embodiments, the invention includes bispecific antigen-binding molecules that specifically bind CD3 and STEAP2. Such molecules may be referred to herein as, for example, anti-CD3/anti-STEAP2 or anti-CD3xSTEAP2 or CD3xSTEAP2 bispecific molecules, etc. The term STEAP2 refers to the human STEAP2 protein unless stated to be in a non-human species (eg, mouse STEAP2, monkey STEAP2, etc.). Sixth transmembrane epithelial prostate antigen 2 (STEAP2; UniProtKB/Swiss-Prot: Q8NFT2.3) is a 490-amino acid protein encoded by the STEAP2 gene, located in the chromosomal region 7q21 in humans.

The aforementioned bispecific antigen-binding molecules that specifically bind a tumor-associated antigen comprise an anti-CD3 antigen-binding molecule that binds to CD3 with weak or undetectable binding affinity, such as exhibiting a K _D of greater than about 100, 300, or 500 nM, as measured by in vitro binding affinity assay.

As used herein, the term antigen-binding molecule means a protein, polypeptide, or molecular complex containing or consisting of at least one complementarity-determining region (CDR) that alone or in combination with one or more additional CDRs and/or framework regions (FRs) specifically binds to a specific antigen. In some embodiments, the antigen binding molecule is an antibody or

- 17 040594 antibody fragment as these terms are defined in the text of this document.

As used herein, the term bispecific antigen-binding molecule means a protein, polypeptide, or molecular complex comprising at least a first antigen-binding domain and a second antigen-binding domain. Each antigen-binding domain in a bispecific antigen-binding molecule contains at least one CDR, which alone or in combination with one or more additional CDRs and/or FRs specifically binds to a particular antigen. In the context of this invention, the first antigen-binding domain specifically binds a first antigen (eg, CD3), and the second antigen-binding domain specifically binds a second, different antigen (eg, PSMA, MUC16, EGFRvIII, or STEAP2).

In some exemplary embodiments of this invention, the bispecific antigen-binding molecule is a bispecific antibody. Each antigen-binding domain of a bispecific antibody contains a heavy chain variable domain (HCVR) and a light chain variable domain (LCVR). In the context of a bispecific antigen-binding molecule comprising first and second antigen-binding domains (eg, a bispecific antibody), the CDRs of the first antigen-binding domain may be denoted by the prefix A1 and the CDRs of the second antigen-binding domain may be denoted by the prefix A2. Thus, the CDRs of the first antigen-binding domain may be referred to herein as A1-HCDR1, A1-HCDR2 and A1-HCDR3; a CDRs of the second antigen-binding domain may be referred to herein as A2-HCDR1, A2-HCDR2 and A2-HCDR3. The first antigen-binding domain and the second antigen-binding domain can be directly or indirectly linked to each other to form a bispecific antigen-binding molecule of the invention. Alternatively, the first antigen-binding domain and the second antigen-binding domain may be associated with a separate multimerizing domain. Linking one multimerizing domain to another multimerizing domain facilitates communication between the two antigen-binding domains, thereby forming a bispecific antigen-binding molecule. As used herein, the term multimerization domain is any macromolecule, protein, polypeptide, peptide, or amino acid that has the ability to bind to a second multimerization domain of the same or similar structure or constitution. For example, the multimerizing domain may be a polypeptide containing an immunoglobulin C _H 3 domain. A non-limiting example of a multimerizing component is the Fc portion of an immunoglobulin (containing a CH2-CH3 domain), such as an IgG Fc domain selected from the IgG1, IgG2, IgG3, and IgG4 isotypes, as well as any allotype within each isotype group.

The bispecific antigen-binding molecules of this invention typically contain two multimerizing domains, such as two Fc domains, each of which is individually part of a separate antibody heavy chain. The first and second multimerizing domains may be of the same IgG isotype, such as, for example, IgG1/IgG1, IgG2/IgG2, IgG4/IgG4. Alternatively, the first and second multimerizing domains may be of different IgG isotypes, such as, for example, IgG1/IgG2, IgG1/IgG4, IgG2/IgG4, etc.

In some embodiments, the multimerization domain is an Fc fragment or an amino acid sequence of 1 to about 200 amino acids in length containing at least one cysteine residue. In other embodiments, the multimerization domain is a cysteine residue or a short cysteine-containing peptide. Other multimerizing domains include peptides or polypeptides containing or consisting of a leucine zipper, a coiled loop motif, or a coiled coil motif.

Any format or technology of a bispecific antibody can be used to make the bispecific antigen-binding molecules of this invention. For example, an antibody or fragment thereof having a first antigen-binding specificity may be operably linked (e.g., by chemical bonding, genetic synthesis, non-covalent association, or otherwise) to one or more other molecular entities, such as another antibody or antibody fragment having a second antigen-binding specificity to obtain a bispecific antigen-binding molecule. Specific exemplary bispecific formats that may be used in the context of the present invention include, without limitation, for example, scFv-based or diary bispecific formats, IgG-scFv fusion, dual variable domain (DVD)-Ig, Quadroma, ridge-in-gap, common light chain (e.g., common lung-in-trough light chain, etc.), CrossMab, CrossFab, (SEED) bodies, leucine zipper, Duobody, IgG1/IgG2, Dual Action Fab (DAF)-IgG, and Mab ² bispecific formats (see, for example, Klein et al. 2012, mAbs, 4:6, 1-11, and references cited therein for an overview of the above formats).

In the context of the bispecific antigen-binding molecules of the invention, multimerizing domains, eg Fc domains, may contain one or more amino acid changes (eg insertions, deletions or substitutions) compared to naturally occurring versions of the wild-type Fc domain. For example, the invention includes bispecific antigen-binding molecules containing one or more modifications in the Fc domain, which results in a modified Fc domain having a modified binding (eg, increased or decreased) between Fc and

- 18 040594

FcRn. In one embodiment, the bispecific antigen binding molecule comprises a modification in the CH2 or CH3 region, wherein the modification increases the affinity of the Fc domain for FcRn in an acidic environment (eg, in the endosome where the pH is about 5.5 to about 6.0). Non-limiting examples of such Fc modifications include, for example, a modification at position 250 (eg, E or Q); 250 and 428 (for example, L or F); 252 (eg L/Y/F/W or T), 254 (eg S or T) and 256 (eg S/R/Q/E/D or T); or a modification at position 428 and/or 433 (eg L/R/S/P/Q or K) and/or 434 (eg H/F or Y); or a modification at position 250 and/or 428; or a modification at position 307 or 308 (eg, 308F, V308F) and 434. In one embodiment, the modification includes modifications 428L (eg, M428L) and 434S (eg, N434S); modifications 428L, 259I (for example, V259I) and 308F (for example, V308F); modifications 433K (for example, H433K) and 434 (for example, 434Y); modifications 252, 254 and 256 (for example, 252Y, 254T and 256E); modifications 250Q and 428L (eg T250Q and M428L) and modifications 307 and/or 308 (eg 308F or 308P).

The invention also includes bispecific antigen-binding molecules comprising a first CH3 domain and a second C _H 3 Ig domain, wherein the first and second C _H 3 Ig domains differ from each other by at least one amino acid, and wherein at least one amino acid difference reduces binding a bispecific antibody with protein A versus a bispecific antibody lacking the amino acid difference. In one embodiment, the first C _H 3 Ig domain binds the A protein and the second Ig C _H 3 domain contains a mutation that reduces or abolishes the binding to the A protein, such as an H95R modification (via IMGT exon numbering, H435R by EC numbering). The second CH3 may additionally contain the modification Y96F (by IMGT; Y436F by EU). Further modifications that can be found in the second CH3 include D16E, L18M, N44S, K52N, V57M and V82I (by IMGT; D356E, L358M, N384S, K392N, V397M and V422I by EU) in the case of IgG1 antibodies; N44S, K52N and V82I (by IMGT; N384S, K392N and V422I by EU) for IgG2 and Q15R antibodies, N44S, K52N, V57M, R69K, E79Q and V82I (by IMGT; Q355R, N384S, K392N, V397M, R409K, E419Q and V422I by EU) in the case of IgG4 antibodies.

In some embodiments, the Fc domain may be chimeric, combining Fc sequences derived from more than one immunoglobulin isotype. For example, a chimeric Fc domain may comprise part or all of the _CH 2 sequence derived from the _CH 2 region of human IgG1, human IgG2, or human IgG4 and part or all of the CH3 sequence derived from human IgG1, human IgG2, or human IgG4. The chimeric Fc domain may also contain a chimeric hinge region. For example, a chimeric hinge may comprise an upper hinge sequence derived from a human IgG1, human IgG2, or human IgG4 hinge region combined with a lower hinge sequence derived from a human IgG1, human IgG2, or human IgG4 hinge region. A specific example of a chimeric Fc domain that can be included in any of the antigen binding molecules set forth herein includes the N- to C-terminus: [ _CH 1 IgG4] [upper IgG4 hinge] - [lower IgG2 hinge] - [C _H 2 IgG4] [ _CH 3 IgG4]. Another example of a chimeric Fc domain that can be _included in any of the antigen binding molecules set forth herein includes N- to C-terminus: _H 2 IgG4] - [ _CH 3 IgG1]. These and other examples of chimeric Fc domains that can be included in any of the antigen binding molecules of this invention are described in PCT International Publication No. WO 2014/121087 A1, published August 7, 2014, which is incorporated herein by reference throughout its completeness. Fc chimeric domains sharing these common structural mechanisms and variants thereof can alter Fc receptor binding, which in turn affects Fc effector function.

In some embodiments, the invention relates to an antibody heavy chain, wherein the heavy chain constant region (CH) contains an amino acid sequence of at least 95%, at least 96%, at least 97%, at least 98% at least 99% identical to any of SEQ ID NO: 182, SEQ ID NO: 183, SEQ ID NO: 184, SEQ ID NO: 185, SEQ ID NO: 186, SEQ ID NO: 187, SEQ ID NO: 188, SEQ ID NO: 189, SEQ ID NO: 190, or SEQ ID NO: 191. In some embodiments, the heavy chain constant region ( _CH ) comprises an amino acid sequence selected from the group consisting of SEQ ID NO: 182, SEQ ID NO: 183, SEQ ID NO: 184, SEQ ID NO: 185, SEQ ID NO: 186, SEQ ID NO: 187, SEQ ID NO: 188, SEQ ID NO: 189, SEQ ID NO: 190 and SEQ ID NO: 191.

In other embodiments, the invention relates to an antibody heavy chain wherein the Fc domain contains an amino acid sequence of at least 95%, at least 96%, at least 97%, at least 98%, at least 99% identical to any of SEQ ID NO: 192, SEQ ID NO: 193, SEQ ID NO: 194, SEQ ID NO: 195, SEQ ID NO: 196, SEQ ID NO: 197, SEQ ID NO: 198, SEQ ID NO : 199, SEQ ID NO: 200, or SEQ ID NO: 201. In some embodiments, the Fc domain comprises an amino acid sequence selected from the group consisting of SEQ ID NO: 192, SEQ ID NO: 193, SEQ ID NO: 194, SEQ ID NO: 195, SEQ ID NO: 196, SEQ ID NO: 197, SEQ ID NO: 198, SEQ ID NO: 199, SEQ ID NO: 200 and SEQ ID NO: 201.

- 19 040594

Other variants of Fc.

In accordance with some embodiments of the present invention, anti-CD3 antibodies and anti-CD3/anti-TAA bispecific antigen-binding molecules are provided, comprising an Fc domain containing one or more mutations that increase or decrease antibody binding to an FcRn receptor, e.g., at acidic pH. compared to neutral pH. For example, the invention includes antibodies containing a mutation in the C _H 2 or _CH 3 region of the Fc domain, wherein the mutation(s) increases the affinity of the Fc domain for FcRn in an acidic environment (e.g., in the endosome where the pH is from about 5.5 to about 6.0). Such mutations can lead to an increase in the serum half-life of the antibody when administered to an animal. Non-limiting examples of such Fc modifications include, for example, a modification at position 250 (eg, E or Q); 250 and 428 (for example, L or F); 252 (eg L/Y/F/W or T), 254 (eg S or T) and 256 (eg S/R/Q/E/D or T); or a modification at position 428 and/or 433 (eg H/L/R/S/P/Q or K) and/or 434 (eg H/F or Y); or a modification at position 250 and/or 428; or a modification at position 307 or 308 (eg, 308F, V308F) and 434. In one embodiment, the modification includes modifications 428L (eg, M428L) and 434S (eg, N434S); modifications 428L, 259I (for example, V259I) and 308F (for example, V308F); modifications 433K (for example, H433K) and 434 (for example, 434Y); modifications 252, 254 and 256 (for example, 252Y, 254T and 256E); modifications 250Q and 428L (for example, T250Q and M428L) and modifications 307 and/or 308 (for example, 308F or 308P).

For example, this invention includes anti-CD3 antibodies and anti-CD3/anti-TAA bispecific antigen-binding molecules containing an Fc domain containing one or more pairs or groups of mutations selected from the group consisting of 250Q and 248L (for example, T250Q and M248L) ; 252Y, 254T and 256E (eg M252Y, S254T and T256E); 428L and 434S (eg M428L and N434S) and 433K and 434F (eg H433K and N434F). All possible combinations of the above Fc domain mutations and other mutations within the variable domains of an antibody described herein are contemplated within the scope of this invention.

Biological characteristics of antibodies and bispecific antigen-binding molecules.

The invention includes bispecific antigen-binding molecules (eg, bispecific antibodies) that are capable of simultaneously binding to human CD3 and human TAA. In some embodiments, the bispecific antigen-binding molecules of the invention specifically interact with cells that express CD3 and/or TAA, such as PSMA, EGFRvIII, or MUC16. A binding arm that interacts with cells that express CD3 may have weak or undetectable binding as measured in a suitable in vitro binding assay. The extent to which a bispecific antigen binding molecule binds cells that express CD3 and/or TAA can be assessed by flow cytometry (FACS) as shown in Example 4 herein.

For example, the invention includes antibodies, antigen-binding fragments, and bispecific antibodies thereof that specifically bind a human T cell line that expresses CD3 but not TAA (e.g., Jurkat), primate T cells (e.g., cynomolgus monkey peripheral blood mononuclear cells [ PBMC]) and/or TAA-expressing cells. This invention includes bispecific antigen-binding molecules that bind any of the aforementioned T cells and T cell lines with an EC ₅₀ value from about 1.8 x 10'8 (18 nM) to about 2.1 x 10' ⁷ (210 nM) or more (t ie lower affinity) and includes bispecific antibodies for which no EC ₅₀ is detected as determined using a FACS binding assay as described in Example 4 or a substantially similar assay. In some embodiments, the antibodies, antigen binding fragments, and bispecific antibodies of the invention bind CD3 with an EC ₅₀ of greater than about 30 nM, greater than about 40 nM, greater than about 45 nM, greater than about 50 nM, greater than about 55 nM , greater than about 60 nM, greater than about 65 nM, greater than about 70 nM, greater than about 75 nM, at least 80 nM, greater than about 90 nM, greater than about 100 nM, greater than about 110 nM, according to at least 120 nM, greater than about 130 nM, greater than about 140 nM, greater than about 150 nM, at least 160 nM, greater than about 170 nM, greater than about 180 nM, greater than about 190 nM, greater than about 200 nM, greater than about 250 nM, greater than about 300 nM, greater than about 500 nM, greater than about 1 μM, greater than about 2 μM, or greater than about 3 μM, or an undetectable binding affinity as measured by binding FACS, for example, using the analysis format as defined in example 4 herein, or, in essence nature of a similar analysis.

The invention also includes antibodies, antigen-binding fragments, and bispecific antibodies thereof that bind to TAA-expressing cells and cell lines, such as PSMA-, EGFRvIII-, STEAP2-, and MiC16-expressing cell lines, with an EC ₅₀ value of less than about 100 nM. or even less than the concentration required for binding (i.e., stronger affinity), such as less than 5.6 nM (5.6x10'9), as determined using a FACS binding assay as described in Example 4, or, by essentially a similar cellular assay. Dan

The invention includes bispecific antigen binding molecules that bind any of the aforementioned tumor cell lines with an EC ₅₀ value of less than about 50 nM, less than about 45 nM, less than about 40 nM, less than about 35 nM, less than about 30 nM, less than about 25 nM, less than about 20 nM, less than about 15 nM, less than about 10 nM, less than about 6 nM, less than about 5 nM, or less than about 1 nM, e.g. using the above assay .

The present invention includes antibodies, antigen-binding fragments, and bispecific antibodies that bind human CD3 with low, weak, or even no detectable affinity. In some embodiments, the invention includes antibodies and antigen-binding antibody fragments that bind human CD3 (e.g., at 37°C) with a KD of greater than about 11 nM, includes antibodies that bind CD3 with a KD of greater than about 100 or 500 nM , and also includes antibodies having no detectable binding affinity as measured by surface plasmon resonance, for example, using the assay format as defined in Example 5 herein. In some embodiments, the antibodies or antigen-binding fragments of this invention bind CD3 with a KD of greater than about 15 nM, greater than about 20 nM, greater than about 25 nM, greater than about 30 nM, greater than about 35 nM, greater than about 40 nM, greater than about 45 nM, greater than about 50 nM, greater than about 55 nM, at least 60 nM, greater than about 65 nM, greater than about 70 nM, greater than about 75 nM, at least 80 nM , greater than about 90 nM, greater than about 100 nM, greater than about 110 nM, at least 120 nM, greater than about 130 nM, greater than about 140 nM, greater than about 150 nM, greater than about 160 nM, greater than greater than about 170 nM, greater than about 180 nM, greater than about 190 nM, greater than about 200 nM, greater than about 250 nM, greater than about 300 nM, greater than about 1 μM, greater than about 2 μM, or greater than about 3 μM, or undetectable binding affinity measured from a surface plasmon resonance assay, e.g. using forms Assay data as defined in Example 5 herein (eg, mAb capture or antigen capture format), or a substantially similar assay.

The invention includes antibodies, antigen-binding fragments, and bispecific antibodies that bind monkey (cynomolgus) CD3 with low, weak, or even no detectable affinity. According to some embodiments, the invention includes antibodies and antigen-binding antibody fragments, and bispecific antibodies thereof that bind human CD3 (e.g., at 37°C) with a KD of greater than about 10 nM, includes antibodies that bind CD3 to a KD of greater than about 100 or 500 nM, and also includes antibodies having no detectable binding affinity as measured by surface plasmon resonance, for example, using the assay format as defined in Example 5 herein. In some embodiments, the antibodies or antigen-binding fragments of this invention bind CD3 with a KD of greater than about 15 nM, greater than about 20 nM, greater than about 25 nM, greater than about 30 nM, greater than about 35 nM, greater than about 40 nM, greater than about 45 nM, greater than about 50 nM, greater than about 55 nM, at least 60 nM, greater than about 65 nM, greater than about 70 nM, greater than about 75 nM, at least 80 nM , greater than about 90 nM, greater than about 100 nM, greater than about 110 nM, at least 120 nM, greater than about 130 nM, greater than about 140 nM, greater than about 150 nM, greater than about 160 nM, greater than greater than about 170 nM, greater than about 180 nM, greater than about 190 nM, greater than about 200 nM, greater than about 250 nM, greater than about 300 nM, greater than about 1 μM, greater than about 2 μM, or greater than about 3 μM, or an undetectable binding affinity measured using a surface plasmon resonance assay, e.g. receiving an assay format as defined in Example 5 herein (eg, mAb capture or antigen capture format), or a substantially similar assay.

The present invention includes antibodies, antigen-binding fragments, and bispecific antibodies that bind human CD3 and induce T cell activation. For example, this invention includes anti-CD3 antibodies that induce human T cell activation with an EC ₅₀ value of less than about 113 pM as measured by an in vitro T cell activation assay, e.g. using the assay format as defined in Example 6 in herein [eg, assessment of percent activated (CD69+) cells from total T cells (CD2+) in the presence of anti-CD3 antibodies], or a substantially similar assay that evaluates a T cell in its activated state. In some embodiments, the antibodies or antigen-binding fragments of the invention induce human T cell activation (e.g., percentage of activated (CD69+) T cells) with an EC50 value of less than about 100 pM, less than about 50 pM, less than about 20 less than about 19 pM, less than about 18 pM, less than about 17 pM, less than about 16 pM, less than about 15 pM, less than about 14 pM, less than about 13 pM, less than about 12 pM, less than about 11 pM, less than about 10 pM, less than about 9 pM, less than about 8 pM, less than about 7 pM, less than about 6 pM, less than about 5 pM, less than about 4 pM,

- 21 040594 less than about 3 pM, less than about 2 pM, or less than about 1 pM, as measured by an in vitro T cell activation assay, for example using the assay format as defined in Example 6 herein, or, by essentially a similar analysis. Ahtu-CD3 antibodies that have weak or undetectable binding to CD3 have the ability to induce T cell activation with high efficiency (i.e. pM range) despite having a weak or undetectable binding affinity to CD3, such as shown in example 6 in this document.

The invention also includes antibodies, antigen-binding fragments, and bispecific antibodies that bind human CD3 and induce T-cell mediated lysis of tumor antigen-expressing cells. For example, the invention includes anti-CD3 antibodies that induce T-cell mediated lysis of tumor cells with an EC ₅₀ of less than about 1.3 nM as measured in a T-cell mediated tumor cell lysis assay, e.g. , as defined in Example 6 herein (e.g., assessment of the extent of tumor antigen-expressing cells, such as PSMA expression, EGFRvIII expression, or MUC16 expression, human PBMC cell count in the presence of anti-CD3 antibodies), or a substantially similar assay. In some embodiments, the antibodies or antigen-binding fragments of the invention induce T cell mediated tumor cell lysis (e.g., PBMC mediated OVCAR3 cell lysis) with an EC ₅₀ value of less than about 1 nM, less than about 400 pM, less than about 250 pM, less than about 100 pM, less than about 50 pM, less than about 40 pM, less than about 30 pM, less than about 20 pM, less than about 10 pM, less than about 9 pM, less than about 8 pM , less than about 7 pM, less than about 6 pM, less than about 5 pM, less than about 4 pM, less than about 3 pM, less than about 2 pM, or less than about 1 pM, as measured in an in vitro lysis assay tumor-mediated T-cells, for example, using the assay format as defined in Example 6 herein, or a substantially similar assay.

The invention also includes antibodies, antigen-binding fragments, and bispecific antibodies that bind CD3 with a dissociative half-life (t _1/2 ) of less than about 10 minutes, as measured by surface plasmon resonance at 25 or 37°C, for example, using the assay format , as defined in example 5 of the present invention, or essentially a similar analysis. In some embodiments, the antibodies or antigen-binding fragments of the invention bind CD3 to t _1/2 in less than about 9 minutes, less than about 8 minutes, less than about 7 minutes, less than about 6 minutes, less than about 5 minutes, less than about 4 minutes, less than about 3 minutes, less than about 2 minutes, less than about 1.9 minutes, or less than about 1.8 minutes, or exhibit very little or no detectable binding as measured by surface plasmon resonance at 25° or 37° C, for example using the assay format as defined in Example 5 herein (eg, mAb or antigen capture format), or a substantially similar assay.

The anti-CD3/anti-TAA bispecific antigen binding molecules of this invention may further exhibit one or more characteristics selected from the group consisting of:

(a) inducing PBMC proliferation in vitro;

(b) activating T cells by inducing IFN-gamma release and regulating CD25 inflammation in human whole blood, and (c) inducing T cell mediated cytotoxicity in anti-TAA resistant cell lines.

This invention includes anti-CD3/anti-TAA bispecific antigen-binding molecules that are capable of destroying tumor antigen-expressing cells in a subject (see, for example, example 7). For example, according to some embodiments, anti-CD3/anti-PSMA, anti-CD3/anti-MUC16, or anti-CD3/anti-STEAP2 bispecific antigen-binding molecules are provided, wherein a single administration of 1 or 10 or 100 μg of the bispecific antigen-binding molecule to a subject ( for example, at a dose of about 0.1 mg/kg, about 0.08 mg/kg, about 0.06 mg/kg, about 0.04 mg/kg, about 0.04 mg/kg, about 0.02 mg/kg kg, about 0.01 mg/kg or less) causes a decrease in the number of tumor antigen-expressing cells in the subject (eg, tumor growth in the subject is suppressed or inhibited) below detectable levels. In some embodiments, a single administration of an anti-CD3/anti-PSMA bispecific antigen-binding molecule at a dose of about 0.4 mg/kg results in a reduction in tumor growth in a subject below the detection level by about day 7, about day 6, about day 5 , about day 4, about day 3, about day 2, or about day 1 after administration of the bispecific antigen-binding molecule to a subject. In some embodiments, a single administration of an anti-CD3/anti-PSMA bispecific antigen-binding molecule of the invention at a dose of at least about 0.01 mg/kg results in the number of PSMA-expressing tumor cells remaining below detectable levels up to at least 7 days, 8 days, 9 days, 10 days, 11 days, 12 days, 13 days, 14 days, 15 days, 16 days, 17 days or more after administration. The expression used in this document is neither

- 22 040594 same detectable levels means that no tumor cells can be detected directly or indirectly, when growing subcutaneously in a subject, using standard instruments and measurement methods, for example, as described in example 7. In some embodiments, a single administration of antiCD3/ anti-MUC16 of the bispecific antigen-binding molecule at a dose of about 10 μg results in suppression of tumor growth in the subject by about 6 days and maintains tumor suppression for at least 26 days after administration of the bispecific antigen-binding molecule to the subject. In subjects receiving a single administration of an anti-CD3/anti-MUC16 bispecific antigen-binding molecule at a dose of about 10 μg at least 7 days after tumor implantation, the bispecific antigen-binding molecule is effective in inhibiting the growth of established tumors in the subject at about 26 days after implantation. subject's tumors. In some embodiments, a single administration of an anti-CD3/anti-MUC16 bispecific antigen-binding molecule of the invention at a dose of at least about 0.1 mg/kg inhibits the growth of MUC16-expressing tumor cells for at least about 7 days, 8 days , 9 days, 10 days, 11 days, 12 days, 13 days, 14 days, 15 days, 16 days, 17 days, 18 days, 19 days, 20 days or more after administration of the bispecific molecule. See, for example, example 8.

In some embodiments, a single administration of the anti-CD3/anti-STEAP2 bispecific antigen binding molecule at a dose of about 0.1 or 0.01 mg/kg maintains tumor growth suppression for at least 46 days after administration of the bispecific antigen binding molecule and tumor to the subject. In some embodiments, a single administration of an anti-CD3/anti-STEAP2 bispecific antigen-binding molecule of the invention at a dose of at least about 0.1 mg/kg, about 0.08 mg/kg, about 0.06 mg/kg , about 0.05 mg/kg, about 0.04 mg/kg, about 0.03 mg/kg, about 0.02 mg/kg, about 0.01 mg/kg or less inhibits the growth of STEAP2-expressing tumor cells for up to at least about 20 days, 30 days, 35 days, 40 days, 45 days or more after administration of the bispecific molecule. See, for example, example 10.

In other embodiments, anti-CD3/anti-TAA bispecific antigen-binding molecules having a target CD3-binding arm having low binding affinity for effector cells exhibit a reduced rate of drug elimination compared to bispecific antibodies containing the same anti-TAA. -binding arm and strong CD3 binding introduced in an in vivo pharmacokinetic study. The results demonstrate that bispecific molecules containing weaker arm binding to the CD3 targeting arm may exhibit favorable drug exposure levels (AUC _last ) and drug elimination profiles (antibody clearance). See, for example, example 9.

The present invention provides anti-CD3/anti-PSMA, anti-CD3/anti-MUC16, and anti-CD3/anti-STEAP2 bispecific antigen-binding molecules (i.e., anti-CD3/anti-TAA bispecific antigen-binding molecules) that exhibit one or more characteristics selected from the group consisting of:

(a) inhibition of tumor growth in immunocompromised mice bearing human prostate cancer xenografts;

(b) inhibition of tumor growth in immunocompromised mice bearing human prostate cancer xenografts;

(c) inhibition of tumor growth of established tumors in immunocompromised mice bearing human prostate cancer xenografts; and (d) reducing tumor growth in established tumors in immunocompromised mice bearing human prostate cancer xenografts (see, for example, Examples 7, 8 and 10).

The present invention also provides anti-CD3/anti-PSMA, anti-CD3/anti-MUC16, and anti-CD3/anti-STEAP2 bispecific antibodies (i.e., anti-CD3/anti-TAA bispecific antibodies) containing i) the first a heavy chain directed to an effector T cell (ie, CD3); and ii) a second heavy chain directed to the target tumor cell, wherein the bispecific antibodies exhibit weak or undetectable binding to effector cells and exhibit suppression of tumor growth and reduced antibody clearance (i.e. elimination) from the body compared to bispecific antibodies that show strong binding to effector cells.

Epitope mapping and related technologies.

The epitope on CD3 to which the antigen-binding molecules of the present invention bind may consist of a single contiguous sequence of 3 or more (e.g., 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20 or more) amino acids of the CD3 protein. Alternatively, an epitope may be composed of a plurality of non-contiguous amino acids (or amino acid sequences) of CD3. Antibodies of the invention may interact with amino acids contained on a single CD3 chain (eg, CD3 epsilon, CD3 delta, or CD3 gamma) or may interact with amino acids on two or more different CD3 chains. As used herein, the term epitope refers to an antigenic determinant that interacts with a specific antigen binding site in the variable region of an antibody molecule known as a paratope. One antigen can have more than one

- 23 040594 epitopes. Thus, different antibodies may bind to different regions on an antigen and may have different biological effects. Epitopes can be either conformational or linear. A conformational epitope is produced by spatially aligned amino acids from different segments of the linear polypeptide chain. A linear epitope is one obtained by adjacent amino acid residues in a polypeptide chain. In certain circumstances, an epitope may contain fragments of saccharides, phosphoryl groups, or sulfonyl groups on the antigen.

Various techniques known to those skilled in the art can be used to determine if the antigen-binding domain of an antibody interacts with one or more amino acids within a polypeptide or protein. Illustrative techniques include, for example, standard epitope cross-competition assays such as those described in Antibodies, Harlow and Lane (Cold Spring Harbor Press, Cold Spring Harb., NY), alanine scan mutation analysis, peptide blot analysis (Reineke, 2004, Methods Mol Biol 248:443-463) and peptide digestion analysis. In addition, methods such as epitope removal, epitope extraction, and chemical modification of antigens can be used (Tomer, 2000, Protein Science, 9:487-496). Another method that can be used to identify amino acids within a polypeptide with which an antigen-binding domain of an antibody interacts is hydrogen-deuterium exchange, detected by mass spectrometry. In general, the hydrogen-deuterium exchange method involves labeling the protein of interest with deuterium and then binding the antibody to the deuterium labeled protein. The protein/antibody complex is then transferred to water so that hydrogen can be exchanged with deuterium in all residues, except for residues protected by the antibody (which remain labeled with deuterium). After dissociation of the antibody, the target protein is subjected to protease digestion and mass spectrometry analysis, thereby revealing deuterium-labeled residues that correspond to the specific amino acids with which the antibody interacts. See, for example, Ehring (1999), Analytical Biochemistry, 267(2):252-259; Engen and Smith (2001), Anal. Chem. 73:256A-265A. X-ray crystallography of the antigen/antibody complex can also be used for epitope mapping purposes.

The invention further includes anti-PSMA antibodies that bind to the same epitope as any specific exemplary antibody described herein (e.g., antibodies containing any of the amino acid sequences as listed in Table 6 herein) . Similarly, the invention also includes anti-PSMA antibodies that compete for binding to PSMA with any of the specific exemplary antibodies described herein (e.g., antibodies containing any of the amino acid sequences as listed in Table 6 herein) . The anti-PSMA antibodies disclosed in US Application No. 15/223434 are incorporated herein by reference.

The invention also includes bispecific antigen-binding molecules comprising a first antigen-binding domain that specifically binds human CD3 and/or cynomolgus CD3 with low or detectable binding affinity, and a second antigen-binding domain that specifically binds human tumor-associated antigen (TAA), the first antigen-binding domain binds to the same epitope on CD3 as any of the specific exemplary CD3-specific antigen-binding domains described herein.

In addition, the invention also includes bispecific antigen-binding molecules comprising a first antigen-binding domain that specifically binds human CD3 and/or cynomolgus monkey CD3 with low or detectable binding affinity, and a second antigen-binding domain that specifically binds human tumor associated antigen (TAA), wherein the first antigen-binding domain competes for binding to CD3 with any of the specific exemplary CD3-specific antigen-binding domains described herein.

Whether a particular antigen-binding molecule (eg, antibody) or its antigen-binding domain binds to the same epitope or competes for binding with a reference antigen-binding molecule of the present invention can be readily determined using conventional methods known in the art. For example, to determine if a test antibody binds to the same epitope on CD3 (or TAA) as a reference bispecific antigen-binding molecule of this invention, the reference bispecific molecule first binds to the CD3 protein (or TAA protein). The ability of the test antibody to bind to the CD3 molecule is then assessed. If the test antibody can bind to CD3 (or TAA) after saturation binding to the reference bispecific antigen-binding molecule, it can be concluded that the test antibody binds to a different CD3 (or TAA) epitope than the control bispecific antigen-binding molecule. On the other hand, if the test antibody cannot bind to the CD3 (or TAA) molecule after saturation binding to the reference bispecific antigen-binding molecule, then the test antibody can bind to the same CD3 (or TAA) epitope as the epitope bound by the reference bispecific antigen-binding molecule. according to the invention. Then there may be

- 24 040594 additional routine experiments (e.g. peptide mutation and binding assay) were performed to confirm whether the observed lack of binding of the test antibody is in fact due to binding to the same epitope as the reference bispecific antigen-binding molecule, or steric blocking (or another phenomenon) is responsible for the lack of observed binding. If the control antibody is one that does not have measurable binding, as shown herein by way of example, then the reference antibody can be mutated back to the germline sequence to determine binding to CD3 in order to compare epitope interactions or compare its binding properties to test antibody as described herein. Experiments of this type can be performed using ELISA, RIA, Biacore, flow cytometry, or any other quantitative or qualitative antibody binding assay available in the art. In accordance with some embodiments of the present invention, two antigen-binding proteins bind to the same (or overlapping) epitope if, for example, a 1-, 5-, 10-, 20-, or 100-fold excess of one antigen-binding protein inhibits binding of the other at least by 50%, but preferably by 75%, 90% or even 99%, as measured in a competitive binding assay (see, for example, Junghans et al., Cancer Res. 1990, 50:1495-1502). Alternatively, two antigen-binding proteins are considered to bind to the same epitope if substantially all amino acid mutations in the antigen that reduce or eliminate the binding of one antigen-binding protein reduce or eliminate the binding of the other. Two antigen-binding proteins are considered overlapping epitopes if only a subset of amino acid mutations that reduce or eliminate the binding of one antigen-binding protein reduce or eliminate the binding of another.

To determine if an antibody or its antigen-binding domain competes for binding to a reference antigen-binding molecule, the binding methodology described above is performed in two directions: antibodies with a CD3 (or TAA) molecule. In the second orientation, the test antibody is allowed to bind to the CD3 (or TAA) molecule under saturation conditions, followed by evaluation of the binding of the reference antigen-binding molecule to the CD3 (or TAA) molecule. If in both orientations only the first (saturating) antigen-binding molecule is able to bind to the CD3 (or TAA) molecule, then it is concluded that the test antibody and the reference antigen-binding molecule compete for binding to CD3 (or TAA). As one of skill in the art will appreciate, an antibody that competes for binding to a reference antigen-binding molecule may not necessarily bind to the same epitope as a control antibody, but may sterically block binding of the reference antibody by binding to an overlapping or contiguous epitope. If the control antibody is one that does not have measurable binding, as shown herein by way of example, then the reference antibody can be mutated back to the germline sequence to determine binding to CD3 to compare epitope interactions or compare its binding properties, or blocking interaction with the test antibody, as described herein.

Obtaining antigen-binding domains and designing bispecific molecules.

Antigen-binding domains specific for particular antigens can be generated by any antibody generation technique known in the art. Once obtained, two different antigen-binding domains specific for two different antigens (eg, CD3 and TAA) can be appropriately positioned relative to each other to obtain a bispecific antigen-binding molecule of this invention using conventional methods. (A discussion of exemplary bispecific antibody formats that can be used to construct the bispecific antigen-binding molecules of this invention is provided elsewhere herein). In some embodiments, one or more individual components (eg, heavy and light chains) of the multispecific antigen-binding molecules of the invention are derived from chimeric, humanized, or fully human antibodies. Methods for making such antibodies are well known in the art. For example, one or more heavy and/or light chains of the bispecific antigen-binding molecules of this invention can be prepared using VELOCIMMUNE™ technology. Using VELOCIMMUNE™ technology (or any other human antibody generation technology), high affinity chimeric antibodies to a specific antigen (eg, CD3 or TAA) are initially isolated with a human variable region and a mouse constant region. Antibodies are characterized and selected for desirable characteristics, including affinity, selectivity, epitope, and so on. The mouse constant regions are replaced with the desired human constant region to generate fully human heavy and/or light chains that can be incorporated into the bispecific antigen binding molecules of the invention.

- 25 040594

Genetically modified animals can be used to create bispecific human antigen-binding molecules. For example, a genetically modified mouse that is unable to rearrange and express an endogenous mouse immunoglobulin light chain variable sequence can be used, wherein the mouse expresses only one or two human light chain variable domains encoded by human immunoglobulin sequences operably linked to kappa(k) constant genes. mice at the endogenous kappa(k) mouse locus. Such genetically modified mice can be used to generate fully bispecific human antigen-binding molecules containing two different heavy chains that are associated with an identical light chain that contains a variable domain derived from one of two different human light chain variable gene segments. (See, for example, US 2011/0195454 for a detailed discussion of such modified mice and their use in the production of bispecific antigen-binding molecules). The antibodies of the invention may contain immunoglobulin heavy chains linked to a common light chain. A conventional light chain can be derived from a cognate light chain, an anti-TAA heavy chain, or derived from a known light chain derived variable region or a public domain derived from a light chain that exhibits heterogeneity or the ability to pair with a wide range of non-cognate heavy chains, i. universal or conventional light chain. Antibodies of the invention may contain immunoglobulin heavy chains linked to a single rearranged light chain. In some embodiments, the light chain is a variable domain derived from a human Vk1-39 gene segment or a human Vk3-20 gene segment. In other embodiments, the light chain comprises a variable domain derived from a human Vk1-39 gene segment rearranged with a human Jk5 segment or a human Jk1 gene segment, or a Vk3-20 gene segment rearranged with a human Jk1 gene segment, or a Vk1-39 rearranged using a segment of the human Jk1 gene.

Bioequivalents.

This invention encompasses antigen-binding molecules having amino acid sequences that differ from those described in the molecules disclosed herein, but which retain the ability to bind or interact with CD3 and/or TAA. Such molecular variants may contain one or more insertions, deletions, or substitutions of amino acids compared to the original sequence, but exhibit biological activity that is substantially equivalent to that of the described bispecific antigen-binding molecules.

This invention includes antigen-binding molecules that are bioequivalent to any of the exemplary antigen-binding molecules set forth herein. Two antigen-binding proteins or antibodies are considered bioequivalent if they are pharmaceutical equivalents or pharmaceutical alternatives whose rate and extent of absorption does not show a significant difference when administered at the same molar dose under similar experimental conditions, as single doses or multiple doses. Certain antigen-binding proteins will be considered equivalents or pharmaceutical alternatives if they are equivalent in their rate of absorption but not their rate of uptake, and yet may be considered bioequivalent because such differences in rate of absorption are intentional and are reported as not essential to achieving effective efficacy. concentrations of the drug in the body, such as for chronic use, and are considered to be physically insignificant for the particular investigational medicinal product.

In one embodiment, two antigen-binding proteins are bioequivalent if there are no clinically significant differences in their safety, purity, and potency.

In one embodiment, two antigen-binding proteins are bioequivalent if the patient can switch one or more times between a reference product and a biological product without an expected increase in the risk of adverse effects, including a clinically significant change in immunogenicity or reduced efficacy compared to ongoing therapy without such a switch.

In one embodiment, two antigen-binding proteins are bioequivalent if they both act on a common mechanism or mechanisms of action for the condition or conditions of use, to the extent that such mechanisms are known.

Bioequivalence can be demonstrated by in vivo and in vitro methods. Measures of bioequivalence include, for example, (a) an in vivo test in humans or other mammals, in which the concentration of an antibody or its metabolites is measured in blood, plasma, serum, or other body fluid over time; (b) an in vitro test that has been correlated with the data and reliably predicts human in vivo bioavailability data; (c) an in vivo test in humans or other mammals, in which the corresponding acute pharmacological effect of the antibody (or its target) is measured as a function of time; and (d) in a well-controlled clinical trial that establishes the safety, efficacy, or bioavailability or bioequivalence of antigen binding

- 26 040594 protein.

Bioequivalent variants of the exemplary bispecific antigen-binding molecules set forth herein may be constructed, for example, by making various residue or sequence substitutions, or deleting terminal or internal residues or sequences not necessary for biological activity. For example, cysteine residues that are not essential for biological activity may be removed or replaced with other amino acids to prevent the formation of unnecessary or incorrect intramolecular disulfide bridges upon renaturation. In other contexts, bioequivalent antigen-binding proteins may include variants of the exemplary bispecific antigen-binding molecules described herein, including amino acid changes that alter the glycosylation characteristics of the molecules, such as mutations that eliminate or remove glycosylation.

Species selectivity and species cross-reactivity.

In accordance with some embodiments of the invention, antigen-binding molecules are provided that exhibit little or no interaction with human CD3 and little or no interaction with CD3 of other species, such as cynomolgus monkey CD3. Also provided are antigen-binding molecules that bind to human TAA but not TAA from other species. The invention also includes antigen-binding molecules that bind to human CD3 and CD3 from one or more non-human species; and/or antigen-binding molecules that bind to human TAA and to TAA from one or more non-human species.

In accordance with some exemplary embodiments of the invention, antigen-binding molecules are provided that bind weakly to human CD3 and/or human TAA and may or may not bind, as the case may be, to one or more mouse, rat, guinea pig CD3 and/or TAA. , hamster, gerbil, pig, cat, dog, rabbit, goat, sheep, cow, horse, camel, cynomolgus monkey, marmoset, rhesus monkey, or chimpanzee. For example, in some illustrative embodiments, the present invention provides bispecific antigen-binding molecules comprising a first antigen-binding domain that weakly binds human CD3 and cynomolgus monkey CD3 and a second antigen-binding domain that specifically binds human PSMA, MUC16, EGFRvIII, or STEAP2.

Immunoconjugates.

The present invention encompasses antigen-binding molecules conjugated to a therapeutic moiety (immunoconjugate), such as a cytotoxin, a chemotherapeutic drug, an immunosuppressant, or a radioisotope. Cytotoxic agents include any agent that is detrimental to cells. Examples of suitable cytotoxic agents and chemotherapeutic agents for the formation of immunoconjugates are known in the art (see, for example, WO 05/103081).

Therapeutic drug and administration.

This invention relates to pharmaceutical compositions containing the antigen-binding molecules of this invention. Pharmaceutical compositions of the invention are formulated with suitable carriers, excipients, and other agents that provide improved transfer, delivery, tolerability, and the like. A variety of suitable formulations can be found in a formulary known to all pharmaceutical chemists: Remington's Pharmaceutical Sciences, Mack Publishing Company, Easton, PA. These formulations include, for example, powders, pastes, ointments, jellies, waxes, oils, lipids, (cationic or anionic) lipids containing vesicles (such as LIPOFECTIN™, Life Technologies, Carlsbad, CA), DNA conjugates, anhydrous absorption pastes, oil-in-water and water-in-oil emulsions, carbovax emulsions (polyethylene glycols of various molecular weights), semi-solid gels and semi-solid mixtures containing carbovax. See also Powell et al. Compendium of excipients for parenteral formulations, PDA (1998), J. Pharm. sci. Technol. 52:238-311.

The dose of antigen binding molecule administered to a patient may vary depending on the age and size of the patient, the target disease, conditions, route of administration, and the like. The preferred dose is usually calculated according to body weight or body surface area. When the bispecific antigen-binding molecule of this invention is used for therapeutic purposes in an adult patient, it may be advantageous to intravenously administer the bispecific antigen-binding molecule of this invention, typically at a single dose of from about 0.01 to about 20 mg/kg body weight, more preferably from about 0.02 to about 7, about 0.03 to about 5, or about 0.05 to about 3 mg/kg body weight. Depending on the severity of the condition, the frequency and duration of treatment can be adjusted. Effective doses and schedules for administering a bispecific antigen-binding molecule can be determined empirically; for example, the patient's progress can be monitored by periodic evaluation, and the dose adjusted accordingly. In addition, interspecies dose scaling can be performed using methods well known in the art (eg, Mordenti et al., 1991, Pharmaceut. Res. 8:1351).

Various delivery systems are known and can be used to administer the pharmaceutical composition of the invention, such as encapsulation in liposomes, microparticles, microcapsules,

- 27 040594 recombinant cells capable of expressing mutant viruses mediated by receptor endocytosis (see, for example, Wu et al., 1987, J. Biol. Chem. 262:4429-4432). Routes of administration include, but are not limited to, intradermal, intramuscular, intraperitoneal, intravenous, subcutaneous, intranasal, epidural, and oral routes. The composition may be administered by any convenient route, for example by infusion or bolus injection by absorption through epithelial or mucosal linings (eg oral mucosa, rectal and intestinal mucosa, etc.) and may be administered together with other biologically active agents . The introduction can be systemic or local.

The pharmaceutical composition of this invention may be administered subcutaneously or intravenously with a standard needle and syringe. In addition, with respect to subcutaneous delivery, a syringe delivery device can easily have applications for delivering the pharmaceutical composition of this invention. Such a syringe delivery device may be reusable or disposable. A reusable syringe delivery device typically uses a replaceable cartridge that contains a pharmaceutical composition. Once all of the pharmaceutical composition within the cartridge has been introduced and the cartridge is empty, the empty cartridge can be easily discarded and replaced with a new cartridge that contains the pharmaceutical composition. The syringe delivery device can then be reused. There is no replaceable cartridge in the disposable syringe delivery device. Rather, a disposable syringe delivery device is filled with a pharmaceutical composition held in a reservoir within the device. Once the reservoir is empty of the pharmaceutical composition, the entire device is discarded.

Numerous reusable syringe delivery devices and autoinjectors have the function of subcutaneously delivering the pharmaceutical composition of this invention. Examples include, but are not limited to, AUTOPEN™ (Owen Mumford, Inc., Woodstock, UK), DISETRONIC™ pen (Disetronic Medical Systems, Bergdorf, Switzerland), HUMALOG MIX 75/25™ pen, pen HUMALOG™, HUMALIN 70/30™ pen (Eli Lilly and Co., Indianapolis, Ind.), NOVOPEN™ I, II and III (Novo Nordisk, Copenhagen, Denmark), NOVOPEN JUNIOR™ (Novo Nordisk, Copenhagen, Denmark) ), the BD™ pen (Becton Dickinson, Frankling Lake, NJ), OPTIPEN™, OPTIPEN PRO™, OPTIPEN STARLET™ and OPTICLIK™ (sanofi-aventis, Frankfurt, Germany), to name but a few. Examples of disposable syringe delivery devices having the function of subcutaneously delivering the pharmaceutical composition of this invention include, but are not limited to, the SOLOSTAR™ pen (sanofi-aventis), FLEXPEN™ (Novo Nordisk) and KWIKPEN™ (Eli Lilly), SURECLICK™ Autoinjector (Amgen, Thousand Oaks, CA), PENLET™ (Haselmeier, Stuttgart, Germany), EPIPEN (Dey, L.P.) and HUMIRA™ pen (Abbott Labs, Abbott Park IL), to name but a few.

In certain situations, the pharmaceutical composition may be delivered in a controlled release system. In one embodiment, a pump may be used (see Langer, formerly; Sefton, 1987, CRC Crit. Ref. Biomed. Eng. 14:201). Alternatively, polymeric materials may be used; see Medical Applications of Controlled Release, Langer and Wise (eds.), 1974, CRC Pres., Boca Raton, Florida. In yet another embodiment, the controlled release system may be located near the target of the composition, requiring only a fraction of the systemic dose (see, e.g., Goodson, 1984, in Medical Applications of Controlled Release, supra, vol. 2, p. 115-138) . Other controlled release systems are reviewed by Langer, 1990, Science, 249:1527-1533.

Injectable preparations may include dosage forms for intravenous, subcutaneous, intradermal and intramuscular injections, drip infusions, etc. These injectable preparations can be obtained by conventional methods. For example, injectable preparations can be prepared, for example, by dissolving, suspending, or emulsifying an antibody or salt thereof, as described above, in a sterile aqueous or oily medium commonly used for injection. As an aqueous injection medium, there are, for example, physiological saline, isotonic saline containing glucose and other auxiliary agents, etc., which can be used in combination with a suitable solubilizing agent such as an alcohol (for example, ethanol) (for example, propylene glycol, polyethylene glycol), non-ionic surfactant (e.g., polysorbate 80, HCO-50 (polyoxyethylene (50 mol) hydrogenated castor oil adduct), etc.) As the oily medium, for example, sesame oil, soybean oil, etc. .d., which can be used in combination with a solubilizing agent such as benzyl benzoate, benzyl alcohol, etc. The injection thus obtained is preferably placed in an appropriate ampoule.

Preferably, the pharmaceutical compositions for oral or parenteral administration described above are formulated into unit dose dosage forms suitable for dosing the active ingredients. Such unit dosage forms include, for example, tablets, pills, capsules, injections (ampoules), suppositories, and the like. The amount of the above antibody is usually from 5 to 500 mg per dosage form in a unit dose; especially in the form of an injection, it is preferred that the above antibody be present in an amount of from about 5 to about

100 mg and about 10 to about 250 mg for other dosage forms.

Therapeutic applications of antigen-binding molecules.

This invention includes methods comprising administering to a subject in need thereof a therapeutic composition comprising an antitumor antibody, or an antigen-binding fragment thereof, or a bispecific antigen-binding molecule that specifically binds weakly or has undetectable binding to CD3 and binds a tumor-associated antigen. The therapeutic composition may contain any of the antibodies or bispecific antigen-binding molecules described herein and a pharmaceutically acceptable carrier or diluent. As used herein, the term subject in need means a human or non-human animal that exhibits one or more symptoms or signs of cancer (e.g., a subject expressing a tumor or suffering from any of the cancers mentioned below) or who otherwise would benefit from inhibition or reduction of tumor activity or depletion of tumor cells (eg PSMA++ prostate cancer cells).

The antibodies and bispecific antigen-binding molecules of the invention (and therapeutic compositions containing them) are suitable, inter alia, for the treatment of any disease or disorder in which stimulating, activating and/or targeting an immune response would be beneficial. In particular, the bispecific antigen binding molecules of this invention can be used to treat, prevent and/or improve any disease or disorder associated with or mediated by a TAA expressing cell, such as PSMA expression or PSMA+ cell activity or proliferation. The mechanism of action by which the therapeutic methods of the invention are achieved involves the lysis of cells expressing tumor associated antigens in the presence of effector cells, eg CDC, apoptosis, ADCC, phagocytosis, or combinations of two or more of these mechanisms. Cells expressing tumor-associated antigens such as PSMA, MUC16, STEAP2, or EGFRvIII that can be inhibited or killed using the bispecific antigen-binding molecules of the invention include, for example, prostate tumor cells.

The antigen binding molecules of this invention can be used to treat, for example, primary and/or metastatic tumors arising in the brain and meninges, head and neck, oropharynx, pulmonary and bronchial tree, gastrointestinal tract, male and female reproductive tract, muscles, bones, skin and appendages, connective tissue, spleen, immune system, hematopoietic cells and bone marrow, liver and urinary tract, kidney, bladder and/or special sensory organs such as the eye. In some embodiments, the bispecific antigen binding molecules of the invention are used to treat one or more of the following cancers: pancreatic carcinoma, head and neck cancer, prostate cancer, malignant gliomas, osteosarcoma, colorectal cancer, breast cancer, gastric cancer (e.g., gastric cancer with MET amplification), malignant mesothelioma, multiple myeloma, ovarian cancer, small cell lung cancer, non-small cell lung cancer, synovial sarcoma, thyroid cancer, breast cancer, melanomaglioma, breast cancer (eg, ductal or intrauterine breast carcinomas) , squamous cell carcinoma, cancer of the esophagus, ulcerative cell carcinoma of the kidney, chromophobic carcinoma of the kidney, (renal) oncocytoma, (renal) transient cell carcinoma, urothelial carcinoma (bladder), adenocarcinoma, or small cell carcinoma (bladder). In accordance with some embodiments of the present invention, the bispecific antibodies are useful in treating a patient suffering from treatment-refractory or treatment-resistant cancer, eg, castration-resistant prostate cancer. In accordance with exemplary embodiments of the invention, methods are provided comprising administering an anti-CD3/anti-PSMA bespecifically antigen-binding molecule disclosed herein to a patient suffering from castration-resistant prostate cancer. Analytical/diagnostic methods known in the art, such as tumor scanning, etc., can be used to determine if a patient contains a castration-resistant tumor.

The invention also includes methods for treating residual cancer in a subject. As used herein, the term residual cancer refers to the existence or persistence of one or more cancer cells in a subject following treatment with an anticancer therapy, such as first line therapy or standard therapy.

According to some aspects, the invention provides methods of treating cancer associated with TAA expression (e.g., prostate cancer associated with PSMA expression or STEAP2 expression, glioblastoma associated with EGFRvIII expression, or ovarian cancer associated with MUC16 expression), comprising administering one or more of the bispecific antigen-binding molecules described elsewhere herein to a subject after the subject has been determined to have cancer. For example, this invention includes methods of treating prostate cancer comprising administering an anti-CD3/anti-TAA bispecific antigen-binding molecule to a patient 1 day, 2 days, 3 days, 4 days, 5 days, 6 days, 1 week, 2 weeks, weeks, or 4 weeks, 2 months, 4 months, 6 months, 8 months, 1 year or more after the subject has received previous therapy.

Combination therapy and formulations.

This invention relates to methods that include the introduction of a pharmaceutical composition containing any of the exemplary antibodies and bispecific antigen-binding molecules described herein, in combination with one or more additional therapeutic agents. Illustrative additional therapeutic agents that can be combined or administered in combination with the antigen binding molecule of this invention include, for example, an antibody against programmed cell death 1 (for example, an anti-PD1 antibody as described in US Patent Application Publication No. US2015/ 0203579A1), an anti-programmed cell death ligand-1 (eg, an anti-PD-L1 antibody as described in US Patent Application Publication No. US2015/0203580A1), an EGFR antagonist (eg, an anti-EGFR antibody [eg, cetuximab or panitumumab] or small molecule EGFR inhibitor [eg, gefitinib or erlotinib]), an antagonist of another EGFR family member such as Her2/ErbB2, ErbB3, or ErbB4 (eg, an anti-ErbB2, anti-ErbB3, or anti-ErbB4 antibody or small molecule inhibitor of ErbB2, ErbB3, or ErbB4), an EGFRvIII antagonist (eg, an antibody that specifically binds EGFRvIII), a cMET antagonist (eg, an anti-cMET antibody), an IGF1R antagonist (eg. , anti-IGF1R antibody), B-raf inhibitor (eg, vemurafenib, sorafenib, GDC-0879, PLX-4720), PDGFR-α inhibitor (eg, anti-PDGFR-a antibody), PDGFR-β inhibitor (eg, anti- - PDGFR-β antibody), VEGF antagonist (e.g. VEGF-Trap, see e.g. US 7087411 (also referred to herein as VEGF inhibitory fusion protein), anti-VEGF antibody (e.g. bevacizumab), small molecule receptor kinase inhibitor VEGF (eg sunitinib, sorafenib or pazopanib)), DLL4 antagonist (eg anti-DLL4 antibody disclosed in US 2009/0142354 such as REGN421), Ang2 antagonist (eg anti-Ang2 antibody disclosed in US 2011/0027286 such as H1H685P), a FOLH1 antagonist (PSMA), a PRLR antagonist (for example, an anti-PRLR antibody), a STEAP1 or STEAP2 antagonist (for example, an anti-STEAP1 antibody or an anti-STEAP2 antibody), a TMPRSS2 antagonist (for example, an anti-TMPRSS2 antibody), MSLN antagonist (eg, anti-MSLN antibody), CA9 antagonist (eg, anti-CA9 antibody), uropl antagonist akines (for example, an antibody against uroplakin), etc. Other agents that may be advantageously administered in combination with the antigen binding molecules of the invention include cytokine inhibitors, including small molecule cytokine inhibitors, and antibodies that bind to cytokines such as IL-1, IL-2, IL-3, IL-4, IL-5, IL-6, IL-8, IL-9, IL-11, IL-12, IL-13, IL-17, IL-18 or their respective receptors. Pharmaceutical compositions of this invention (for example, pharmaceutical compositions containing the anti-CD3/anti-PSMA bispecific antigen-binding molecule described herein) may also be administered as part of a therapeutic regimen comprising one or more therapeutic combinations selected from ICE: ifosfamide ( eg Ifex®), carboplatin (eg Paraplatin®), etoposide (eg Etopophos®, Toposar®, VePesid®, VP-16); DHAP: dexamethasone (eg, Decadron®), cytarabine (eg, Cytosar-U®, cytosine-arabinoside, ara-C), cisplatin (eg, Platinol®-AQ), and ESHAP: etoposide (eg, Etopophos®, Toposar®, VePesid®, VP-16), methylprednisolone (eg Medrol®), high dose cytarabine, cisplatin (eg Platinol®-AQ).

The invention also includes therapeutic combinations comprising any of the antigen binding molecules mentioned herein and an inhibitor of one or more of VEGF, Ang2, DLL4, EGFR, ErbB2, ErbB3, ErbB4, EGFRvIII, cMet, IGF1R, B-raf, PDGFR- α, PDGFR-β, FOLH1 (PSMA), PRLR, STEAP1, STEAP2, TMPRSS2, MSLN, CA9, uroplakin, or any of the above cytokines, wherein the inhibitor is an aptamer, antisense, ribozyme, siRNA, peptibody, nanoparticle, or antibody fragment (eg Fab fragment, F(ab')2 fragment, Fd fragment, Fv fragment, scFv, dAb fragment, or other modified molecules such as diabodies, tribodies, tetrabodies, minibodies, and minimal recognition units). The antigen binding molecules of the invention may also be administered and/or combined in combination with antivirals, antibiotics, analgesics, corticosteroids and/or NSAIDs. The antigen binding molecules of the invention may also be administered as part of a treatment regimen that also includes radiation therapy and/or conventional chemotherapy.

Additional therapeutically active component(s) can be administered immediately before, simultaneously with or shortly after administration of the antigen-binding molecule of this invention; (for the purposes of this disclosure, such administration regimens are considered to be the administration of an antigen-binding molecule in combination with an additional therapeutically active component).

The invention includes pharmaceutical compositions in which the antigen-binding molecule of the invention is combined with one or more additional therapeutic active ingredient(s) described elsewhere herein.

Introduction modes.

- 30 040594

In accordance with some embodiments of the present invention, multiple doses of a bispecific antigen-binding molecule (eg, an anti-TAA antigen-binding molecule) may be administered to a subject over a period of time. Methods in accordance with this aspect of the invention include sequential administration of several doses of an antigen-binding molecule of the invention to a subject. As used herein, the term sequential administration means that each dose of an antigen-binding molecule is administered to a subject at a different point in time, such as on different days separated by a predetermined interval (eg, hours, days, weeks, or months). The present invention includes methods that include sequentially administering to a patient a single initial dose of an antigen-binding molecule followed by one or more secondary doses of an antigen-binding molecule and optionally followed by one or more tertiary doses of an antigen-binding molecule.

The terms initial dose, secondary doses and tertiary doses refer to the time sequence of administration of the antigen-binding molecule of the invention. Thus, the initial dose is the dose that is administered at the beginning of the treatment regimen (also called the base dose); secondary doses - doses administered after the initial dose; and tertiary doses are doses administered after secondary doses. Initial, secondary and tertiary doses may contain the same amount of antigen-binding molecule, but usually may differ from each other in terms of frequency of administration. However, in some embodiments, the amount of antigen-binding molecule contained in the initial, secondary, and/or tertiary doses differs from one another (eg, adjusted up or down as needed) during treatment. In some embodiments, two or more (eg, 2, 3, 4, or 5) doses are administered at the start of a treatment regimen as loading doses, followed by doses that are administered less frequently (eg, maintenance doses).

In one illustrative embodiment of the present invention, each secondary and/or tertiary dose is administered from 1 to 26 weeks (for example, 1, 11/2, ₂ , 21/2, 3, 31/2, ₄ , 41/2, 5, 51 /2, 6, 61/2, 7, 71/2, 8, 81/ _{2 ,} 9, 91/2, 10, 101/2, 11, 111/2, 12, 121/2, 13, 131/2 , 14, 141/2, 15, 151/2, 16, 161/2, 17, 171/2, 18, 181/2, 19, 191/2, 20, 201/2, 21, 211/2, 22 , 221/2, 23, 231/2, 24, 241/2, 25, 251/2, 26, 261/2 or more) after the immediately preceding dose. The phrase immediately preceding dose as used herein means, in a sequence of multiple administrations, the dose of an antigen-binding molecule that is administered to a patient prior to the administration of the very next dose in the sequence without intermediate doses.

Methods in accordance with this aspect of the invention may include administering any number of secondary and/or tertiary doses of an antigen-binding molecule (eg, a bispecific anti-TAA antigen-binding molecule) to a patient. For example, in some embodiments, only one secondary dose is administered to the patient. In other embodiments, two or more (eg, 2, 3, 4, 5, 6, 7, 8 or more) secondary doses are administered to the patient. Similarly, in some embodiments, only one tertiary dose is administered to the patient. In other embodiments, two or more (eg, 2, 3, 4, 5, 6, 7, 8 or more) tertiary doses are administered to the patient.

In embodiments involving multiple secondary doses, each secondary dose may be administered at the same frequency as the other secondary doses. For example, each secondary dose may be administered to a patient 1-2 weeks after the immediately preceding dose. Similarly, in embodiments involving multiple tertiary doses, each tertiary dose may be administered at the same frequency as the other tertiary doses. For example, each tertiary dose may be administered to a patient 2-4 weeks after the immediately preceding dose. Alternatively, the frequency with which secondary and/or tertiary doses are administered to a patient may vary over the course of treatment. The frequency of administration may also be adjusted during the course of treatment by the physician, depending on the needs of the individual patient after clinical examination.

Examples

The following examples are provided to provide those skilled in the art with a complete disclosure and description of the methods of making and using the methods and compositions of the invention and are not intended to limit the scope of the invention which the inventors consider to be their invention. Efforts have been made to ensure accuracy with regard to the numbers used (eg quantities, temperatures, etc.), but some experimental errors and deviations should be taken into account. Unless otherwise indicated, parts are parts by weight, molecular weight is average molecular weight, temperature is in degrees Celsius, and pressure is at or near atmospheric pressure.

Example 1 Creation of anti-CD3 antibodies.

The following procedures were aimed at detecting antibodies that specifically recognized CD3 (T-cell co-receptor) as an antigen.

A pool of anti-CD3 antibodies was obtained by immunization of genetically modified mice. Briefly, mice genetically engineered to express reverse chimeric (human variable region, mouse constant region) and immunoglobulin heavy chains associated with a single rearranged light chain (e.g., Vk1-39/J or Vk3-20/J) were immunized with CD3 antigen.

- 31 040594 and produced B cells that included a variety of human VH rearrangements to express a diverse repertoire of high affinity antigen-specific antibodies. Some of the exemplary antibodies described herein were recombinantly generated and expressed the same Vk1-39JK5 light chain sequence (LCVR shown in SEQ ID NO: 162), while other antibodies that recombinantly expressed the cognate light chain of one of the arms of the heavy chains (for example, a shoulder targeting a tumor).

The generated antibodies were tested for affinity for human and cynomolgus CD3 antigen in an in vitro binding assay, and for example, one anti-CD antibody designated CD3-VH-P (HCVR shown in SEQ ID NO: 154) was identified among several others. which were found to bind to human and cynomolgus CD3 and had an EC ₅₀ of 1 to 40 nM affinity (+++) as determined by FACS titration of Jurkat cells and cynomolgus T cells, respectively. See, for example, the FACS binding experiments described below in Example 4.

The amino acid residues of the CD3-VH-P germline were then identified and the antibody, designated CD3-VH-G, was designed to contain only germline structures. Other antibody derivatives have been developed by known molecular cloning techniques to replace amino acid residues in steps based on differences between the germline sequence and the CD3-VH-P sequence. Each antibody derivative was assigned a CD3-VH-G designation number. See table. 1 and FIG. 1.

Bispecific antibodies containing the first binding agent derived from engineered anti-CD3 antibodies with the symbols and descriptions presented in table. 1 and a second binding arm derived from anti-TAA antibodies were generated and tested for monovalent affinity for CD3-bearing cells in a FACS assay (as described in Example 4). The results of the monovalent binding affinity of these bispecific antibodies are shown in the two right columns of the table. 1. In specific examples, bispecific antibodies having a TAA binding arm and a CD3 binding arm designated CD3-VH-G, CD3-VH-G5 and CD3-VH-G20, respectively, bound Jurkat cells with an EC _{50 of} 2.7 E- 08, non-detectable binding and 5.5 E-07, respectively.

Table 1

Mutations for CDRs based on the germline sequence and corresponding FACS binding affinities for each ___________________engineered antibody___________________

Antibody Designation CD3-VH Description of mutations compared to the CD3-VH-G* antibody JURKAT Javanese T cells O toque CD3-VH-G CD3-VH-G2 CD3-VH-G3 CD3-VH-G4 CD3-VH-G5 Framework regions (FR) only germline (GL); CDR-CD3-VHP All GL (FR and CDR). All GL (FR and CDR). Addition of A33S. All GL (FR and CDR). Addition of Y105K. All GL (FR and CDR). Addition of A33S and Y105K. (+ + +) (-) (-) (-) (-) (+++) (-) (-) (-) (-) CD3-VH-G8 Framework regions of the germline. Addition of K58I. (+ + +) (+) CD3-VH-G9 Framework regions of the germline. Adding Y99D (+) (-) CD3-VH- Framework regions of the germ line. (+) (-)

- 32 040594

G10 Addition of H108Y. CD3-VHG11 Framework regions of the germ line. Addition of L111M. (+ + +) (+) CD3-VHG12 Framework regions of the germline. Adding K58I, Y99D (+ +) (+/-) CD3-VHG13 Framework regions of the germline. Adding K58I, H108Y (+ +) (+) CD3-VHG14 Framework regions of the germline. Adding K58I, L111M (+ + +) (++) CD3-VHG15 Framework regions of the germline. Adding Y99D,—H108Y (+) (-) CD3-VHG16 Framework regions of the germline. Adding Y99D, L111M (++) (+/-) CD3-VHG17 Framework regions of the germline. Adding H108Y, L111M (+/-) (+/-) CD3-VHG18 Framework regions of the germline. Adding K58I, Y99D, H108Y (+/-) (-) CD3-VHG19 Framework regions of the germline. Adding K58I, Y99D, L111M (+/-) (-) CD3-VHG2 0 Framework regions of the germline. Adding K58I, H108Y, L111M (+/-) (+/-) CD3-VHG21 Framework regions of the germline. Adding Y99D, H1O8Y,-L111M (+/-) (-)

* Sequential numbering based on the mature 7221G protein (CD3-VH-G).

While CD3-VH-G and some other modified antibodies retained their binding affinity as shown in FACS analyses, several anti-CD3 antibodies bound in vitro to human or cynomolgus monkey CD3 with weak (+/-) to ( -) immeasurable. Binding affinities, binding kinetics, and other biological properties to elucidate toxicity and pharmacokinetic (pK) profiles were subsequently investigated for bispecific antibodies containing exemplary anti-CD3 antibodies prepared according to the methods of this example and are detailed in the examples below.

Example 2 Heavy and Light Chain Variable Regions (CDR Amino Acid and Nucleotide Sequences).

Amino acid and nucleotide sequences were determined for each antibody heavy chain sequence.

Each antibody heavy chain derived from the IGHV39*01/D5-12*01/J6*02 germline sequence (SEQ ID NO: 181) was assigned a G number designation for consensus nomenclature. In table. 2 shows the amino acid sequence identifiers for the heavy chain variable regions and CDRs of engineered anti-CD3 antibodies of the invention. The corresponding nucleic acid sequence identifiers are given in table. 3. Amino acid and nucleic acid sequence identifiers for the light chain variable region and CDRs for constructing each recombinant antibody are also identified in Table 1 below. 4 and 5, respectively.

- 33 040594

table 2

Heavy chain amino acid sequence identifiers

Antibody Designated e CD3-VH SEQID NO: HCVR CDR1 CDR2 CDR3 CD3-VH-G 2 4 6 8 CD3-VH-G2 10 12 14 16 CD3-VH-G3 18 20 22 24 CD3-VH-G4 26 28 thirty 32 CD3-VH-G5 34 36 38 40 CD3-VH-G8 42 44 46 48 CD3-VH-G9 50 52 54 56 CD3-VH-G10 58 60 62 64 CD3-VH-G11 66 68 70 72 CD3-VH-G12 74 76 78 80 CD3-VH-G13 82 84 86 88 CD3-VH-G14 90 92 94 96 CD3-VH-G15 98 100 102 104 CD3-VH-G16 106 108 110 112 CD3-VH-G17 114 116 118 120 CD3-VH-G18 122 124 126 128 CD3-VH-G19 130 132 134 136 CD3-VH-G20 138 140 142 144 CD3-VH-G21 146 148 150 152 CD3-VH-P 154 156 158 160

Table 3

Heavy chain nucleic acid sequence identifiers

Antibody Designation CD3-VH SEQID NO: HCVR CDR1 CDR2 CDR3 CD3-VH-G 1 3 5 7 CD3-VH-G2 9 eleven 13 15 CD3-VH-G3 17 19 21 23 CD3-VH-G4 25 27 29 31 CD3-VH-G5 33 35 37 39 CD3-VH-G8 41 43 45 47 CD3-VH-G9 49 51 53 55 CD3-VH-G10 57 59 61 63 CD3-VH-G11 65 67 69 71 CD3-VH-G12 73 75 77 79 CD3-VH-G13 81 83 85 87 CD3-VH-G14 89 91 93 95 CD3-VH-G15 97 99 101 103 CD3-VH-G16 105 107 109 111 CD3-VH-G17 113 115 117 119 CD3-VH-G18 121 123 125 127 CD3-VH-G19 129 131 133 135 CD3-VH-G20 137 139 141 143 CD3-VH-G21 145 147 149 151 CD3-VH-P 153 155 157 159

-34040594

Table 4

Light chain amino acid sequence identifiers

Antibody Designated e ULC SEQID NO: LCVR CDR1 CDR2 CDR3 VK1-39JK5 162 164 166 168

Table 5

Light chain nucleic acid sequence identifiers

Antibody Designation ULC SEQID NO: LCVR CDR1 CDR2 CDR3 VK1-39JK5 161 163 165 167

Control antibody 1, designated CD3-L2K, was constructed from a known anti-CD3 antibody (ie anti-CD3 antibody L2K as set forth in WO 2004/106380).

The isotype control antibody indicated in the examples below is an isotype-identical antibody (modified with IgG4) that interacts with an irrelevant antigen, i.e. FelD1 antigen.

Example 3 Generation of bispecific ULC antibodies that bind CD3 and tumor associated antigens (TAA).

Bispecific antibodies containing an anti-CD3 specific binding domain and an anti-TAA specific binding domain such as PSMA, EGFRvIII, MUC16, or STEAP2 were constructed using standard molecular biology techniques using the heavy chain from the described anti-CD3 antibody in this case. a heavy chain from an anti-TAA antibody; and a conventional light chain or a universal light chain (ULC). The anti-TAA antibodies used to construct the bispecific antibodies of this invention were obtained by immunizing genetically modified mice.

A summary of the constituent parts of the antigen-binding domains of various bispecific antibodies obtained in accordance with this example, is given in table. 6-8. All bispecific antibodies were made with a modified (chimeric) IgG4 Fc domain as described in U.S. Patent Application Publication No. US20140243504A1, published August 28, 2014. Exemplary EGFRvIIIxCD3 bispecific antibodies can be made using any of the heavy chain and light chain variable regions ( or CDR) of any of the EGFRvIII antibodies discussed in US Patent Application Publication No. US20150259423, which is incorporated herein by reference in its entirety, in conjunction with the variable regions or CDRs of any of the anti-CD3 antibodies discussed herein.

Table 6

Creation of PSMAxCD3 bispecific antibodies

bispecific identifier about antibodies Antigen binding domain anti-PSMA antibodies Antigen binding ing domain anti-CDS antibodies Common light chain variable region

- 35 040594

Varna white heavy chain region heavy chain varnable region BSPSMA/CD3-003 PSMA-VH-B CD3-VH-G VK1-39JK5 BSPSMA/CD3-200 CD3-VH-G2 BSPSMA/CD3-300 CD3-VH-G3 BSPSMA/CD3-400 CD3-VH-G4 BSPSMA/CD3-004 CD3-VH-G5 BSPSMA/CD3-800 CD3-VH-G8 BSPSWCD3-900 CD3-VH-G9 BSPSMA/CD3-1000 CD3-VH-G10 BSPSMA/CD3-1100 CD3-VH-G11 BSPSMA/CD3-1200 CD3-VH-G12 BSPSWCD3-1300 CD3-VH-G13 BSPSWCD3-1400 CD3-VH-G14 BSPSMA/CD3-1500 CD3-VH-G15 BSPSMA/CD3-1600 CD3-VH-G16 BSPSMA/CD3-1700 CD3-VH-G17 BSPSWCD3-1800 CD3-VH-G18 BSPSMA/CD3-1900 CD3-VH-G19 BSPSWCD3-005 CD3-VH-G20 BSPSWCD3-2100 CD3-VH-G21

Table 7

Creation of bispecific antibodies EGFRvIIIxCD3

Bispecific antibody identifier Antigen binding domain anti-EGFRvIII antibodies Antigen binding ing domain anti-POPs antibodies Common light chain variable region heavy chain varnable region heavy chain varnable region BSV3/CD3-001 EGFRvIII-VH-A CD3-VH-G EGFRvIII-VL- A BSV3/CD3-002 CD3-VH-G5 BSV3/CD3-003 CD3-VH-G9 BSV3/CD3-004 CD3-VH-G10

- 36 040594

Table 8

Creation of bispecific antibodies MUC16xCD3 Bispecific antibody identifier Antigen-binding domain of an anti-MUC16 antibody Antigen binding domain anti-POPs antibodies Common light chain variable region heavy chain varnable region heavy chain varnable region BSMUC16/CD3- 001 MUC16-VH-A CD3-VH-G MUC16-VL-A BSMUC16/CD3002 CD3-VH-G5 BSMUC16/CD3- 003 CD3-VH-G9 BSMUC16/CD3- 004 CD3-VH-G10 BSMUC16/CD3- 005 CD3-VH-G20 Table 9

Creation of STEAP2xCD3 bispecific antibodies

Bispecific antibody identifier Antigen binding domain anti-STEAP2 antibodies Antigen binding domain anti-POPs antibodies Common light chain variable region heavy chain varnable region heavy chain varnable region BSSTEAP2/CD3-001 STEAP2-VH-A CD3-VH-G STEAP2-VL-A BSSTEAP2/CD3-002 CD3-VH-G5 BSSTEAP2/CD3-003 CD3-VH-G20

Each of the exemplary bispecific antibodies was tested in various bioassays as described below.

Example 4 Binding affinities of exemplary bispecific antibodies measured by FACS analysis.

In this example, the ability of CD3xTAA bispecific antibodies to bind to CD3-expressing human and cynomolgus monkey cell lines was determined by FACS. In addition, the ability of these bispecific antibodies to bind to specific specific (TAA-specific) cell lines was confirmed. As described above, the various bispecific antibodies of the invention use one specific TAA binding arm (PSMA, EGFRvIII, MUC16, or STEAP2, see Example 3, Tables 6-8) paired with one of the anti-CD3 binding arm panels (see examples 1 and 2 above) and a common light chain. As also shown in Example 5 by surface plasmon resonance, CD3xTAA bispecific antibodies exhibit a range of affinities for the soluble human hCD3ε/δmFc protein heterodimer.

Briefly, 2x105 cells/well of Jurkat cells expressing human CD3, cynomolgus monkey cells specifically expressing T or TAA were incubated with serial dilutions of the bispecific antibodies for 30 min at 4°C. After incubation, the cells were washed and goat F(ab') ₂ anti-human Fcy PE labeled secondary antibodies (Jackson Immunolabs) were added to the cells for an additional 30 minutes. Cells were then washed, resuspended in cold PBS+1% BSA and analyzed by flow cytometry on a BD FACS Canto II.

For FACS analysis, cells were separated by forward scatter height versus forward scatter area by single event selection followed by side and front scatterers. EC ₅₀ for cell binding titration was determined using software

- 37 040594

PRISM™ (GraphPad Software, Inc., La Jolla, CA). Values were calculated using 4-parameter non-linear regression analysis.

Table 10A

FACS binding on CD3 and PSMA-specific cell lines

Bispecific antibody identifier Anti-POPsbinding shoulder Jurkat T cells of the cynomolgus macaque B16F10.9/PSM A B EU ₅₀ [M] EU ₅₀ [M] EU ₅₀ [M] BSPSMA/CD3-003 BSPSMA/CD3-200 BSPSMA/CD3-300 BSPSMA/CD3-400 BSPSMA/CD3-004 BSPSMA/CD3-800 BSPSMA/CD3-900 BSPSMA/CD3-1000 BSPSMA/CD3-1100 BSPSMA/CD3-1200 BSPSMA/CD3-1300 BSPSMA/CD3-1400 BSPSMA/CD3-1500 BSPSMA/CD3-1600 BSPSMA/CD3-1700 BSPSMA/CD3-1800 BSPSMA/CD3-1900 BSPSMA/CD3-005 BSPSMA/CD3-2100 FACS binding CD3-VH-G CD3-VH-G2 CD3-VH-G3 CD3-VH-G4 CD3-VH-G5 CD3-VH-G8 CD3-VH-G9 CD3-VH-G10 CD3-VH-G11 CD3-VH-G12 CD3-VH-G13 CD3-VH-G14 CD3-VH-G15 CD3-VH-G16 CD3-VH-G17 CD3-VH-G18 CD3-VH-G19 CD3-VH-G20 CD3-VH-G21 on CD3 and EGF 1.65Е-08 NS NS NS Very weak 1.93Е-08 2.74Е-07 2.77Е-07 1.83Е-08 4.72E-08 1.02E-07 3.19E-08 9,ZOE-08 5.68E-08 2.00E-07 1.26E-07 6.07Е-08 2.10E-07 1.06E-07 RvIII-spec 1.42E-08 NS NS NS NS 1.96E-08 NS NS 8.90E-07 NS 2.17E-06 1.70E-07 NS NS 3.35E-06 NS NS 6.14Е-06 NS physical glue 2.26E-09 1.88Е-09 1.90E-09 1.72E-09 1.31Е-09 1.31Е-09 1.43Е-09 1.19E-09 1.03Е-09 1.16E-09 1.25E-09 1,ZOE-09 1.21E-09 1.03Е-09 1.34Е-09 2.16E-09 1.35Е-09 2.09Е-09 1.14Е-09 Table 10 precise lines Bispecific antibody identifier Ahth-CD3 linking arm Jurkat Javanese macaque T cells U87/ EGFRvIII EU ₅₀ [M] ec ₅₀ [M] EU ₅₀ [M] BSV3/CD3-001 CD3-VH-G 1.46E- 09 nt 2.40E-09 BSV3/CD3-002 CD3-VH-G5 Very nt 5.60E-09 weak

Table 10C

FACS binding on CD3 and MUC16-specific cell lines

Bispecific antibody identifier Anti-CD3 binding arm Jurkat Javanese macaque T cells OVCAR3 (MUC16+) EU ₅₀ [M] EU ₅₀ [M] EU ₅₀ [M] BSMUC16/CD3- 001 CD3-VH-G 3.21E- 09 NT 1.20E-09 BSMUC16/CD3- 002 CD3-VH-G5 Very weak nt 2.69Е-09

As shown in Table. 10A, the CD3 binding arms of each CD3xPSMA bispecific antibody display the range of binding affinity for human CD3 expressing Jurkat cells (EC ₅₀ range 15-300 nM). Importantly, the CD3 arms, which exhibit weak binding to the heterodimeric human CD3 protein at surface plasmon resonance (see Table 11), also correlate with weak and unobservable binding on Jurkat cells (i.e.

- 38 040594

CD3-VH-G2, CD3-VH-G3, CD3-VH-G5). No detectable binding or no detectable binding in a FACS assay or equivalent assay means that the affinity between the antibody and its target antigen is beyond the detection limits of the assay (eg, >1 μM). Several CD3-binding arms also showed cross-reactivity with cynomolgus monkey T cells. All bispecific antibodies tested showed similar cell binding on the respective PSMA, EGFRvIII, and MUC16 cell lines, confirming that bispecific pairing with individual CD3 arms did not affect or reduce TAA-specific binding (TAA-specific binding was less than or equal to 5.6 nM ( high affinity) in all tested examples).

Antibodies showing weak or undetectable binding to human CD3, as well as showing weak or undetectable binding to cynomolgus monkey CD3, are considered to be beneficial for avidity-induced bispecific mating according to the present invention and were further tested for in vitro and in vivo cytotoxicity.

Example 5 Binding affinities of exemplary bispecific antibodies measured by surface plasmon resonance analysis.

Binding affinities and kinetic constants of bispecific anti-TAA antibodies x CD3 to soluble heterodimeric hCD3ε/δmFc protein (hCD3ε = UniProtKB/Swiss-Prot: P07766.2; SEQ ID NO: 169; hCD3δ = UniProtKB/Swiss-Prot: P04234.1, SEQ ID NO: 170) was determined by surface plasmon resonance at 37° C. using either antigen capture format (Table 11) or antibody capture format (data not shown). In this example, BSPSMA/CD3 bispecific antibodies were used because these pairs represented the use of a broader panel of antibodies for CD3 binding. The measurements were carried out on a Sierra Sensors MASS-1 instrument.

In the antigen capture format, the surface of the high density amine sensor MASS-1 was derivatized with a mouse polyclonal anti-IgG2a antibody (Southern Biotech). Soluble heterodimeric CD3 protein was captured and appropriate antibodies were administered against the captured antigen.

Rate constants of kinetic association (k _a ) and dissociation (kd) were determined by processing and adjusting the data to a 1:1 binding model using MASS-1 AnalyserR2 curve adjustment software. Binding dissociation equilibrium constants (KD) and dissociative half-lives (t _1/2 ) were calculated from the kinetic rate constants:

K _D (M) \u003d kd / k _a and t _1/2 (min) \u003d (ln2 / (60 * kd).

Table 11

Affinities of bispecific anti-CD3 antibodies to soluble human CD3

Binding at 37°C / Antigen Capture Format Bispecific antibody identifier Corresponding anti-POP antigen binding ka (ms' ¹ ) kd (C ¹ ) K _D (M) PM (min)

- 39 040594

BSPSMA/CD3-003 HCVR identifier CD3-VH-G 1.32E+0 5 7.62E- 04 5.78E- 09 15, 2 BSPSMA/CD3-200 CD3-VH-G2 NS NS NS NS BSPSMA/CD3-300 CD3-VH-G3 NS NS NS NS BSPSMA/CD3-400 CD3-VH-G4 NS NS NS NS BSPSMA/CD3-004 CD3-VH-G5 NS NS NS NS — 5.95E+0 1.15E- 1.94E- BSPSMA/CD3-800 CD3-VH-G8 100 4 03 08 4.38E+0 4.95E- 1.13E- BSPSMA/CD3-900 CD3-VH-G9 2.3 4 03 07 3.44E+0 6.37E- 1.85E- BSPSMA/CD3-1000 CD3-VH-G10 1.8 4 03 07 — 9.21E+0 1.02E- 1, NOT- BSPSMA/CD3-1100 CD3-VH-G11 11.3 4 03 08 3.85E+0 2.47E- 6.42E- BSPSMA/CD3-1200 CD3-VH-G12 4.7 4 03 08 — 2.03E+0 2.48E- 1.22E- BSPSMA/CD3-1300 CD3-VH-G13 4.7 4 03 07 — 6.21E+0 3.31E- 5,ZZE- BSPSMA/CD3-1400 CD3-VH-G14 3.5 4 03 08 — 7.36E+0 6, NOT- 8.29E- BSPSMA/CD3-1500 CD3-VH-G15 1.9 4 03 08 — 6.43E+0 2.43E- 3.78E- BSPSMA/CD3-1600 CD3-VH-G16 4.7 4 03 08 4.70E+0 3.07E- 6.52E- BSPSMA/CD3-1700 CD3-VH-G17 3, 8 4 03 08 BSPSMA/CD3-1800 CD3-VH-G18 NS NS NS NS 4.43E+0 5.09E- 1.15E- BSPSMA/CD3-1900 CD3-VH-G19 2.3 4 03 07 — 1.73E+0 5.77E- 3.34E- BSPSMA/CD3-005 CD3-VH-G20 20 4 03 07 3.02E+0 2.34E- 7.75E- BSPSMA/CD3-2100 CD3-VH-G21 4.9 4 03 08 3.68E+0 2.66E- 7.22E- Control 1 CD3-L2K 4.3 5 03 09

HC: No binding found.

As shown in Table. 11, all generated aHTu-CD3x-aHTu-PSMA bispecific antibodies retain very weak binding to soluble CD3 in the surface plasmon resonance binding assay, e.g. having a KD value of greater than 11 to 334 nM, which is weaker than that of the anti-CD3 bispecific arm. derived from the germline framework regions -CD3-VH-G.

Several bispecific antibodies showed KD values greater than 50 nM and some KD values were greater than 100 nM (>1 x 10 ^-7 ) (i.e. BSPSMA/CD3-900, BSPSMA/CD3-1000, BSPSMA/CD3-1900), KD values greater than 300 nM (>3x10 ^-7 ) (i.e. BSPSMA/CD3-005) and even beyond the detection limit of the assay (>500 nM; >5x10 ^-7 ), i.e. shows no detectable binding to soluble human CD3 (i.e. BSPSMA/CD3-200, BSPSMA/CD3-300, BSPSMA/CD3-400, BSPSMA/CD3-004 and BSPSMA/CD3-1800).

Example 6 T cell activation and tumor-specific cytotoxicity exhibited by the bispecific antibodies of the invention measured in vitro.

In this example, the specific killing of PSMA, EGFRvIII, or MUC16 expressing TAA target cells in the presence of CD3-based bispecific antibodies was examined by flow cytometry. As previously reported, the bispecific antibodies exhibited a range of affinities for the CD3 protein and CD3 expressing cell lines (ie, weak, moderate, and strong binding). The same group of bispecific antibodies were tested for their ability to induce naive human T cells to redirect lysis to cells expressing the target.

Briefly, PSMA-expressing (C4-2, 22Rv1 and TRAMPC2_PSMA), EGFRvIn-expressing (U87/EGFRvIII), or MUC16-expressing (OVCAR3) cell lines were labeled with 1 μM Violet Cell Tracker fluorescent tracking dye.

After labeling, the cells were seeded overnight at 37°C. Separately, human PBMCs were plated in RPMI supplemented medium at 1x106 cells/ml and incubated overnight at 37°C to enrich lymphocytes by destroying adherent macrophages, dendritic cells and some monocytes. The following day, target cells were co-incubated with adherent depleted naïve PBMCs (effector/target ratio 4:1) and serial dilutions of the appropriate bispecific antibodies or isotype control (concentration range: 66.7 nM to 0.25 pM) for 48 hours at 37°C. Cells were removed from the cell culture plate using enzyme-free cell dissociation buffer and analyzed by FACS.

For FACS analysis, cells were stained with a far red dead/live cell tracker (Invitrogen). 5x105 counting pellets were added to each well just prior to FACS analysis. For each sample, 1x104 pellets were collected. To assess the specificity of lysis, cells were separated by living violet-stained populations. The percentage of the living population was recorded and used to calculate normalized survival.

T cell activation was assessed by incubating cells with directly conjugated anti-CD2 and CD69 antibodies and by reporting the percentage of activated (CD69+) T cells out of total (CD2+) T cells.

As the results in Table. 12A-12C, depletion of TAA-expressing cells was observed using anti-PSMA, anti-EGFRvIn, or anti-MUC16xCD3 bispecific antibodies. Most of the bispecific antibodies tested activated and directed human T cells to deplete target cells with EC ₅₀ in the picomolar range. In addition, the observed lysis (depletion) of the target cell was associated with upregulation of CD69 cells on CD2+ T cells with picomolar (pM) EC ₅₀ .

Importantly, the results of this example demonstrate that several bispecific antibodies that used CD3 binding that exhibited weak or unobservable binding to CD3 protein or CD3-expressing cells (i.e., CD3-VH-G5) still retained the ability to activate T-cells and showed strong cytotoxicity of tumor antigen-expressing cells.

- 41 040594

Table 12A

Cytotoxicity and T cell activation properties of selected PSMAxCD3 bispecific antibodies

Identifier of obispecific antibody Anti-CD3 bind her shoulder cell depletion C4-2 EC ₅₀ [M] 22RV1 cell depletion EC ₅₀ [M] TgashpC2 . PSMA cell depletion EC ₅₀ [M] T cell activation EC ₅₀ [M] BSPSMA/ CD3-VH-G 1,OSE-11 nt 6.43E-12 1.23E-12 CD3-003 — Absences Absence ie CD3-VH-G2 NT NT BSPSMA/ activity active CD3-200 ti BSPSMA/ CD3-VH-G3 NT Very weak NT 1.85E-11 CD3-300 BSPSMA/ CD3-VH-G4 NT Very weak NT Very CD3-400 weak BSPSMA/ CD3-VH-G5 2.15E-11 6.31E-12 1.15E-11 1.34Е-11 CD3-004 BSPSMA/ CD3-VH-G8 NT NT 9.27E-12 1.76E-12 CD3-800 BSPSMA/ CD3-VH-G9 NT NT 3.50E-12 1.12E-12 CD3-900 BSPSMA/ CD3-VH- NT NT 5.97E-12 1.28E-12 CD3-1000 GIO BSPSMA/ CD3-VH- NT NT 3.86E-12 1.11E-12 CD3-1100 Gil BSPSMA/ CD3-VH- 8.74Е-12 NT NT 2.31E-12 CD3-1300 G13 BSPSMA/ CD3-VH- 7.37E-12 2.07Е-12 NT 3.89E-12 CD3-1700 G17 BSPSMA/ CD3-VH- 1.39E-11 8.32E-12 NT 6.11E-12 | CD3-005 G20 1

NT = not tested.

Table 12B

Cytotoxicity and T cell activation properties of selected EGFRvIIIxCD3 bispecific antibodies

Bispecific antibody identifier Anti-POP binding shoulder U87_EGFRvIII depletion of EC ₅₀ cells [M] Activation of T cells EC ₅₀ [M] BSV3/CD3-001 CD3-VH-G 3.64Е-10 3,ZZE-11 BSV3/CD3-002 CD3-VH-G5 1,ZOE-09 1.13E-10

Table 12C

Cytotoxicity and T cell activation properties of selected MUC16xCD3 bispecific antibodies

Bispecific antibody identifier Anti-POP binding shoulder OVCAR3 cell depletion EC ₅₀ [M] T cell activation EC ₅₀ [M] BSV3/CD3-001 CD3-VH-G 2.24Е-11 5.88E- 12 BSV3/CD3-002 CD3-VH-G5 3 _Z 06E-11 1.01E- eleven

- 42 040594

Example 7 Bispecific aHmu-PSMA/aHTu-CD3 antibodies show strong anti-tumor efficacy in vivo.

To determine the in vivo efficacy of exemplary anti-PSMA/anti-CD3 bispecific antibodies identified as having weak or no detectable binding affinity for human and cynomolgus monkey CD3, studies were performed in immunocompromised mice bearing human prostate cancer xenografts. Additional studies were also performed in immunocompetent mice bearing mouse prostate cancer xenografts engineered to express human PSMA.

Efficacy of anti-PSMA/anti-CD3 bispecific antibodies in human tumor xenograft models.

To evaluate the in vivo efficacy of anti-PSMA/anti-CD3 bispecific antibodies in human tumor xenograft studies in NOD SCID gamma (NSG) mice (Jackson Laboratories, Bar Harbor, Maine), human peripheral blood mononuclear cells (PBMCs) were co-implanted with tumor cells. human prostate 22Rv1 or C4-2 cells that endogenously express PSMA.

Briefly, 4 x 10 ⁶ 22Rv1 cells or 5 x 10 ⁶ C4-2 cells (MD Anderson, TX) were co-implanted subcutaneously with 1 x 10 ⁶ human PBMC (ReachBio, LLC., Seattle, WA) in a 50:50 mixture with Matrigel (BD Biosciences ) in the right flank of male NSG mice. In the C4-2 study, mice were injected intraparenterally on days 0, 4 and 7 after tumor implantation with 0.1 mg/kg BSPSMA/CD3-003 or BSPSMA/CD3-005.

In an additional xenogeneic model, bispecific anti-PSMA/anti-CD3 antibodies were tested in mice inoculated with human CD34+ hematopoietic stem cells. Briefly, newborn SIRPa BALB/c-Rag2-IL2rY-(BRG) mice were inoculated with hCD34 ⁺ fetal liver cells. 3-6 months later, hCD34-grafted SIRPa BRG mice were then implanted with C4-2 cells (5 x 10 ⁶ subcutaneously in Matrigel). After 8 days, mice were treated with 10 μg of BSPSMA/CD3-004 or isotype control antibody, followed by 2x/week doses throughout the study.

In all studies, tumor size was measured twice a week using calipers and tumor volume calculated as volume=(length x width ² )/2.

As the results in Table. 13, the bispecific antibodies tested in the xenogeneic models described above were effective in suppressing tumor growth compared to the isotype control.

Table 13

Suppression of tumor growth by introducing anti-PSMA/anti-CD3 bispecific antibodies into xenogeneic mouse models

Xenogeneic model: suppression of tumor growth

Tumor model/mouse strain N No. mouse/treatment group Bispecific antibody identifier Dose Final tumor volume (mm ³ ) Mean ± SD C4-/2 NSG 5 BSPSMA/CD3-003 0.1 mg/kg on days 0.4 and 7 0±0 5 BSPSMA/CD3-005 0±0 5 Isotype control 960±660 С4-2/ SIRPa Balb/c-Rag2IL2ry-BRG grafted with hCD34+ HSC 5 BSPSMA/CD3-004 1.0 µg/mouse 2x/week 70 ± 60 5 Isotype control 260±180

Efficacy of anti-PSMA/anti-CD3 bispecific antibodies in an immunocompetent tumor model.

In addition, anti-PSMA/CD3 bispecific antibodies were evaluated for antitumor activity in an immunocompetent model (U.S. Provisional Application No. 62/083653 filed November 24, 2014). Mice humanized for CD3 tri-chain (δγε) were also humanized for PSMA and implanted with an alternative murine prostate cancer cell line TRAMP-C2 transfected with human PSMA.

Prior to the study, a variant of the TRAMP-C2_hPSMAv tumor cell lines #1 was produced. Briefly, 7.5 x 10 ⁶ TRAMP-C2_hPSMA cells were implanted subcutaneously in the right flank of male mice humanized for CD3 and PSMA. The tumor was excised and cut into 3 mm fragments and subsequently implanted in the right flank of new humanized male mice. The tumor formed by the implanted tumor fragments was then harvested and disaggregated into a single cell suspension. These cells (TRAMP-C2_hPSMAv #1) were then cultured in vitro with G418 selected. 4x10 ⁶ cells of this variant cell line were then implanted in the right flank of PSMA/CD3 humanized male mice for bispecific antibody efficacy studies.

Humanized PSMA/CD3 mice implanted with TRAMPC2_hPSMAv #1 were treated with 100 or 10 μg of the bispecific anti-P8MA/anti-SDR antibody BSPSMA/CD3-004 or isotype control 2x/week starting from the day of tumor implantation. Serum cytokine levels 4 hours after injection, as well as spleen T-cell levels, were also examined. The study was terminated on the 27th day.

As the results in Table. 14, the anti-P8MA/anti-COP bispecific molecule tested by BSPSMA/CD3-004 was effective in significantly slowing down tumor growth in all treatment groups. Minimal release of cytokines was observed after administration of BSPSMA/CD3-004, possibly due to weak binding of anti-SVZ antibodies. Both antibodies tested demonstrated antitumor efficacy without depleting T cells in the spleen.

Table 14

Efficacy of anti-P8MA/anti-COP bispecific antibodies in an immunocompetent syngeneic model

Mean serum cytokine concentrations, (pg/ml) The level of T cells of the spleen%, (mean+SD)* Tumor model/mouse strain Bispecific antibody identifier Dose (µg/mouse) 2x/week* N No. mouse/treatment group Tumor volume (mm ³ ) at day 27 (Mean ± SD) IFN g TNFa IL- 2 IL- 12p70 IL- 6 CD4+ CD8+ TRAMP-C2/ P S ^ ^um _C D3 ^h ™ BSPSMA/CD3- 004 100 4 50 ± 60 thirty 60 60 40 370 8.0±1.0 12.0±2.0 10 5 380±650 10 50 50 10 330 8.0±3.0 14.0±4.0 isotypic control 100 5 1740±560 4 thirty thirty 10 230 5.0±1.0 8.0 ± 2.0

* Mice were treated with antibody or isotype control 2x/week from the day of tumor implantation.

# Measured as the percentage of CD4 ⁺ or CD8 ⁺ cells in the spleen from living mCD45 ⁺ cells.

Thus, the bispecific anti-PSMA antibodies/SEZs of the present invention exhibit strong antitumor efficacy in both immunocompromised and immunocompetent tumor models despite having low or undetectable binding to the CD3 antigen.

Example 8 Anti-MiC16/anti-SDR bispecific antibodies show strong anti-tumor efficacy in vivo.

To determine the in vivo efficacy of exemplary anti-MiC16/anti-SDR bispecific antibodies identified as having weak or no detectable binding affinity for human and cynomolgus monkey POPs, studies were performed in immunocompromised mice bearing human prostate cancer xenografts. The efficacy of selected bispecific antibodies was tested both in dosing regimens and in therapeutic treatments.

Efficacy of anti-MiC16/anti-COP bispecific antibodies in human tumor xenograft models.

To evaluate the in vivo efficacy of anti-MuC16/anti-POP bispecific antibodies in human tumor xenograft studies, NOD SCID gamma (NSG) mice (Jackson Laboratories, Bar Harbor, Maine) were pre-implanted with human peripheral blood mononuclear cells (PBMC; ReachBio LLC). ., Seattle, WA) and then introduced ascitic cells from the human ovarian cancer cell line OVCAR-3 (American Type Culture Collection, Manassas, VA) were transduced with luciferase (OVCAR-3/Luc). OVCAR-3 cells endogenously express MUC-16.

Briefly, NSG mice were injected intraperitoneally (ip) with 5.0x10 ⁶ human PBMC. After 8 days, 1.5x10 ⁶ ascitic cells from the OVCAR-3/Luc cell line previously passaged in vivo were injected i.b. NSG mice inoculated with PBMC. In the immediate treatment group, mice were injected ip. in a day

-44040594 implantation of OVCAR-3/Luc aHTu-MUC16/aHmu-CD3, anti-BSMUC16/anti-CD3-001 or anti-BSMUC16/aHTu-CD3-005 cells bispecific antibodies or isotype control at 10 µg/mouse (N=5 mice /treatment group). In the therapeutic dose model, mice were administered ip. on day 7 after tumor implantation, the bispecific or control MUC16/CD3 antibodies described above at a dose of μg/mouse (N=5/treatment group).

In all studies, tumor growth was monitored using bioluminescent imaging (BLI). Mice were injected i.b. luciferase substrate D-luciferin suspended in PBS (150 mg/kg) and visualized under isoflurane anesthesia after 10 min. BLI was performed using the Xenogen IVIS system (Perkin Elmer, Hopkinton, MA) and BLI signals were generated using Living Image software (Xenogen/Perkin Elmer). Regions of interest were drawn around each cell mass and photon intensities were recorded as photons(p)/seconds(s)/cm ² /steradian(srad). For the immediate therapy group, data are presented as BLI levels at day 26 post-tumor implantation (Table 15). For the treatment group, data are presented as fold changes in BLI between day 6 (day 1 pre-treatment) and study endpoint (day 26 post-tumor implantation, Table 16).

As the results show, both BSMUC16/CD3-001 and BSMUC16/CD3-005 showed similar efficacy in tumor growth suppression compared to isotype control when BLI was measured on day 26 in the immediate dosing model. Both anti-MUC16/anti-CD3 bispecific antibodies also inhibited the growth of mature tumors on day 7 after tumor implantation compared to controls. Thus, the anti-MUC16/anti-CD3 bispecific antibodies of the present invention exhibit strong antitumor efficacy in several models.

Table 15

Efficacy of anti-MUC16/anti-CD3 bispecific antibodies in an immunocompromised xenograft model. __________________Immediate dosing__________________

Tumor Model/Mice Strain/Dose Bispecific antibody identifier N Mean bioluminescent radiation (photons/sec/sam ² /steradian) 26 days (Mean ± SD) OVCAR- Z/Luc/ NSG/10 µg/mouse BSMUC16/CD3- 001 5 1.4x10 ³ ± 3.5x10 ² BSMUC16/CD3- 005 5 1.5x10 ³ ± 9.7x10 ² isotypic control 5 2, OxiO ⁷ ± 1.0x10 ^e

- 45 040594

Table 16

Efficacy of aHTu-Muc16/aHmu-CD3 bispecific antibodies against in an immunocompromised xenograft model. Therapeutic treatment

Tumor Model/Mouse Strain/Dose Bispecific antibody identifier N Fold change in average bioluminescent radiation [p/s/cm ² /sr] at day 26 compared to day 6 (Mean ± SD) OVCAR-3/Luc/ NSG/10 µg/mouse BSMUC16/CD3-001 5 2.0±5.0 BSMUC16/CD3-005 5 0.01±0.02 isotypic control 5 21.0±8.0

Example 9 Pharmacokinetic evaluation of anti-MUC16 x anti-CD3 bispecific antibodies.

The pharmacokinetics of the anti-MUC16anti-CD3 bispecific antibodies BSMUC16/CD3-001 and BSMUC16/CD3-005 and isotype controls were evaluated in humanized MUC16xCD3 mice (mice homozygous for human MUC16 and CD3 expression, MUC16 ^hu/hu xCD3 ^{hu/ hu} ), humanized CD3 mice (mice homozygous for human CD3 expression, CD3 ^hu / ^hu ) and a wild-type control (75% C57BL, 25% 129Sv) mouse strain (WT). The cohort contained 4-5 mice per test antibody and per mouse strain. All mice received a single intraperitoneal (ip) dose of 0.4 mg/kg. Blood samples were collected at 3 and 6 hours, 1, 3, 7, 14 and 28 days after dosing. Blood was processed to serum and frozen at -80°C until analysis.

Circulating antibody concentrations were determined by total anti-human IgG antibody assay using GyroLab xPlore™ (Gyros, Uppsala, Sweden). Briefly, a biotinylated goat anti-human IgG polyclonal antibody (Jackson ImmunoResearch, West Grove, PA) was placed on streptavidin-coated beads on a Gyrolab Bioaffy 200 CD (Gyros) to capture human IgG present in serum. After capturing the affinity column, bound human IgG antibody in the samples was detected using Alexa-647 labeled with goat anti-human IgG (Jackson ImmunoResearch). The fluorescent signal on the column, necessary for the detection of bound IgG and response units (RU), was read by the instrument. Sample concentrations were determined by interpolation from a standard curve that was fitted using a 5-parameter logistic curve using the Gyrolab Evaluator software.

PK parameters were determined by non-compartmental analysis (NCA) using Phoenix®WinNonlin® version 6.3 software (Certara, LP, Princeton, NJ) and an extravascular dosing model. Using the appropriate mean concentration values for each antibody, all PK parameters including observed maximum serum concentration (C _max ), estimated half-life (t _1/2 ) and area under the concentration versus time curve to the last measured concentration (AUC _last ) were determined using the linear trapezoidal rule with linear interpolation and uniform weighting.

After v.b. administration of antibodies in TD mice, the overall time profiles of BSMUC16/CD3-001, BSMUC16/CD3-005 IgG concentration and isotype control were similar, characterized first by transient drug distribution followed by a single phase of drug elimination throughout the remainder of the study. The maximum serum concentration (C _max ) and estimated drug exposure (AUC _last ) from the three antibodies were comparable (1.3 times each other).

After v.b. administration of antibodies to mice CD3 ^hu/hu BSMUC16/CD3-001, BSMUC16/CD3-005 and isotype control had comparable concentrations of C _max (4.6, 3.6 and 4.1 µg/ml, respectively). BSMUC16/CD3-005 and the isotype control showed similar drug elimination curves, while BSMUC16/CD3-001 showed more rapid drug elimination than both, suggesting that binding to the human CD3 target induces clearance. The final antibody concentration for BSMUC16/CD3-001 was 0.03 µg/mL, which is about 28 times less than the final antibody concentrations determined for isotype control (0.85 µg/mL) and 22 times less than BSMUC16/CD3- 005 (0.66 µg/ml).

- 46 040594

In doubly humanized MUC16 ^hu/hu xCD3 ^hu/hu mice, the Muc16xCD3 bispecific antibody and the isotype control antibody had comparable C _max concentrations (C _max in the range of 4.5-6.9 μg/ml). Both bispecific antibodies exhibited higher drug removal than the isotype control, confirming a target-mediated effect. The final antibody concentration for BSMUC16/CD3-001 and BSMUC16/CD3-005 was about 29 times and 2.9 times less than the final antibody concentration determined for the isotype control (0.86 μg/ml).

A summary of the data for the total concentrations of anti-MUC16hanty-CD3 bispecific antibodies and isotype control antibodies is presented in table. 17. The average PK parameters are described in Table. 18A and 18B. Mean total antibody concentrations versus time are shown in FIG. 2A-2C. In conclusion, the MUC16xCD3 bispecific antibodies showed similar _Cmax and drug elimination curves in TD mice, but BSMUC16/CD3-001 showed higher elimination rates than BSMUC16/CD3-005 and isotype control in single humanized CD3 mice and mice with double humanization of MUC16/CD3. Since the bispecific antibodies administered in this PK study consist of the same anti-MUC16 binding arm, the results demonstrate that the binding strength of the CD3 targeting arm may play a role in drug exposure levels (AUC _last ) and drug elimination rates. Neither BSMUC16/CD3-001 nor BSMUC16/CD3-005 bind mouse MUC16 or mouse CD3.

Table 17

Mean serum total IgG concentrations after a single 0.4 mg/kg intraperitoneal injection of BSMUC16/CD3-001, BSMUC16/CD3-005 and isotype control antibodies in TD mice, humanized CD3 mice, and __________ humanized MUC16xCD3 mice ___________

Antibody Time (e) Total mAb concentration in mouse serum DT CD3hu ^/hu MUC16 ^hu/hu x CD3 ^hu/hu Mean (µg/mL) +/- SD Mean ee (mcg/ml) +/- SD Average e (mcg/m l) +/- SD 0.13 5, 39 0.34 4.30 0.29 6, 77 1.52 0.25 5, 80 0.36 4.26 1.07 6, 63 1, 06 BSMUC16 100 4.13 0.43 2.87 0.71 4.89 0.53 /CD3- 3.00 3, 19 0.53 1.44 0.27 2.50 0.22 001 7.00 2, 61 0.73 0.72 0.13 1.20 0.22 14.00 1.44 0.69 0.18 0.05 0.28 0.08 21.00 0.93 BUT 0.07 0.02 0.06 0.05

- 47 040594

28.00 0.60 BUT 0.04 0.01 0.03 0.02 BSMUC16 /CD3005 0.13 4.23 0.62 3.35 1.15 4.35 0.24 0.25 4.53 0.55 3.40 0.96 4.45 0.49 100 3.47 0.32 2.72 0.42 3.00 0.61 3.00 2.51 0.13 1.95 0.37 1.98 0.41 7.00 2.02 0.24 2.31 0.67 1.58 0.36 14.00 1, 19 0.17 1.01 0.23 0.78 0.26 21.00 1, 19 0.29 1.19 0.11 0.66 0.29 28.00 0.71 0.20 0.66 0.28 0.30 0.22 Isotype control 0.13 5, 07 1.16 5, 43 1.30 6.56 0.70 0.25 5, 91 1.10 5, 67 1.91 6.48 0.90 100 2, 64 0.24 2.98 1, 14 2.82 0.30 3.00 2.05 0.06 2.29 0.83 1.57 0.37 7.00 1.80 0.25 2.14 0.85 1.96 0.37 14.00 1.22 0.28 1.48 0.66 1.34 0.37 21.00 1.20 0.58 1.43 0.72 1.24 0.44 28.00 0.73 0.24 0.85 0.29 0.86 0.41

Time: (g, when noted) = time in hours after single injection;

D = study day;

SD = standard deviation;

NA = not determined due to the exclusion of mice with negative drug clearance titers.

Table 18A

Summary of pharmacokinetic parameters: CD3 ^hu / ^hu humanized mice

Pair- meter Edi- threads Mouse DT Mouse CD3 ^hu/hu Isotype control BSMUC16 / CD3-001 BSMUC1 6/ CD3005 Isotype control BSMUC16 /SOZOO! BSMUC16 /CD3005 With ^max mcg/m l 5 ± 3 6±0.4 5±0.5 4.1±3 4.6± 0.8 3.5±1 t _1/2 d 11 ± 4 7 ± 3 12±2 14±0.5 3.9± 0.6 11±5 AUC _last d»MKT / ml 35±18 40 ± 11 45±5 49 ± 20 16±3 36 ± 13

_Cmax = peak concentration;

AUC = area under the concentration-time curve;

AUClast = AUC calculated from the time until the last positive concentration;

T _1/2 = estimated half-life;

d = day.

- 48 040594

Table 18B

Summary of pharmacokinetic parameters: Double humanized MUC16 ^hu/hu xCD3 ^hu/hu ______ mice

Parameter Edi- threads Mouse DT MUC16 ^hu/hu x CD3 ^hu/hu mice Isotype control BSMUC16/CD3001 BSMUC16/CD3005 Isotype control BSMUC16/CD3001 BSMUC16/CD3005 With _tach mcg/ml 5 ± 3 6±0.4 5±0.5 6.7 ± 0.7 6.9 ± 1 4.5±4 t _1/2 d 11 ± 4 7 ± 3 12±2 12.9±4 3.3± 0.8 8.2±4 AUC _last s!*mcg/ml 35 ± 18 40 ± 11 45±5 46 ± 10 27 ± 3 34 ± 11

C _max = peak concentration;

AUC = area under the concentration-time curve;

AUClast =AUC calculated from the time until the last positive concentration;

T1 / 2 = estimated half-life; d = day.

Example 10 Anti-STEAP2/anti-CD3 bispecific antibodies show strong antitumor efficacy in vivo.

To determine the in vivo efficacy of exemplary anti-STEAP2/anti-CD3 bispecific antibodies identified as having weak or no detectable binding affinity for human and cynomolgus monkey CD3, studies were performed in immunocompromised mice bearing human prostate cancer xenografts.

To evaluate the in vivo efficacy of anti-STEAP2/anti-CD3 bispecific antibodies in human tumor xenograft studies in NOD SCID gamma (NSG) mice (Jackson Laboratories, Bar Harbor, Maine), human peripheral blood mononuclear cells (PBMC; ReachBio LLC.) were co-implanted. , Seattle, WA) together with C4-2 human prostate tumor cells (MD Anderson Cancer Center, Houston, TX) that endogenously express STEAP2.

Briefly, 5.0x106 C4-2 cells were co-implanted subcutaneously (SC) with 1.25x106 human PBMC in a 50:50 mixture of Matrigel (BD Biosciences, San Jose, CA) in the right flank of male NSG mice. Mice were treated intraperitoneally (ip) on the day of implantation (immediate treatment model) with anti-STEAP2/anti-CD3 bispecific antibodies BSSTEAP2/CD3-001, BSSTEAP2/CD3-002, or BSSTEAP2/CD3-003 or isotype control antibody ( that do not bind C4-2 tumor cells) at a dose of 0.1 or 0.01 mg/kg (N=5 mice/group).

Tumor size was measured twice a week using calipers and tumor volume calculated as volume=(length x width ² )/2. Data are presented as tumor size (mm ³ ) at the endpoint of the study, 46 days after tumor implantation (Table 19).

As the results shown in table. 19, BSSTEAP2/CD3-001, BSSTEAP2/CD3-002, and BSSTEAP2/CD3-003 significantly inhibited tumor growth compared to isotype controls when tumor sizes were measured at study endpoint. Importantly, the antiSTEAP2/anti-CD3 bispecific antibodies were effective in inhibiting C4-2 tumor growth even at the lowest dose of 0.1 mg/kg.

-

Claims (20)

Table 19 Efficacy of aHmu-STEAP2/aHTu-CD3 bispecific antibodies in an immunocompromised xenograft model. Immediate dosing Tumor model/ Mouse strain/ Bispecific antibody identifier Dose (mg/kg) N Tumor size (mm ³ ) 4 6 days after tumor implantation (Mean ± SD) С4-2 / NSG BSSTEAP2/CD3-001 0, 1 5 18.0±14.0 0.01 5 23.0±220 BSSTEAP2/CD3-002 oh 1 5 15.0±12.0 0.01 5 17.0±8.0 BSSTEAP2/CD3-003 oh 1 5 19.0 ± 12.0 0.01 5 25.0 ± 21.0 bispecific control oh 1 5 1020.0 ± 922.0 This invention should not be limited in scope to the specific embodiments described herein. Indeed, various modifications of the invention in addition to those described herein will become apparent to those skilled in the art from the foregoing description and accompanying figures. Such modifications are intended to be covered by the appended claims. CLAIM 1. A cytotoxic bispecific antibody comprising a first antigen-binding arm that exhibits weak to undetectable binding to human CD3 and cynomolgus CD3, and a second antigen-binding arm that binds a tumor-associated antigen, wherein said first arm contains a heavy chain variable region (HCVR) containing complementarity determining regions, HCDR1-HCDR2-HCDR3, containing, respectively, the amino acid sequences of SEQ ID NO: 36-38-40 or SEQ ID NO: 140-142-144, and a light chain, where the specified light chain contains a light chain variable region (LCVR ) containing complementarity determining regions LCDR1-LCDR2-LCDR3 containing, respectively, the amino acid sequences of SEQ ID NO: 164-166-168, or said light chain is a cognate light chain of the second antigen-binding arm that binds the tumor-associated antigen. 2. The bispecific antibody of claim 1, wherein said LCVR contains the amino acid sequence of SEQ ID NO: 162. 3. A bispecific antibody according to claim 1 or 2, wherein said first antigen-binding arm comprises complementarity determining regions HCDR1-HCDR2-HCDR3-LCDR1-LCDR2-LCDR3 containing, respectively, the amino acid sequences of SEQ ID NO: 36-38-40-164- 166-168. 4. A bispecific antibody according to claim 1 or 2, wherein said first antigen-binding arm comprises complementarity determining regions HCDR1-HCDR2-HCDR3-LCDR1-LCDR2-LCDR3, respectively containing the amino acid sequences of SEQ ID NO: 140-142-144-164- 166-168. 5. The bispecific antibody of claim 2, wherein said bispecific antibody comprises a pair of amino acid sequences HCVR and LCVR (HCVR/LCVR) comprising the amino acid sequences of SEQ ID NO: 34/162. 6. The bispecific antibody of claim 2, wherein said bispecific antibody comprises a pair of amino acid sequences HCVR and LCVR (HCVR/LCVR) comprising the amino acid sequences of SEQ ID NO: 138/162. 7. The use of a cytotoxic bispecific antibody for the treatment of cancer in a subject, wherein said bispecific antibody comprises a first antigen-binding arm that exhibits weak to undetectable binding to human CD3 and cynomolgus CD3, and a second antigen-binding arm that binds a tumor-associated antigen, wherein said first arm contains a heavy chain variable region (HCVR) containing complementarity determining regions HCDR1-HCDR2-HCDR3 containing, respectively, the amino acid sequences of SEQ ID NO: 36-38-40 or SEQ ID NO: 140-142-144, and a light chain, where said light chain contains - 50 040594 light chain variable region (LCVR) containing complementarity determining regions, LCDR1-LCDR2-LCDR3 containing respectively amino acid sequences SEQ ID NO: 164-166-168 or said light chain is a cognate light chain of a second antigen-binding arm that binds a tumor-associated antigen. 8. Use according to claim 7, wherein said bispecific antibody exhibits T cell activation in vitro. 9. Use according to claim 7 or 8, wherein said tumor-associated antigen is expressed on a human tumor cell. 10. Use according to any one of claims 7-9, wherein said bispecific antibody induces T cell mediated tumor cell lysis with an EC ₅₀ value of less than about 1.3 nM as measured in the T cell mediated tumor cell lysis assay in vitro. 11. Use according to any one of claims 7-10, wherein said tumor-associated antigen is selected from the group consisting of AFP, ALK, BAGE proteins, BIRC5 (survivin), BIRC7, β-catenin, brc-abl, BRCA1, BORIS, CA9, carbonic anhydrase IX, caspase-8, CALR, CCR5, CD19, CD20 (MS4A1), CD22, CD30, CD40, CDK4, CEA, CTLA4, cyclin-B1, CYP1B1, EGFR, EGFRvIII, ErbB2/Her2, ErbB3, ErbB4, ETV6- AML, EpCAM, EphA2, Fra-1, FOLR1, GAGE proteins (e.g. GAGE-1, -2), GD2, GD3, GloboH, glypican-3, GM3, gp100, Her2, HLA/B-raf, HLA/k -ras, HLA/MAGE-A3, hTERT, LMP2, MAGE proteins (e.g. MAGE-1, -2, -3, -4, -6 and -12), MART-1, mesothelin, ML-IAP, Muc1, Muc2, Muc3, Muc4, Muc5, Muc16 (CA-125), MUM1, NA17, NY-BR1, NY-BR62, NY-BR85, NY-ESO1, OX40, p15, p53, PAP, PAX3, PAX5, PCTA-1 , PLAC1, PRLR, PRAME, PSMA (FOLH1), RAGE proteins, Ras, RGS5, Rho, SART-1, SART-3, STEAP1, STEAP2, TAG-72, TGF-β, TMPRSS2, Thompson-nouvelle antigen (Tn) , TRP-1, TRP-2, tyrosinase and uroplakin-3. 12. Use according to claim 11, wherein the tumor associated antigen is CD20, EGFRvIII, PSMA (FOLH1), STEAP2, or MUC16. 13. Use according to any one of claims 7-12, wherein said LCVR contains the amino acid sequence of SEQ ID NO: 162. 14. Use according to any one of claims 7-13, wherein said first antigen-binding arm comprises complementarity determining regions HCDR1-HCDR2-HCDR3-LCDR1-LCDR2-LCDR3 containing, respectively, the amino acid sequences of SEQ ID NO: 36-38-40-164 -166-168. 15. Use according to any one of claims 7-13, wherein said first antigen-binding arm comprises complementarity determining regions HCDR1-HCDR2-HCDR3-LCDR1-LCDR2-LCDR3 containing, respectively, the amino acid sequences of SEQ ID NO: 140-142-144-164 -166-168. 16. Use according to claim 13, wherein said bispecific antibody comprises an HCVR and LCVR amino acid sequence pair (HCVR/LCVR) comprising the amino acid sequences of SEQ ID NO: 34/162. 17. Use according to claim 13, wherein said bispecific antibody comprises an HCVR and LCVR amino acid sequence pair (HCVR/LCVR) comprising the amino acid sequences of SEQ ID NO: 138/162. 18. A pharmaceutical composition for the treatment of cancer in a subject, comprising a cytotoxic bispecific antibody according to any one of claims 1 to 6 and a pharmaceutically acceptable carrier or diluent. 19. A method of treating cancer in a subject, comprising administering to the subject a pharmaceutical composition according to claim 18. 20. The method of claim 19 wherein the cancer is selected from the group consisting of pancreatic cancer, melanoma, glioblastoma, head and neck cancer, prostate cancer, malignant gliomas, osteosarcoma, colorectal cancer, gastric cancer, malignant mesothelioma, multiple myeloma, cancer ovarian cancer, small cell lung cancer, non-small cell lung cancer, synovial sarcoma, thyroid cancer, breast cancer, melanomaglioma, breast cancer, squamous cell carcinoma, esophageal cancer, clear cell renal cell carcinoma, chromophobe renal cell carcinoma, renal oncocytoma, renal transient cell carcinoma carcinoma, urothelial carcinoma, adenocarcinoma, small cell carcinoma, or primary or metastatic tumors of the immune system. -