WO2020006544A1

WO2020006544A1 - Humanized antibodies against human tim-3 and uses thereof

Info

Publication number: WO2020006544A1
Application number: PCT/US2019/040029
Authority: WO
Inventors: Yu-Chen Yang; Chia-Hua Li; Shu-Hui Lin; Ai-Hsuan TSAI; I-Jung WU; Hsien-Yu TSAI; Szu-Liang LAI; Mu-Shung LO; Yu-Hsun LO; Chien-Tsun Kuan
Original assignee: Development Center For Biotechnology; Dcb-Usa Llc
Priority date: 2018-06-29
Filing date: 2019-06-29
Publication date: 2020-01-02
Also published as: TW202016145A

Abstract

A humanized anti-human T-cell immunoglobulin domain and mucin domain 3 (TIM-3) antibody, or a binding fragment thereof, wherein the antibody or the binding fragment includes a heavy-chain variable domain that comprises framework region sequences derived from a human immunoglobulin and the following complementarity determining region (CDR) sequences: HCDRI (SEQ ID NO: 3), HCDR2 (SEQ ID NO: 4), HCDR3 (SEQ ID NO: 5), and a light-chain variable domain that includes framework region sequences from a human immunoglobulin and the following CDR sequences: LCDR1 (SEQ ID NO: 6), LCDR2 (SEQ ID NO: 7), and LCDR3 (SEQ ID NO: 8).

Description

HUMANIZED ANTIBODIES AGAINST HUMAN TIM-3 AND USES THEREOF

BACKGROUND OF INVENTION

Field of the Invention

[0001] The present invention relates to methods for generation and use of humanized antibodies that bind specifically to the human T-cell immunoglobulin domain and mucin domain 3 (TIM-3).

Background Art

[0002] Immune responses play important roles in staving off cancers. Therefore, T cell exhaustion may be associated with tumor growth in hosts. T cell exhaustion may arise from many mechanisms. A programmed cell death molecule (PD-l) is a marker of the exhausted T cells. Blockade of PD-l interactions with its ligand (PD-l ligand, or PD1L, or PD-L1) can partially restore T cell functions.

[0003] In addition to PD-l, T cell immunoglobulin mucin 3 (TIM-3) expression was found on CD8⁺ tumor-infiltrating lymphocytes in mice bearing solid tumors. TIM-3 was originally found to be a T helper (Th) 1 -specific type I membrane protein. TIM-3, an immune checkpoint, regulates macrophage activation and plays a vital role in Thl immunity and tolerance induction.

[0004] All TIM-3 expressing tumor-infiltrating lymphocytes are also found to express

PD-l. These TIM-3 and PD-l expressing lymphocytes account for a major fraction of T cells that infiltrate tumors. In addition, these TIM-3 and PD-l expressing lymphocytes also exhibit the most severe exhausted phenotype: they fail to proliferate and also fail to produce IL-2, TNF, and IFN-g. (Sakuishi et al, J. Exp. Med., 2010, 207(10): 2187-2194).

[0005] More recently, TIM-3 has also been found to play a role in the regulation of other cells, such as Thl7 cells, CD4(+) CD25(+) regulatory T cells (T_regs), CD8(+) T cells and certain innate immune cells.

[0006] Because the TIM-3 pathway is involved in the pathogenesis of autoimmune diseases, chronic viral infections, and cancers, there is a need to find agents that can inhibit or block the TIM-3 signaling pathway.

SUMMARY OF INVENTION

[0007] Embodiments of the invention relate to humanized antibodies that can bind specifically with TIM-3, thereby inhibiting the functions of TIM-3. TIM-3 signaling pathway is found to be associated with T cell exhaustion. Therefore, antibodies of the invention can be used to prevent or treat diseases or conditions associated with T cell exhaustion, such as autoimmune diseases and cancers.

[0008] One aspect of the invention relates to humanized antibodies against human

T-cell immunoglobulin domain and mucin domain 3 (TIM-3), or a binding fragment (e.g., scFv, Fab, or (Fab)₂) thereof. A humanized anti-human T-cell immunoglobulin domain and mucin domain 3 (TIM-3) antibody, or a binding fragment thereof, in accordance with one embodiment of the invention, may comprise a heavy-chain variable domain that comprises framework region sequences derived from a human immunoglobulin and the following complementarity determining region (CDR) sequences: HCDRI (SEQ ID NO: 3), HCDR2 (SEQ ID NO: 4), HCDR3 (SEQ ID NO: 5), and a light-chain variable domain that comprises framework region sequences from a human immunoglobulin and the following CDR sequences: LCDR1 (SEQ ID NO: 6), LCDR2 (SEQ ID NO: 7), and LCDR3 (SEQ ID NO: 8). The framework region sequences in the heavy-chain variable domain may be derived from corresponding framework region sequences of IGHVl-2*02, and the framework region sequences in the light-chain variable domain may be derived from corresponding framework region sequences of IGVK4-l *0l .

[0009] An anti-human T-cell immunoglobulin domain and mucin domain 3 (TIM-3) antibody, or a binding fragment thereof, in accordance with embodiments of the invention, may comprise a heavy-chain variable domain having the sequence of SEQ ID NO: 10, or 20, or 2, 1 and a light-chain variable domain having the sequence of SEQ ID NO: 9 or 22.

[0010] In accordance with embodiments of the invention, an antibody, or a binding fragment thereof, of the invention is useful in diagnosis, prognosis, and treatment of cancers that express cell- surface TIM-3. These cancers, for example, may include lung, liver, esophageal cancer, and solid tumors.

[0011] Other aspects and advantages of the invention will be apparent from the following description and the appended claims.

BRIEF DESCRIPTION OF DRAWINGS

[0012] FIG. 1 shows the sequences of the heavy chain and light chain variable regions of a monoclonal anti-human TIM-3 antibody, 8C11, in accordance with one embodiment of the invention. The complementarity determining region (CDR) sequences are shown as bold-faced and underlined sequences. The framework sequences are dispersed between and flanking the CDR sequences.

[0013] FIG. 2 shows the full-length (including the signal peptide shown as underlined) sequences of the heavy chain (SEQ ID NO: 14) and light chain (SEQ ID NO: 13) of a humanized anti-human TIM-3 antibody, Hu8Cl l, in accordance with one embodiment of the invention. The light-chain variable domain (VL; SEQ ID NO: 9) and the heavy-chain variable domain (VH; SEQ ID NO: 10) are shown in bold types in the respective sequences and the complementarity determining region (CDR) sequences in the variable domains, which are identical to those shown in FIG. 1, are shown in boxes.

[0014] FIG. 3A shows nucleotide and protein sequences for the heavy-chain viable domain for Hu8Cl 1 (nucleotide: SEQ ID NO: 15; protein: SEQ ID NO: 10), and FIG. 3D shows nucleotide and protein sequences for the light-chain viable domain for Hu8Cl l (nucleotide: SEQ ID NO: 18; protein: SEQ ID NO:9). Three back mutated mutant sequences are also shown: FIG. 3B shows Hu8Cl l-HBl (nucleotide: SEQ ID NO: 16; protein: SEQ ID NO:20), FIG. 3C shows Hu8Cl l-HB2 (nucleotide: SEQ ID NO: 17; protein: SEQ ID NO:2l), and FIG. 3E shows Hu8Cl l-LB3 (nucleotide: SEQ ID NO: 19; protein: SEQ ID NO:22). [0015] FIG. 4 shows that the 8C11 mAh and a control mAh 2E2 are specific for Tim3 antigen and will not bind Timl or Tim4.

[0016] FIG. 5 shows results of FACS analysis using activated T cells. The results show that the mouse mAh 8C11, humanized mAh Hu8Cl 1, and the control mAh 2E2 all bind tightly to the activated T cells.

[0017] FIG. 6A shows a standard curve for dot-blot analysis. FIG. 6B shows dot blot analysis of the heavy-chain alanine mutants, and FIG. 6C shows the dot blot analysis of the light-chain alanine mutants.

[0018] FIG. 7 shows representative results from FACS analysis of heavy-chain alanine mutants binding to human Tim3. These results, together with similar results for other alanine mutants, show that the bindings of G26A, R99A, S102A, and W105A mutants were significantly compromised, indicating that these residues are critical for binding.

[0019] FIG. 8 shows representative results from FACS analysis of light-chain alanine mutants binding to human Tim3. These results, together with similar results for other alanine mutants, show that the bindings of S26A, V29A, S30A, T31A, H91A, N93A, and W96A mutants were significantly compromised, indicating that these residues are critical for binding.

[0020] FIG. 9 shows 8C11 mAh cannot bind monkey TIM-3 extracellular domain

(ECD), while the control antibody 2E2 can. These results indicate that 8C11 mAh is specific for human TIM-3, but not for monkey TIM-3.

[0021] FIG. 10 shows that VH-S101A and VH-S102A mutants of Hu8Cl l in the heavy-chain CDRs change the properties of the antibody such that the mutant antibodies can bind monkey TIM-3.

[0022] FIG. 11A shows a treatment schedule for testing the effects of Hu8Cl l in a xenograft lung cancer mouse model.

[0023] FIG. 11B shows the effect of cancer cell growth inhibition using the xenograft model by Hu8Cl l, cyclophosphamide, and a combination therapy using Hu8Cl l and cyclophosphamide. The results show that the combination therapy has a synergistic effect.

DEFINITIONS

[0024] As used herein, "complementarity-determining region (CDR)" refers to the three hypervariable sequences in the variable region that are involved in antigen recognition. The three CDRs are dispersed by four "framework" regions in the light or heavy chain variable region. The CDRs are primarily responsible for binding to an epitope of an antigen. The CDRs of each chain are typically referred to as CDR1, CDR2, and CDR3, numbered sequentially starting from the N-terminus, and are also typically identified by the chain in which the particular CDR is located. Thus, a VH CDR3 is located in the variable domain of the heavy chain of the antibody in which it is found, whereas a VL CDR1 is the CDR1 from the variable domain of the light chain of the antibody in which it is found. These may be abbreviated as HCDR1, HCDR2, HCDR3, LCDR1, LCDR2, and LCDR3, respectively, wherein H denotes the heavy chain and L denotes the light chain. [0025] The sequences of the framework regions (FR) of different light or heavy chains are relatively conserved within a species. The framework regions of an antibody (i.e., the combined framework regions of the constituent light and heavy chains) serve as a scaffold to position and align the CDRs in the three-dimensional space. The four FRs that flank the 3 CDRs may be referred to as FR1, FR2, FR3, and FR4. Antibodies of the invention may comprise framework region sequences“derived from” a human immunoglobulin. As used herein, the term“derived from” includes framework sequences that are identical or have at least 90% identity to the sequences of original framework sequences of a human antibody. In some embodiments, antibodies of the invention may comprise framework region sequences “selected from” a human immunoglobulin. As used herein, the term“selected from” includes framework sequences that are identical to the sequences of original framework sequences of a human antibody.

[0026] The amino acid sequences of the CDRs and framework regions (FRs) can be determined using various well-known definitions in the art, e.g., Rabat, international ImMunoGeneTics database (IMGT). The CDR regions described herein are based on Rabat definitions.

DETAILED DESCRIPTION

[0027] Embodiments of the invention relate to humanized antibodies that specifically bind human TIM-3. These humanized anti-human TIM-3 antibody can be used to treat, and/or diagnose immune or cancerous diseases.

[0028] Immune responses play important roles in staving off cancers. However, in chronic viral infections and cancers, it has been found that T cell exhaustion is associated with these diseases. T cell exhaustion may arise from many mechanisms. A programmed cell death molecule (PD-l) is a marker of the exhausted T cells. Blockade of PD-l interactions with its ligand (PD-l ligand, or PD1L, or PD-L1) can partially restore T cell functions, suggesting that blockade of PD-l and PD-L1 interactions can be an effective way to restore immune functions, thereby treating or preventing diseases associated with PD-l and/or PD-L1 mediated immune suppression or exhaustion..

[0029] In addition to PD-l, T cell immunoglobulin mucin 3 (TIM-3), which is an immune checkpoint marker, is also found to be involved in T cell exhaustion. TIM-3 was originally found to be a T helper (Th) 1 -specific type I membrane protein. TIM-3, an immune checkpoint, regulates macrophage activation and plays a vital role in Thl immunity and tolerance induction.

[0030] TIM-3 expression was also found on CD8⁺ tumor-infiltrating lymphocytes in mice bearing solid tumors. All TIM-3 expressing tumor-infiltrating lymphocytes are also found to express PD-l. These TIM-3 and PD-l expressing lymphocytes account for a major fraction of the T cells that infiltrate tumors. In addition, these TIM-3 and PD-l expressing lymphocytes also exhibit the most severe exhausted phenotype: they fail to proliferate and also fail to produce IL-2, TNF, and IFN-g. (Sakuishi et al, J. Exp. Med., 2010, 207(10): 2187-2194). TIM-3 pathway may interact with PD-l pathway in the dysfunctional CD8⁺ T cells and Tregs in cancers. Following PD-l inhibition, TIM-3 is mainly expressed on activated CD8⁺ T cells and suppresses macrophage activation. Upregulation of TIM-3 was observed in tumors progressing after anti-PD-1 therapy. (Koyarna S. et al. (February 2016), “Adaptive resistance to therapeutic PD-1 blockade is associated with upregulation of alternative immune checkpoints,” Nature Communications , 7: 10501, doi:

10.1038/ncommsl0501). This seems to be a form of adaptive resistance to immunotherapy.

[0031] More recently, TIM-3 has been found to play a role in the regulation of other cells, such as Thl7 cells, CD4(+) CD25(+) regulatory T cells (T_regs), CD8(+) T cells and certain innate immune cells. Because the TIM-3 pathway is involved in the pathogenesis of autoimmune diseases, chronic viral infections, and cancers, antibodies of the invention can be used in the treatments of these disease.

[0032] In accordance with embodiments of the invention, several clones of antibodies against human TIM-3 were generated. Briefly, BALB/c mice were primed with purified recombinant human TIM-3 antigen (as a fusion protein with a 6xHis tag). The splenocytes were harvested and then cultured after fusing with Fo cells. The hybridoma cells secreting monoclonal antibodies that can recognize TIM-3 antigen were selected by a TIM-3 antigen-based ELISA. The selected clones were verified by FACS assays. Many clones were isolated and analyzed. The variable domain sequences were determined.

[0033] One exemplary antibody, 8C11, was further investigated. The epitopes on the extracellular domain of TIM-3 for 8C11 binding are found to be in the regions of RKGDVSL (SEQ ID NO: 11) and/or EKFNLKL (SEQ ID NO: 12). It was found that 8C11 can bind with TIM-3 to interfere with the interactions between TIM-3 and its ligand, Galectin-9. In addition, 8C11 binding to TIM-3 does not induce T cell death (apoptosis), i.e., binding of the antibody to TIM-3 does not trigger the TIM-3 signaling pathway. Therefore, there is no risk of inducing T cell exhaustion by an antibody of the invention. This is important because to be useful as a therapeutic agent, an antibody that blocks the interaction between TIM-3 and its ligand (galectin-9) should not itself activates TIM-3 signaling pathway, leading to T cell exhaustion/apoptosis.

[0034] In addition, antibodies of the invention can enhance IFN-g and TNF-a sections by T cell, thereby enhancing immune responses. These antibodies are able to suppress tumor growths in animal models. Therefore, antibodies of the invention can be used to treat cancers, such as lung, breast, pancreas, liver, colorectal, or prostate cancer.

[0035] While the above described antibodies are useful, they are of mouse origin.

These mouse antibodies may have complications (e.g., adverse immune responses) when used on humans. In addition, these mouse antibodies will not induce antibody-dependent cellular toxicity (ADCC) or complement-dependent cytotoxicity (CDC) reactions, which help to eliminate cancer cells or infected cells. Thus, to further develop these antibodies for clinical use, these antibodies have been humanized. Furthermore, the humanized antibodies have been optimized by back mutations and mutations in the CDR sequences.

[0036] The humanized antibodies have been analyzed for their bindings to various Tim-family members and Tim-3 from different species. It was found that humanized antibodies of the invention are specific for TIM-3 and would not bind other TIM family members (e.g., TIM-1 and TIM-4) and that certain antibodies (e.g., hu8Cl l) are species specific (e.g., binds only human TIM-3), while other variants can be made to also bind TIM-3 from other species (e.g., monkey). In addition, the humanized antibodies have been tested for their abilities to inhibit tumor growth in xenograft system. It was found that antibodies of the invention are effective in inhibiting tumor growths in vivo and combination therapy with other chemotherapeutic agents (e.g., cyclophosphamide) can provide synergistic effects. The following will describe these with specific examples. One skilled in the art would appreciate that these examples are for illustration only and are not meant to limit the scope of the invention.

Cloning of 8C11 monoclonal antibody

[0037] Cloning of the gene encoding the 8C11 mAh is performed in accordance with the methods described below. While the following procedures may set out specific conditions and parameters, one skilled in the art would appreciate that this is only an example for illustrating embodiments of the invention and that other modifications and variations are possible without departing from the scope of the invention.

cDNA cloning of antibody genes and preparation

[0038] The hybridoma was cultured in a defined medium. When the cell number reached about lOxlO⁶ cells/ml, the cells were harvested by centrifugation, and then TRIzol kit was added to extract total RNA in accordance the instruction manual (Thermo Fisher Scientific, Waltham, MA, USA). Cloning the variable regions of the antibody cDNAs was performed using a mouse Ig-primer set in accordance with the instruction manual from the supplier. The first strand cDNA was prepared using 5 micrograms of total RNA as a template, 50 ng/ml of random primers, and 10 mM dNTP, which were mixed in diethylpyrocarbonate (DEPC)-treated water in a PCR tube.

[0039] The reaction mixture was incubated at 65°C for 5 min, and then placed on ice.

Ten (10) mΐ cDNA synthesis mixture containing 2 mΐ of lOx RT buffer, 4 mΐ of 25 mM MgCh, 2 mΐ of DTT, 1 mΐ of 4 units of RNaseOUT (Thermo Fisher Scientific), and 1 mΐ of 200 units of Superscript III RT, was mixed gently and collected by brief centrifugation. The reaction tube was incubated for 10 min at 25°C, followed by 50 min at 50°C. The reaction was terminated by heating at 85°C for 5 min, and then the tube was chilled on ice. The tube was briefly centrifuged to collect the reaction product, and 1 mΐ of RNase H was added, and the mixture was incubated for 20 min at 37°C.

[0040] A reaction mixture having a composition of 5 mΐ of cDNA, 5 mΐ of lOx reaction buffer, 1 mΐ of lOmM dNTP mix, Imΐ of 2.5 unit Taq polymerase, and 1 mΐ of forward and reverse primers was prepared in a final volume of 50 mΐ with double distilled water and subjected to PCR. For amplification of the light or and heavy chain of an antibody, 94°C as the first step, and then a cycle of 94°C for 30 sec, 55°C for 30 sec, and 72°C for 1 min was repeated 30 times. After the reaction cycles, the final step was 72°C for 10 min. The reaction mixture was analyzed by 2% agarose gel. Products with the predicted molecular weights were ligated into a cloning vector, and then used to determine the nucleotide sequences.

[0041] Based on the sequence information, antibody sequences were translated into protein sequences using the ExPASY-Translation tool (ExPASY Bioinformatics Resource Portal). Resulting sequences of anti-human TIM-3 antibody 8C11 comprise a heavy chain amino acid sequence and a light chain sequence. The complementarity determining regions (CDR) in these sequences were determined by the method of Rabat et al,“Sequences of proteins of immunological interest,” Fifth Edition, NIH Publication 91-3242, Bethesda Md. (1991), vols. 1-3.

[0042] As shown in FIG. 1, the amino acid sequences of 8C11 antibody contains a heavy-chain variable region (SEQ ID NO: l) and a light-chain variable region (SEQ ID NO:2). The framework regions (FR1, FR2, FR3, and FR4) and CDRs (HCDR1 SEQ ID NO:3, HCDR2 SEQ ID NO:4, HCDR3 SEQ ID NO:5, LCDR1 SEQ ID NO:6, LCDR2 SEQ ID NO: 7, and LCDR3 SEQ ID NO: 8) are indicated.

Humanization of anti-TIM3 antibody 8C11 mAb

[0043] To make anti-TIM3 antibodies useful in the clinics, 8C11 was selected for humanization. Humanization starts by selecting human antibody sequences that are highly homologous to the 8C11 sequence. The homology search may use the entire sequences of 8C11, only the variable domain sequences of 8C11, or only the framework sequences of 8C11. As an example, the VH and VL sequences of mAb 8C11, shown below, may be used to search the IMGT database for human antibody sequences with the highest homologies.

8C11 VH (CDR sequences are shown in boxes: HCDR1, HCDR2, and HCDR3):

EVQLQQSGPELVKPGASVKMSCKAS|GYTFTDYYMN|WVKQSHGKSLEWIG|R|

|K ATLT VDKSLNT AYMQLN SLTSED S AVYY C AR|RD|

V S A (SEQ ID NO: 1)

8C11 VL (CDR sequences are shown in boxes: LCDR!, LCDR2, and LCDR3):

DIVMTQSHKFMSTSVGDRVSITCKASQDVSTAVVWYQQKPGQSPKILIFSPS

YRYT|GVPDRFTGSGSGTEFTFTIS S VQAEDLAVYY C|QQHYNIPWT[FGGGTK LEIRR (SEQ ID NO: 2)

[0044] Alternatively, one may use only the framework sequences of 8C11 (i.e., without the CDR sequences) in the homology search of the IMGT database.

[0045] Based on these searches, IGHVl-2*02 (IMGT accession No. X62106) and

IGVK4-l*0l (IMGT accession No. Z00023) are selected as the most homologous human sequences for the heavy-chain variable domain and light-chain variable domain, respectively. The humanization of 8C11 is then based on the framework sequences from these two sequences.

[0046] Then, the CDR sequences from 8C11 are used to replace the corresponding

CDR sequences in the human antibody sequences. The CDR regions in the human sequence can be derived from analysis using methods known in the art, such as the Rabat method. The grafting of the mouse CDR sequences into the human antibody sequence can be accomplished using PCR or similar techniques.

[0047] FIG. 2 shows the full-length light chain (VL + VC; SEQ ID NO: 13) and full-length heavy chain (VH + CH1 + CH2 + CH3; SEQ ID NO: 14) of the humanized antibody, Hu8Cl l, wherein the light-chain variable domain (VL) sequence (SEQ ID NO:9) and heavy-chain variable domain (VH) sequence (SEQ ID NO: 10) are shown in bold types.

[0048] Next, the humanized anti-Tim3 antibody (i.e., Hu8Cl l) was sub-cloned into antibody expression vectors. Any suitable vectors known in the art may be used, such as TGEX vectors from Antibody Design Labs (San Diego, CA). The expression vectors were then used to transfect cells, such as FreeStyle 293-F cells (Thermo Fisher Scientific), to express the antibodies.

3) Back mutations to improve bindings.

[0049] Grafting of mouse CDR sequences onto human frameworks/antibodies results in variable domains (VH and VL) containing sequences from two different sources (i.e., mouse CDR sequences + human framework sequences). Such heterologous domains may not have the optimal sequences and may have structural perturbation that may impact antibody bindings. Therefore, affinities of the antibodies may not be the best. To improve the binding affinities, some amino acids in the variable domains may be mutated back to the mouse residues.

[0050] In one approach to selecting amino acids for back mutations, critical amino acid residues that may impact antibody bindings may be analyzed by computer modeling. In the modeling, one may also apply prior knowledge learned from antibody engineering based on successful cases. Based on the modeling, selective amino acid substitutions may be performed, for example, to replace the amino acid residues in Hu8Cl l with the corresponding amino acid residues in the original murine 8C11 (i.e., back mutations). Then, the mutant antibodies may be cloned, expressed, and assessed for their bindings to select for improved antibodies.

[0051] FIGs. 3 A-3E show the nucleotide and protein sequences of the heavy-chain and light-chain regions for Hu8Cl l and several back-mutation variants of Hu8Cl l. FIG. 3 A shows Hu8Cl l-H sequences (nucleotide: SEQ ID NO: 15; Protein: 10) of the heavy-chain variable region sequences for the initial humanized mAh Hu8Cl l. FIG. 3D shows Hu8Cl l-L sequences (nucleotide: SEQ ID NO: 18; protein:9) of the light-chain region sequence for the initial humanized mAh Hu8Cl l. FIGs. 3B and 3C show two improved heavy-chain variants (Hu8Cl l-HBl; nucleotide: SEQ ID NO: 16; protein: SEQ ID NO:20) and (Hu8Cl l-HB2; nucleotide: SEQ ID NO: 17; protein: SEQ ID NO:2l) are also shown, as well as one improved light-chain variant (Hu8Cl 1-LB3; nucleotide: SEQ ID NO: 19; protein: SEQ ID NO:22).

[0052] The Hu8Cl l-HBl (SEQ ID NO:20) variant contains back mutations (replacing with the corresponding residues in Hu8Cl l with the original residues in the murine 8C11 mAh) at amino-acid residues 38, 48, 66, 67, 71, 73, and 76. The Hu8Cl l-HB2 (SEQ ID NO:2l) variant contains back mutations at amino-acid residues 69, 71, 73, and 76. The Hu8Cl l-LB3 (SEQ ID NO:22) variant contains back mutations at amino-acid residues 46 and 49. [0053] Antibodies containing different combinations of variant heavy chains and light chains are produced in 293 cells, purified, and tested for their bindings to human-Tim3. The combinations include Hu8Cl l-BlB3 (containing Hu8Cl l-HBl heavy chain and Hu8Cl l-LB3 light chain), Hu8Cl l-B2B3 (containing Hu8Cl l-HB2 heavy chain and Hu8Cl l-LB3 light chain), and Hu8Cl 1-HB3 (containing the original Hu8Cl 1 heavy chain and Hu8Cl 1-LB3 light chain). The binding assays may be performed using any methods known in the art, such as ELISA, BIAcore, or FACS analysis. For example, Table 1 shows the results ofBIAcore assays for these combinational variants, as well as the original murine mAb (801 -MM) and the initial humanized mAb (Hu8Cl l-HH).

TABLE 1

[0054] As shown in Table 1, the initial humanized mAb (Hu8Cl l-HH) has a binding affinity slightly worse than that of the original murine mAb (8C11-MM). The various combinations of the back-mutated variants have slight improvements in the binding affinities. For example, the Hu8Cl l-BlB3 mutant has an affinity that is essentially the same as the murine mAb (8C1 l-MM). The fact that these back-mutation variants retain the biding avidities suggests that the humanized anti-human TIM-3 antibodies can accommodate certain degree of mutations in the framework sequences. Thus, antibodies of the invention may comprise framework region sequences“derived from” a human immunoglobulin. As used herein, the term“derived from” includes framework sequences that are identical, or have at least 90% identity, to the sequences of original framework sequences of a human antibody.

[0055] To test the specificity of the humanized antibody, we tested the binding of

Hu8Cl l to different Tim molecules (i.e., Timl, Tim3, and Tim4). The functions of different Tim molecules in immune regulations are complex, and different Tim molecules may have different effects. Therefore, in order to use anti-Tim3 antibodies for reversing T cell exhaustion, it would be desirable that the antibodies are specific for Tim3 and would not act on Timl or Tim4. As shown in FIG. 4, both Hu8Cl l and a control antibody (Anti-Tim3 Antibody, clone 2E2, which is available from commercial source, such as Millipore Sigma) bind specifically to Tim3 only and do not bind Timl or Tim4. This result indicates that Hu8Cl 1 or its variants would be safe to use to reverse T cell exhaustion by interfering with the interactions between Tim3 and its ligand (galectin-9). [0056] In addition to the BIAcore binding assays, the antibodies have also been assayed using FACS analysis. Briefly, human T cells were first activated with anti-CD3 and anti-CD28 coated magnetic beads, which are commercially available (such as Dynabeads® Human T-Activator CD3/CD28 from Thermo Fisher Scientific). The anti-CD3 and anti-CD28 antibodies provide primary and co-stimulatory signals, optimized for efficient T cell activation and expansion.

[0057] T cells activation using the magnetic beads coated with anti-CD3/anti-CD28 were performed following the commercial provider’s procedures. For example, using Dynabeads® Human T-Activator CD3/CD28 from Thermo Fisher Scientific, the following are exemplary procedures:

[0058] For example, start with 8 x 10⁴ purified T cells in 100-200 pL medium in a

96-well tissue culture plate. Add 2 pL pre-washed and resuspended Dynabeads® to obtain a bead-to-cell ratio of 1 : 1. Incubate in a humidified C0₂ incubator at 37°C. Harvest the activated T cells and remove magnetic beads before FACS analysis.

[0059] After activation, the T cells were incubated with the test antibodies, including the commercially available 2E2 antibody as a positive control. Bindings of the antibodies with T cells were allowed to proceed in a humidified CO2 incubator at 37°C for 1 hour. Then, a fluorescence-labeled secondary antibody (i.e., APC-labeled anti-mouse IgG or FITC-labeled anti-human IgG) were added. Secondary antibody bindings were allowed to proceed for 1 hour. Then, the cells were analyzed with FACS. As shown in FIG. 5, all antibodies, 2E2, 8C11, and Hu8Cl l, bind well to the T cells. This result indicates that the humanized Hu8Cl l retains the binding avidity to activated T cells.

4) CDR affinity optimization and alanine scanning of critical amino acid residues.

[0060] In addition to the above-described back mutations in the framework regions, antibody affinities may be further improved by optimizing CDR sequences. First, we performed alanine scanning of the residues in CDRs to identify residues that are critical for bindings. The various alanine-substituted variants were generated by site-directed mutagenesis and transiently expressed in F293 cells.

[0061] The expression of these mutants may be checked for proper expression, for example using dot blotting. Briefly, a standard curve within the range of expected protein production may be established using a serially diluted Hu8Cl 1, as shown in FIG. 6 A.

[0062] The following Table summarizes the results from dot-blot analysis of alanine mutants of the heavy-chain CDR residues (FIG. 6B):

[0063] The following Table summarizes the results from dot-blot analysis of alanine mutants of the light-chain CDR residues (FIG. 6C):

[0064] The various alanine mutants were constructed in expression vectors, which are transiently expressed in suitable cells, such as 293 cells or CHO cells. Then, the binding affinities of the mutated variants may be analyzed with any suitable methods, such as by ELISA, Biacore, or FACS analysis. FIG. 7 shows representative FACS results of heavy-chain alanine mutants, indicating that G26A mutant lost binding while Y27A retains the binding to Tim3. All alanine mutants (G26A, Y27A, T28A, F29A, T30A, D31A, Y32A, Y33A, M34 A, N35A, R50A, V51A, P53A, S54A, N551, G56A, G57A, N63A, F64A, K65A, G66A, R99A, D100A, S101A, S102A, G103A, Y104A, W105A, F106A, and Y108A) were similarly tested. From these heavy-chain alanine scanning and FACS analyses, it was found that among the heavy-chain mutants, G26A, R99A, S102A, and W105A mutants all lost the bindings, indicating that these residues in the heavy-chain CDRs are critical for Tim-3 bindings. [0065] Similarly, FIG. 8 shows representative FACS results of light-chain alanine mutants, indicating that K24A mutant retains the binding while S26A mutant lost binding to Tim3. All alanine mutants (K24A, S26A, Q27A, V29A, S30A, T31A, V33A, V34A, S50A, P51A, S52A, Y53A, R54A, T56A, Q89A, Q90A, H91A, Y92A, N93A, I94A, P95A, W96A, and T97A) were similarly tested. From the light-chain alanine scanning and FACS analyses, it was found that among the light-chain mutants, S26A, V29A, S30A, T31A, H91A, and N93A all lost their bindings to Tim3, indicating that these residues in the light-chain CDRs are critical for bindings to Tim-3.

[0066] As noted above (FIG. 4 A), mAh 8C11 and Hu8Cl l are specific for Tim-3 and would not bind with Tim-l and Tim-4. In addition, these antibodies are found to be species specific. To test this, Monkey TIM-3 with a green fluorescence protein (GFP) label was constructed (TIM-3 ECD-GFP). The expression vector for transient expression of this recombinant protein was transfected into F293 cells. The transfected F293cells expressing TIM-3 ECD-GFP were reacted with antibodies, including the control 2E2 mAh and 8C11 mAh, and analyzed with FACS. Wild-type F293 cells and anti-mouse IgG APC-conjugate were used as negative controls. As shown in FIG. 9, mAh 2E2 can bind monkey TIM-3, whereas 8C11 mAh could not. These results indicate that 8C11 mAh is species specific; it binds human TIM-3, but not monkey TIM-3.

[0067] The various alanine mutants in the CDR regions were also tested for their binding with monkey TIM-3. It was unexpectedly found that some Hu8Cl l mutants containing alanine mutations in the heavy-chain CDRs can bind monkey TIM-3. As an example, FIG. 10 shows that VH-S101A mutant and VH-S102A mutant of Hu8Cl l can bind to monkey TIM-3. Similarly, VH-T28A, and VH-W105A could also bind monkey TIM-3. These results indicate that these CDR mutations (T28A, S101A, S102A, and W105A) changed the binding properties of Hu8Cl 1.

[0068] Similar FACS analyses were performed with light-chain alanine mutants. It was found that S26A mutant in the light-chain CDR could bind monkey TIM-3, indicating that this mutant also has a changed binding property.

[0069] The following tables summary the FACS and dot-blot analysis results for the various heavy-chain mutants of Hu8Cl 1 :

O indicates binding; X indicates no binding; N indicates not available.

[0070] The following tables summary the FACS and dot-blot analysis results for the various light-chain mutants of Hu8Cl 1 :

O indicates binding; X indicates no binding; N indicates not available.

The xenograft animal model illustrates the ability of tumor growth inhibition by Hu8Cl 1

[0071] Tumor growth is often associated with T cell exhaustion. T cell exhaustion results in reduced immune responses, which permit the cancer cells to grow unchecked. Because anti-TIM-3 antibodies are found to reverse or minimize the conditions of T cell exhaustion, e.g., blocking Galectin-9/TIM-3 binding and enhancing the secretions of IFN-g and TNF-a by T cells, it is likely that these antibodies can prevent or slow the growth of cancer cells. To test this, the following experiment was performed. [0072] As shown in FIG. 11 A, NOD . Cg-Prkdc ^{c l}II2 rg""¹" ^/Ί Y ck N arl mice (5 to

6-week-old) were injected with highly metastatic human lung cancer cells, CL1-5, (a total of 10⁶ cells/lOO mΐ in PBS) subcutaneously (s.c.) on day 0 of the experiment. Mice were randomly separated into groups and received different treatments.

[0073] As shown in FIG. 11 A, 0.5x l0⁶ human fresh PBMCs were injected (i.p.) on days 13 and 20, while treatments with humanized anti-TIM-3 mAh (Hu8Cl 1, 1 mpk, 3 mpk, or 10 mg/kg per injection) were injected on days 13, 15, 17, 20, 23, and 25. For groups receiving cyclophosphamide, cyclophosphamide (50 mg/kg per injection) were administered on day 12.

[0074] Survival and xenograft-versus-host reaction were monitored daily up to 28 days.

Animals that developed clinical symptoms of xenogeneic graft versus host disease (xGVHD) (>15% weight loss, hunched posture, reduced mobility, fur loss, tachypnea) were sacrificed, and an endpoint of survival was recorded. Mice were sacrificed on day 28 and s.c. tumors were removed, weighed, and processed for immune histochemistry (IHC) and FACS analysis.

[0075] As shown in FIG. 11B, cyclophosphamide or anti-TIM3 each are capable of inhibiting tumor growth. Cyclophosphamide is a chemotherapeutic agent. The combination therapy using cyclophosphamide and anti-TIM3 antibody resulted in synergistic effects and the tumor growth is significantly suppressed. These results indicate that antibodies of the invention can be effective cancer therapeutic agents, particularly when used with another immune suppressant or cancer therapeutics.

[0076] Some embodiments of the invention relate to the uses of antibodies of the invention in treating diseases or disorders associated with T cell exhaustion mediated by TIM-3. Such diseases include cancers, such as lung, breast, pancreas, liver, colorectal, or prostate cancers. In accordance with embodiments of the invention, a method for treating such cancer may comprise administering an effective amount of an antibody of the invention to a subject in need thereof. An effective amount is the amount needed to effect the treatments. One skilled in the art would appreciate that the effective amount would depend on the disease, the patient (age, weight), dosage form, route of administration, etc. One skilled in the art can determine the effective amount without undue experimentation. The administration may be by any suitable means, including injections, such as subcutaneous injection, intramuscular injection, intravenous injection, intraperitoneal injection, etc.

[0077] While the invention has been described with respect to a limited number of embodiments, those skilled in the art, having benefit of this disclosure, will appreciate that other embodiments can be devised which do not depart from the scope of the invention as disclosed herein. Accordingly, the scope of the invention should be limited only by the attached claims.

Claims

CLAIMS What is claimed is:

1. A humanized anti-human T-cell immunoglobulin domain and mucin domain 3 (TIM-3) antibody, or a binding fragment thereof, wherein the antibody or the binding fragment comprises a heavy-chain variable domain that comprises framework region sequences derived from a human immunoglobulin and the following complementarity determining region (CDR) sequences: HCDR1 (SEQ ID NO: 3), HCDR2 (SEQ ID NO: 4), HCDR3 (SEQ ID NO: 5), and a light-chain variable domain that comprises framework region sequences from a human immunoglobulin and the following CDR sequences: LCDR1 (SEQ ID NO: 6), LCDR2 (SEQ ID NO: 7), and LCDR3 (SEQ ID NO: 8).

2. The anti-human T-cell immunoglobulin domain and mucin domain 3 (TIM-3) antibody, or the binding fragment thereof, according to claim 1, wherein the framework region sequences in the heavy-chain variable domain are derived from corresponding framework region sequences of IGHVl-2*02.

3. The anti-human T-cell immunoglobulin domain and mucin domain 3 (TIM-3) antibody, or the binding fragment thereof, according to claim 1, wherein the framework region sequences in the light-chain variable domain are derived from corresponding framework region sequences of IGVK4-1*01.

4. The anti-human T-cell immunoglobulin domain and mucin domain 3 (TIM-3) antibody, or the binding fragment thereof, according to claim 1, wherein the heavy-chain variable domain comprises the sequence of SEQ ID NO: 10, or 20, or 21.

5. The anti-human T-cell immunoglobulin domain and mucin domain 3 (TIM-3) antibody, or the binding fragment thereof, according to claim 1, wherein the light-chain variable domain comprises the sequence of SEQ ID NO: 9 or 22.

6. The anti-human T-cell immunoglobulin domain and mucin domain 3 (TIM-3) antibody, or the binding fragment thereof, according to claim 1, wherein the heavy-chain variable domain comprises the sequence of SEQ ID NO: 10, or 20, or 21 and the light-chain variable domain comprises the sequence of SEQ ID NO: 9 or 22.

7. A pharmaceutical composition for use in treating cancer, wherein the pharmaceutical composition comprises the anti-human T-cell immunoglobulin domain and mucin domain 3 (TIM-3) antibody, or the binding fragment thereof, according to any one of claims 1-6.

8. The pharmaceutical composition according to claim 7, the pharmaceutical composition further comprises a chemotherapeutic agent.

9. The pharmaceutical composition according to claim 8, the chemotherapeutic agent is cyclophosphamide.

10. The pharmaceutical composition according to any one of claims 7-9, wherein the cancer is lung, breast, pancreas, liver, colorectal, or prostate cancer.