Proteins binding CD33, NKG2d and CD16

Abstract

To provide multi-specific binding proteins that bind CD33, the NKG2D receptor, and CD16, as well as pharmaceutical compositions and therapeutic methods useful for the treatment of cancer.SOLUTION: A protein comprises: (a) a first antigen-binding site that binds NKG2D; (b) a second antigen-binding site that binds CD33; and (c) an antibody Fc domain or a portion thereof sufficient to bind CD16, or a third antigen-binding site that binds CD16.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] (CROSS REFERENCE TO RELATED APPLICATIONS) This application claims the benefit of priority from U.S. Provisional Application No. 62 / 461,145, filed February 20, 2017, the entire contents of which are incorporated by reference herein for all purposes.

[0002] (Sequence Listing) This application contains a Sequence Listing which has been submitted electronically in ASCII format and is incorporated herein by reference in its entirety. The ASCII copy was created on February 19, 2018, is named DFY-007PC_SL.txt, and is 98,304 bytes in size.

[0003] FIELD OF THE PRESENT ART The present invention relates to multispecific binding proteins that bind to CD33, the NKG2D receptor, and CD16. [Background technology]

[0004] (background) Cancer remains a major health problem, despite the extensive research efforts and scientific advances reported in the literature to treat this disease. Some of the most frequently diagnosed cancers in adults include prostate cancer, breast cancer, and lung cancer. Hematological malignancies are less frequent than solid cancers, but have poorer survival rates. Current treatment options for these cancers are either not effective for all patients, or may have substantial adverse side effects, or both. Other types of cancer also remain difficult to treat using existing treatment options.

[0005] Cancer immunotherapy is desirable because it is highly specific and can use the patient's own immune system to promote the destruction of cancer cells. Fusion proteins such as bispecific T cell engagers are cancer immunotherapies described in the literature that bind to tumor cells and T cells to promote the destruction of tumor cells. Antibodies that bind to certain tumor-associated antigens and certain immune cells have been described in the literature. See, for example, International Publication Nos. WO 2016 / 134371 and WO 2015 / 095412.

[0006] Natural killer (NK) cells are components of the innate immune system and constitute approximately 15% of circulating lymphocytes. NK cells infiltrate virtually all tissues and were initially characterized by their ability to effectively kill tumor cells without the need for prior sensitization. Activated NK cells kill target cells by means similar to cytotoxic T cells, i.e., via cytolytic granules containing perforin and granzymes, as well as via death receptor pathways. Activated NK cells also secrete inflammatory cytokines and chemokines, such as IFN-γ, that enhance the recruitment of other leukocytes to target tissues.

[0007] NK cells respond to signals through a variety of activating and inhibitory receptors on their surface. For example, when NK cells encounter healthy self-cells, their activity is inhibited through activation of killer cell immunoglobulin-like receptors (KIRs). Alternatively, when NK cells encounter foreign or cancer cells, they are activated through activating receptors (e.g., NKG2D, NCR, DNAM1). NK cells are also activated by the constant regions of several immunoglobulins through the CD16 receptor on their surface. The overall susceptibility of NK cells to activation depends on the sum of stimulatory and inhibitory signals.

[0008] CD33 is a member of the sialic acid-binding immunoglobulin-like lectins. As a transmembrane receptor expressed primarily on cells of the myeloid lineage, CD33 regulates inflammation and immune responses through its effect of dampening tyrosine kinase-driven signaling pathways. For example, CD33 has been shown to constitutively suppress the production of proinflammatory cytokines such as IL-1β, TNF-α, and IL-8 by human monocytes.

[0009] CD33 is associated with hematopoietic cancers. CD33 is broadly expressed on blast cells of almost all acute myeloid leukemias (AMLs). Furthermore, hematopoietic cancer stem and / or progenitor cells express CD33 + It has been found that CD33 is a marker of hematopoietic differentiation and has been shown to potentially eradicate malignant stem and / or progenitor cells in such cases while sparing normal hematopoietic stem cells. In addition to its expression in AML, CD33 has also been found in other myeloid neoplasms (e.g., myelodysplastic syndromes and myeloproliferative neoplasms), and in a subset of B-cell and T-cell acute lymphoblastic leukemia (ALL) / lymphoblastic lymphoma. This expression pattern has led to the use of CD33-directed therapeutic agents in patients with malignancies including AML, myelodysplastic syndromes, chronic myelomonocytic leukemia, myeloid blast crisis of chronic myeloid leukemia, and ALL. Summary of the Invention

[0010] (overview) The present invention provides a multispecific binding protein that binds to CD33 on cancer cells and to the NKG2D and CD16 receptors on natural killer cells. Such proteins can engage two or more NK activating receptors and can block the binding of natural ligands to NKG2D. In certain embodiments, the proteins can stimulate NK cells in humans and in other species such as rodents and cynomolgus monkeys. Various aspects and embodiments of the invention are described in further detail below.

[0011] Thus, one embodiment of the invention provides a protein incorporating a first antigen-binding site which binds NKG2D, a second antigen-binding site which binds CD33, and an antibody Fc domain, a sufficient portion thereof which binds CD16, or a third antigen-binding site which binds CD16. Each of these antigen-binding sites may incorporate an antibody heavy chain variable domain and an antibody light chain variable domain (e.g., arranged as in an antibody or fused together into an scFv), or one or more of the antigen-binding sites may incorporate a VF domain, such as a camelid antibody. H H antibodies, or V antibodies such as those found in cartilaginous fishes NAR It may also be a single domain antibody such as an antibody.

[0012] The first antigen-binding site binds to NKG2D and in one embodiment may incorporate a heavy chain variable domain related to SEQ ID NO: 1, for example, by having an amino acid sequence that is at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 1, and / or by incorporating amino acid sequences that are identical to the CDR1 (SEQ ID NO: 54), CDR2 (SEQ ID NO: 55), and CDR3 (SEQ ID NO: 56) sequences of SEQ ID NO: 1. Alternatively, the first antigen-binding site may incorporate a heavy chain variable domain related to SEQ ID NO: 41, and a light chain variable domain related to SEQ ID NO: 42. For example, the heavy chain variable domain of the first antigen-binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 41 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO: 57), CDR2 (SEQ ID NO: 58), and CDR3 (SEQ ID NO: 59) sequences of SEQ ID NO: 41. Similarly, the light chain variable domain of the second antigen-binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 42 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO: 60), CDR2 (SEQ ID NO: 61), and CDR3 (SEQ ID NO: 62) sequences of SEQ ID NO: 42. In other embodiments, the first antigen-binding site can incorporate a heavy chain variable domain related to SEQ ID NO: 43 and a light chain variable domain related to SEQ ID NO: 44. For example, the heavy chain variable domain of the first antigen-binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 43 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO: 63), CDR2 (SEQ ID NO: 64), and CDR3 (SEQ ID NO: 65) sequences of SEQ ID NO: 43.Similarly, the light chain variable domain of the second antigen binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99% or 100%) identical to SEQ ID NO:44 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO:66), CDR2 (SEQ ID NO:67) and CDR3 (SEQ ID NO:68) sequences of SEQ ID NO:44.

[0013] In some embodiments, the first antigen binding site can incorporate a heavy chain variable domain related to SEQ ID NO: 45 and a light chain variable domain related to SEQ ID NO: 46, e.g., by having an amino acid sequence at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 45 and an amino acid sequence at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 46, respectively. In another embodiment, the first antigen binding site can incorporate a heavy chain variable domain related to SEQ ID NO: 47 and a light chain variable domain related to SEQ ID NO: 48, e.g., by having an amino acid sequence that is at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 47 and an amino acid sequence that is at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 48, respectively.

[0014] In some embodiments, the first antigen-binding site can incorporate a heavy chain variable domain related to SEQ ID NO:69 and a light chain variable domain related to SEQ ID NO:70. For example, the heavy chain variable domain of the first antigen-binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO:69 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO:71), CDR2 (SEQ ID NO:72), and CDR3 (SEQ ID NO:73) sequences of SEQ ID NO:69. Similarly, the light chain variable domain of the second antigen-binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 70 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO: 74), CDR2 (SEQ ID NO: 75), and CDR3 (SEQ ID NO: 76) sequences of SEQ ID NO: 70. In some embodiments, the first antigen-binding site can incorporate a heavy chain variable domain related to SEQ ID NO: 77 and a light chain variable domain related to SEQ ID NO: 78. For example, the heavy chain variable domain of the first antigen-binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 77 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO: 69), CDR2 (SEQ ID NO: 80), and CDR3 (SEQ ID NO: 81) sequences of SEQ ID NO: 77. Similarly, the light chain variable domain of the second antigen-binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 78 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO: 82), CDR2 (SEQ ID NO: 83), and CDR3 (SEQ ID NO: 84) sequences of SEQ ID NO: 78.

[0015] In some embodiments, the first antigen-binding site can incorporate a heavy chain variable domain related to SEQ ID NO: 85 and a light chain variable domain related to SEQ ID NO: 86. For example, the heavy chain variable domain of the first antigen-binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 85 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO: 87), CDR2 (SEQ ID NO: 88), and CDR3 (SEQ ID NO: 89) sequences of SEQ ID NO: 85. Similarly, the light chain variable domain of the second antigen binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99% or 100%) identical to SEQ ID NO:86 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO:90), CDR2 (SEQ ID NO:91) and CDR3 (SEQ ID NO:92) sequences of SEQ ID NO:86.

[0016] In some embodiments, the first antigen-binding site can incorporate a heavy chain variable domain related to SEQ ID NO: 133 and a light chain variable domain related to SEQ ID NO: 134. For example, the heavy chain variable domain of the first antigen-binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 133 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO: 135), CDR2 (SEQ ID NO: 136), and CDR3 (SEQ ID NO: 137) sequences of SEQ ID NO: 133. Similarly, the light chain variable domain of the second antigen binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99% or 100%) identical to SEQ ID NO:134 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO:138), CDR2 (SEQ ID NO:139) and CDR3 (SEQ ID NO:140) sequences of SEQ ID NO:134.

[0017] The second antigen-binding site may optionally incorporate a heavy chain variable domain related to SEQ ID NO: 93 and a light chain variable domain related to SEQ ID NO: 94. For example, the heavy chain variable domain of the second antigen-binding site may be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 93 and / or may incorporate amino acid sequences identical to the CDR1 (SEQ ID NO: 95), CDR2 (SEQ ID NO: 96), and CDR3 (SEQ ID NO: 97) sequences of SEQ ID NO: 93. Similarly, the light chain variable domain of the second antigen binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99% or 100%) identical to SEQ ID NO:94 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO:98), CDR2 (SEQ ID NO:99) and CDR3 (SEQ ID NO:100) sequences of SEQ ID NO:94.

[0018] Alternatively, the second antigen-binding site may incorporate a heavy chain variable domain associated with SEQ ID NO: 101 and a light chain variable domain associated with SEQ ID NO: 102. For example, the heavy chain variable domain of the second antigen-binding site may be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 101 and / or may incorporate amino acid sequences identical to the CDR1 (SEQ ID NO: 103), CDR2 (SEQ ID NO: 104), and CDR3 (SEQ ID NO: 105) sequences of SEQ ID NO:101. Similarly, the light chain variable domain of the second antigen binding site may be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99% or 100%) identical to SEQ ID NO:58 and / or may incorporate amino acid sequences identical to the CDR1 (SEQ ID NO:106), CDR2 (SEQ ID NO:107) and CDR3 (SEQ ID NO:108) sequences of SEQ ID NO:102.

[0019] In another embodiment, the second antigen-binding site can incorporate a heavy chain variable domain related to SEQ ID NO: 109 and a light chain variable domain related to SEQ ID NO: 110. For example, the heavy chain variable domain of the second antigen-binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 59 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO: 111), CDR2 (SEQ ID NO: 112), and CDR3 (SEQ ID NO: 113) sequences of SEQ ID NO: 109. Similarly, the light chain variable domain of the second antigen binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99% or 100%) identical to SEQ ID NO:110 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO:114), CDR2 (SEQ ID NO:115) and CDR3 (SEQ ID NO:116) sequences of SEQ ID NO:110.

[0020] In another embodiment, the second antigen-binding site can incorporate a heavy chain variable domain related to SEQ ID NO: 117 and a light chain variable domain related to SEQ ID NO: 118. For example, the heavy chain variable domain of the second antigen-binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 117 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO: 119), CDR2 (SEQ ID NO: 120), and CDR3 (SEQ ID NO: 121) sequences of SEQ ID NO: 117. Similarly, the light chain variable domain of the second antigen binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO:118 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO:122), CDR2 (SEQ ID NO:123), and CDR3 (SEQ ID NO:124) sequences of SEQ ID NO:118.

[0021] In another embodiment, the second antigen-binding site can incorporate a heavy chain variable domain related to SEQ ID NO: 125 and a light chain variable domain related to SEQ ID NO: 126. For example, the heavy chain variable domain of the second antigen-binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%) identical to SEQ ID NO: 125 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO: 127), CDR2 (SEQ ID NO: 128), and CDR3 (SEQ ID NO: 129) sequences of SEQ ID NO: 125. Similarly, the light chain variable domain of the second antigen binding site can be at least 90% (e.g., 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99% or 100%) identical to SEQ ID NO:126 and / or can incorporate amino acid sequences identical to the CDR1 (SEQ ID NO:130), CDR2 (SEQ ID NO:131) and CDR3 (SEQ ID NO:132) sequences of SEQ ID NO:126.

[0022] In some embodiments, the second antigen-binding site incorporates a light chain variable domain having an amino acid sequence identical to the amino acid sequence of a light chain variable domain present in the first antigen-binding site.

[0023] In some embodiments, the protein incorporates a sufficient portion of an antibody Fc domain to bind CD16, where the antibody Fc domain comprises a hinge and CH2 domain, and / or an amino acid sequence at least 90% identical to amino acid sequence 234-332 of a human IgG antibody.

[0024] Formulations containing one of these proteins; cells containing one or more nucleic acids that express these proteins, and methods of enhancing tumor cell death using these proteins are also provided.

[0025] Another aspect of the invention provides a method of treating cancer in a patient, comprising administering to a patient in need thereof a therapeutically effective amount of a multispecific binding protein described herein. Exemplary cancers for treatment using a multispecific binding protein include, for example, those in which the cancer is selected from the group consisting of AML, myelodysplastic syndrome, chronic myelomonocytic leukemia, myeloid blast crisis of chronic myelogenous leukemia, and ALL. [Brief description of the drawings]

[0026] BRIEF DESCRIPTION OF THE DRAWINGS [Figure 1] Figure 1 is a representation of a heterodimeric multispecific antibody. Each arm can represent either an NKG2D binding domain or a CD33 binding domain. In some embodiments, the NKG2D and CD33 binding domains can share a common light chain.

[0027] [Diagram 2] Figure 2 is a representation of a heterodimeric multispecific antibody. The NKG2D or CD33 binding domains can be in scFv format (right arm).

[0028] [Diagram 3] FIG. 3 is a line graph demonstrating the binding affinity of NKG2D binding domains (listed as clones) to human recombinant NKG2D in an ELISA assay.

[0029] [Figure 4] FIG. 4 is a line graph demonstrating the binding affinity of NKG2D binding domains (listed as clones) to cynomolgus monkey recombinant NKG2D in an ELISA assay.

[0030] [Diagram 5]FIG. 5 is a line graph demonstrating the binding affinity of NKG2D binding domains (listed as clones) to mouse recombinant NKG2D in an ELISA assay.

[0031] [Figure 6] FIG. 6 is a bar graph demonstrating binding of NKG2D binding domains (listed as clones) to EL4 cells expressing human NKG2D by flow cytometry showing fold mean fluorescence intensity (MFI) over background.

[0032] [Figure 7] FIG. 7 is a bar graph demonstrating binding of NKG2D binding domains (listed as clones) to EL4 cells expressing mouse NKG2D by flow cytometry showing fold mean fluorescence intensity (MFI) over background.

[0033] [Figure 8] FIG. 8 is a line graph demonstrating the specific binding affinity of the NKG2D binding domain (listed as clones) to recombinant human NKG2D-Fc by competition with the natural ligand ULBP-6.

[0034] [Figure 9] FIG. 9 is a line graph demonstrating the specific binding affinity of NKG2D binding domains (listed as clones) to recombinant human NKG2D-Fc by competition with the natural ligand MICA.

[0035] [Figure 10] FIG. 10 is a line graph demonstrating the specific binding affinity of the NKG2D binding domain (listed as clones) to recombinant murine NKG2D-Fc by competition with the natural ligand Rae-1 delta.

[0036] [Figure 11]FIG. 11 is a bar graph showing activation of human NKG2D by NKG2D binding domains (listed as clones) by quantitating the percentage of TNF-α positive cells expressing human NKG2D-CD3 zeta fusion protein.

[0037] [Figure 12] FIG. 12 is a bar graph showing activation of mouse NKG2D by NKG2D binding domains (listed as clones) by quantitating the percentage of TNF-α positive cells expressing mouse NKG2D-CD3 zeta fusion protein.

[0038] [Figure 13] FIG. 13 is a bar graph showing activation of human NK cells by NKG2D binding domains (listed as clones).

[0039] [Figure 14] FIG. 14 is a bar graph showing activation of human NK cells by NKG2D binding domains (listed as clones).

[0040] [Figure 15] FIG. 15 is a bar graph showing activation of mouse NK cells by NKG2D binding domains (listed as clones).

[0041] [Figure 16] FIG. 16 is a bar graph showing activation of mouse NK cells by NKG2D binding domains (listed as clones).

[0042] [Figure 17] FIG. 17 is a bar graph showing the cytotoxic effect of NKG2D binding domains (listed as clones) on tumor cells.

[0043] [Figure 18]FIG. 18 is a bar graph showing the melting temperatures of NKG2D binding domains (listed as clones) as measured by differential scanning fluorimetry.

[0044] [Figure 19] Figures 19A-19C are bar graphs of synergistic activation of NK cells using CD16 and NKG2D binding. Figure 19A demonstrates levels of CD107a; Figure 19B demonstrates levels of IFNγ; Figure 19C demonstrates levels of CD107a and IFNγ. Graphs show mean (n=2) ± SD. Data are representative of 5 independent experiments using 5 different healthy donors.

[0045] [Figure 20] Figure 20 is a representation of TriNKET in the Triomab format, a trifunctional bispecific antibody that maintains an IgG-like shape. This chimera consists of two half antibodies, each with one light and one heavy chain, derived from two parent antibodies. The Triomab format may be a heterodimeric construct containing half a rat antibody and half a mouse antibody.

[0046] [Figure 21] Figure 21 is a representation of a KiH Common Light Chain (LC) form of TriNKET incorporating knobs-into-holes (KIH) technology. KiH is a heterodimer containing two Fabs that bind targets 1 and 2, and an Fc stabilized by a heterodimerization mutation. TriNKET in KiH format may be a heterodimeric construct with two fabs that bind target 1 and target 2, containing two different heavy chains and a common light chain that pairs with both heavy chains.

[0047] [Figure 22]Figure 22 is a representation of TriNKET in the form of a Dual Variable Domain Immunoglobulin (DVD-Ig™) that combines the target binding domains of two monoclonal antibodies via a flexible natural linker, resulting in a tetravalent IgG-like molecule. DVD-Ig™ is a homodimeric construct in which a variable domain targeting antigen 2 is fused to the N-terminus of the variable domain of a Fab targeting antigen 1. The construct contains a conventional Fc.

[0048] [Diagram 23] Figure 23 is a representation of TriNKET in orthogonal Fab interface (Ortho-Fab) form, a heterodimeric construct containing two Fabs that bind target 1 and target 2 fused to an Fc. LC-HC pairing is ensured by the orthogonal interface. Heterodimerization is ensured by mutations in the Fc.

[0049] [Figure 24] FIG. 24 is a representation of TrinKET in the 2-in-1 Ig format.

[0050] [Diagram 25] Figure 25 is a representation of the ES form of TriNKET, a heterodimeric construct containing two different Fabs that bind target 1 and target 2 fused to an Fc. Heterodimerization is ensured by electrostatic steering mutations in the Fc.

[0051] [Figure 26] Figure 26 is a representation of a Fab arm exchanged form of TriNKET: an antibody in which the Fab arms have been exchanged by replacing the heavy chain and the associated light chain (half molecule) with a heavy-light chain pair from another molecule, resulting in a bispecific antibody. The Fab arm exchanged form (cFae) is a heterodimer containing two Fabs that bind targets 1 and 2, and an Fc that is stabilized by a heterodimerization mutation.

[0052] [Figure 27]FIG. 27 is a representation of the SEED Body form of TriNKET, a heterodimer containing two Fabs that bind targets 1 and 2, and an Fc that is stabilized by a heterodimerization mutation.

[0053] [Figure 28] Figure 28 is a representation of the LuZ-Y form of TriNKET in which a leucine zipper is used to induce heterodimerization of two different HCs. The LuZ-Y form is a heterodimer containing two different scFabs that bind targets 1 and 2 fused to an Fc. Heterodimerization is ensured through a leucine zipper motif fused to the C-terminus of the Fc.

[0054] [Figure 29] FIG. 29 is a representation of TriNKET in Cov-X-Body form.

[0055] [Diagram 30] Figures 30A-30B are representations of the κλ-Body form of TriNKET, a heterodimeric construct with two different Fabs fused to an Fc stabilized by heterodimerization mutations: Fab1, which targets antigen 1, contains a kappa LC, while the second Fab, which targets antigen 2, contains a lambda LC. Figure 30A is an exemplary representation of one form of κλ-Body; Figure 30B is an exemplary representation of another κλ-Body.

[0056] [Diagram 31] Figure 31: Construct of an Oasc-Fab heterodimer comprising a Fab that binds target 1 and a scFab that binds target 2 fused to an Fc. Heterodimerization is ensured by mutations in the Fc.

[0057] [Diagram 32]Figure 32 shows DuetMab, a heterodimeric construct containing two different Fabs that bind antigens 1 and 2, and an Fc stabilized by heterodimerization mutations. Fab1 and 2 contain differential S-S bridges that ensure correct light chain (LC) and heavy chain (HC) pairing.

[0058] [Diagram 33] Figure 33 is a CrossmAb, which is a heterodimeric construct with two different Fabs that bind targets 1 and 2 fused to an Fc stabilized by heterodimerization. The CL and CH1 domains, and the VH and VL domains are switched, e.g., CH1 is fused in tandem with VL, while CL is fused in tandem with VH.

[0059] [Diagram 34] Figure 34 is Fit-Ig, a homodimeric construct in which a Fab that binds antigen 2 is fused to the N-terminus of the HC of a Fab that binds antigen 1. This construct contains wild-type Fc.

[0060] [Diagram 35] Figures 35A-35B are binding profiles of CD33-targeting TriNKET to NKG2D expressed on EL4 cells. Figure 35A shows the binding of TriNKET compared to monoclonal antibodies containing the corresponding NKG2D binding domain. Figure 35B shows the binding profile of CD33-targeting TriNKET containing six different NKG2D binding domains.

[0061] [Fig. 36A-B]Figures 36A and 36B are binding profiles of CD33-targeting TriNKET to CD33 expressed on MV4-11 human AML cells. Figure 36C is a binding profile of CD33-targeting TriNKET and a CD33 monoclonal antibody to CD33 expressed on Molm-13 human AML cells. Figure 36D is a binding profile of CD33-targeting TriNKET and a CD33 monoclonal antibody to CD33 expressed on the human AML cell line MV4-11. [Fig. 36C-D] Figures 36A and 36B are binding profiles of CD33-targeting TriNKET to CD33 expressed on MV4-11 human AML cells. Figure 36C is a binding profile of CD33-targeting TriNKET and a CD33 monoclonal antibody to CD33 expressed on Molm-13 human AML cells. Figure 36D is a binding profile of CD33-targeting TriNKET and a CD33 monoclonal antibody to CD33 expressed on the human AML cell line MV4-11.

[0062] [Figure 37] Figures 37A-37B are line graphs demonstrating TriNKET-mediated activation of resting or IL-2-activated human NK cells in co-culture with the CD33-expressing human AML cell line MV4-11. Figure 37A shows TriNKET-mediated activation of resting human NK cells. Figure 37B shows TriNKET-mediated activation of IL-2-activated human NK cells from the same donor. NK cells alone, NK cells in co-culture with MV4-11 cells without TriNKET, and TriNKET targeting CD20 were used as controls.

[0063] [Figure 38] Figures 38A-38C are histograms showing the expression of high affinity FcRγI (CD64) in three human AML cell lines, the Molm-13 cell line (Figure 38A), the MV4-11 cell line (Figure 38B), and the THP-1 cell line (Figure 38C).

[0064] [Fig. 39A-B] Figures 39A-39B are line graphs of CD33 monoclonal antibody or TriNKET mediated activation of human NK cells in co-culture with either Molm-13 (Figure 39B) or THP-1 (Figure 39A) cells. Figure 39C shows activation of human NK cells by TriNKET in co-culture with MV4-11 human AML cell line. HER2-TriNKET was used as a control. [Figure 39C] Figures 39A-39B are line graphs of CD33 monoclonal antibody or TriNKET mediated activation of human NK cells in co-culture with either Molm-13 (Figure 39B) or THP-1 (Figure 39A) cells. Figure 39C shows activation of human NK cells by TriNKET in co-culture with MV4-11 human AML cell line. HER2-TriNKET was used as a control.

[0065] [Fig. 40A-B] Figures 40A-40C are line graphs of human NK cytotoxicity against three human AML cell lines mediated by TriNKET targeting CD33 and the corresponding CD33 monoclonal antibody. Figure 40A shows that the CD33 monoclonal antibody showed reduced efficacy against MV4-11 cells, which express CD64 at lower levels than THP-1. Figure 40B demonstrates that the CD33 monoclonal antibody showed good efficacy against Molm-13 cells, which do not express CD64. Figure 40C demonstrates that the CD33 monoclonal antibody had no effect on THP-1 cells. [Figure 40C]Figures 40A-40C are line graphs of human NK cytotoxicity against three human AML cell lines mediated by TriNKET targeting CD33 and the corresponding CD33 monoclonal antibody. Figure 40A shows that the CD33 monoclonal antibody showed reduced efficacy against MV4-11 cells, which express CD64 at lower levels than THP-1. Figure 40B demonstrates that the CD33 monoclonal antibody showed good efficacy against Molm-13 cells, which do not express CD64. Figure 40C demonstrates that the CD33 monoclonal antibody had no effect on THP-1 cells.

[0066] [Diagram 41] FIG. 41 shows TriNKET-mediated cytotoxicity of resting human NK cells against Molm-13 cells.

[0067] [Diagram 42] Figure 42A is a bar graph showing that B cells from healthy donors are protected from TriNKET-mediated lysis targeting CD33. Figure 42B is a bar graph showing that autologous CD33+ bone marrow cells are protected from TriNKET-mediated NK cell responses targeting CD33 and were therefore resistant to TriNKET-mediated lysis. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0068] (Detailed Description) The present invention provides multispecific binding proteins that bind to CD33 on cancer cells and the NKG2D and CD16 receptors on natural killer cells to activate natural killer cells, pharmaceutical compositions comprising such multispecific binding proteins, and therapeutic methods using such multispecific proteins and pharmaceutical compositions, including for the treatment of cancer. Various aspects of the invention are described below in separate chapters; however, aspects of the invention described in one particular chapter should not be limited to any particular chapter.

[0069] To facilitate the understanding of this invention, a number of terms and phrases are defined below.

[0070] As used herein, the words "a" and "an" mean "one or more" and include plurals unless the context requires otherwise.

[0071] As used herein, the term "antigen-binding site" refers to the portion of an immunoglobulin molecule that participates in antigen binding. In human antibodies, the antigen-binding site is formed by amino acid residues of the N-terminal variable ("V") regions of the heavy ("H") and light ("L") chains. Three highly divergent stretches within the V regions of the heavy and light chains are called "hypervariable regions" and are sandwiched between more conserved adjacent stretches known as "framework regions" or "FRs". The term "FR" therefore refers to the amino acid sequences naturally found between and adjacent to the hypervariable regions in immunoglobulins. In human antibody molecules, the three hypervariable regions of the light chain and the three hypervariable regions of the heavy chain are arranged with respect to each other in three-dimensional space to form an antigen-binding surface. This antigen-binding surface is complementary to the three-dimensional surface of the bound antigen, and the three hypervariable regions of each of the heavy and light chains are called "complementarity determining regions" or "CDRs". In some animals, such as camelids and cartilaginous fish, the antigen-binding site is formed by a single antibody chain providing a "single domain antibody". The antigen-binding site can be present in an intact antibody, an antigen-binding fragment of an antibody which retains the antigen-binding surface, or a recombinant polypeptide such as an scFv which uses a peptide linker to connect the heavy chain variable domain to the light chain variable domain in a single polypeptide.

[0072] As used herein, the term "tumor-associated antigen" refers to any antigen associated with cancer, including, but not limited to, proteins, glycoproteins, gangliosides, carbohydrates, lipids, etc. Such antigens can be expressed on malignant cells or in the tumor microenvironment, such as tumor-associated vasculature, extracellular matrix, mesenchymal stroma, or immune infiltrates.

[0073] As used herein, the terms "subject" and "patient" refer to an organism to be treated by the methods and compositions described herein. Such organisms preferably include, but are not limited to, mammals (e.g., murine, simian, equine, bovine, porcine, canine, feline, etc.), and more preferably include humans.

[0074] As used herein, the term "effective amount" refers to an amount of a compound (e.g., a compound of the present invention) sufficient to produce a beneficial or desired result. An effective amount can be administered in one or more administrations, applications, or dosages, and is not intended to be limited to a particular formulation or route of administration. As used herein, the term "treating" includes any effect that results in the improvement of a condition, disease, disorder, etc., or the amelioration of symptoms thereof, such as, for example, alleviation, reduction, modulation, amelioration, or elimination.

[0075] As used herein, the term "pharmaceutical composition" refers to a combination of an active agent with a carrier, inert or active, that makes the composition particularly suitable for in vivo or ex vivo diagnostic or therapeutic applications.

[0076] As used herein, the term "pharmaceutical acceptable carrier" refers to any standard pharmaceutical carrier, such as phosphate buffered saline solution, water, emulsions (e.g., oil / water or water / oil emulsions), and various types of wetting agents. The composition may also contain stabilizers and preservatives. For examples of carriers, stabilizers, and adjuvants, see, for example, Martin's Remington's Pharmaceutical Sciences, 15th Edition, Mack Publ. Co., Easton, PA (1975).

[0077] As used herein, the term "pharmaceutical acceptable salt" refers to any pharmaceutical acceptable salt (e.g., acid or base) of a compound of the present invention that, upon administration to a subject, can provide the compound of the present invention, or an active metabolite or residue thereof. As known to those skilled in the art, the "salt" of a compound of the present invention may be derived from inorganic or organic acids and bases. Exemplary acids include, but are not limited to, hydrochloric acid, hydrobromic acid, sulfuric acid, nitric acid, perchloric acid, fumaric acid, maleic acid, phosphoric acid, glycolic acid, lactic acid, salicylic acid, succinic acid, toluene-p-sulfonic acid, tartaric acid, acetic acid, citric acid, methanesulfonic acid, ethanesulfonic acid, formic acid, benzoic acid, malonic acid, naphthalene-2-sulfonic acid, benzenesulfonic acid, and the like. Other acids, such as oxalic acid, while not themselves pharmaceutical acceptable, may be utilized in the preparation of salts useful as intermediates in obtaining the compounds of the present invention and their pharmaceutical acceptable acid addition salts.

[0078] Exemplary bases include alkali metal (e.g., sodium) hydroxides, alkaline earth metal (e.g., magnesium) hydroxides, ammonia, and bases of the formula NW 4 + (Wherein, W is C 1~4 Examples of suitable compounds include, but are not limited to, compounds in which the aryl group is an alkyl group.

[0079] Exemplary salts include, but are not limited to, acetate, adipate, alginate, aspartate, benzoate, benzenesulfonate, bisulfate, butyrate, citrate, camphorate, camphorsulfonate, cyclopentanepropionate, digluconate, dodecylsulfate, ethanesulfonate, fumarate, flucoheptanoate, glycerophosphate, hemisulfate, heptanoate, hexanoate, hydrochloride, hydrobromide, hydroiodide, 2-hydroxyethanesulfonate, lactate, maleate, methanesulfonate, 2-naphthalenesulfonate, nicotinate, oxalate, palmoate, pectinate, persulfate, phenylpropionate, picrate, pivalate, propionate, succinate, tartrate, thiocyanate, tosylate, undecanoate, and the like. Other examples of salts include Na + , N.H. 4 + , and N.W. 4 + (Wherein, W is C 1~4 and the anion of a compound of the present invention complexed with a suitable cation, such as an alkyl group.

[0080] For therapeutic use, the salts of the compounds of the invention are contemplated as pharma- ceutical acceptable. However, salts of acids and bases that are non-pharma- ceutical acceptable may also find use, for example, in the preparation or purification of a pharma- ceutical acceptable compound.

[0081] Throughout the description, where compositions are described as having, containing, or comprising specific components, or where processes and methods are described as having, containing, or comprising specific steps, it is contemplated that there are additionally compositions of the invention that consist essentially of or consist of the specified components, and processes and methods of the invention that consist essentially of or consist of the specified process steps.

[0082] As a general matter, compositions specifying percentages are by weight unless otherwise specified. Further, if a variable is not accompanied by a definition, the preceding definition of that variable will control.

[0083] I. Proteins The present invention provides a multispecific binding protein that binds to CD33 on cancer cells and to the NKG2D and CD16 receptors on natural killer cells to activate the natural killer cells. The multispecific binding protein is useful in the pharmaceutical compositions and methods of treatment described herein. Binding of the multispecific binding protein to the NKG2D and CD16 receptors on natural killer cells enhances the activity of the natural killer cells toward the destruction of the cancer cells. Binding of the multispecific binding protein to CD33 on cancer cells brings the cancer cells into the vicinity of the natural killer cells, which facilitates the direct or indirect destruction of the cancer cells by the natural killer cells. Further, a description of exemplary multispecific binding proteins is provided below.

[0084] The first component of the multispecific binding protein binds to NK cells, γδ T cells and CD8 + The multispecific binding proteins bind to cells that express the NKG2D receptor, including, but not limited to, αβ T cells. Upon binding to NKG2D, the multispecific binding proteins may block natural ligands such as ULBP6 and MICA from binding to and activating the NKG2D receptor.

[0085] The second component of the multispecific binding protein binds to cells that express CD33, including, but not limited to, AML, myelodysplastic syndromes, chronic myelomonocytic leukemia, myeloid blast crisis of chronic myelogenous leukemia, and ALL.

[0086] The third component of the multispecific binding protein binds to cells that express CD16, an Fc receptor on the surface of leukocytes, including natural killer cells, macrophages, neutrophils, eosinophils, mast cells, and follicular dendritic cells.

[0087] The multispecific binding proteins described herein can take a variety of formats. For example, one format is a heterodimeric multispecific antibody comprising a first immunoglobulin heavy chain, a first immunoglobulin light chain, a second immunoglobulin heavy chain, and a second immunoglobulin light chain (FIG. 1). The first immunoglobulin heavy chain comprises a first Fc (hinge-CH2-CH3) domain, a first heavy chain variable domain, and optionally a first CH1 heavy chain domain. The first immunoglobulin light chain comprises a first light chain variable domain and a first light chain constant domain. The first immunoglobulin light chain forms an antigen binding site with the first immunoglobulin heavy chain that binds to NKG2D. The second immunoglobulin heavy chain comprises a second Fc (hinge-CH2-CH3) domain, a second heavy chain variable domain, and optionally a second CH1 heavy chain domain. The second immunoglobulin light chain comprises a second light chain variable domain and a second light chain constant domain. The second immunoglobulin light chain, together with the second immunoglobulin heavy chain, forms an antigen-binding site that binds to CD33. The first Fc domain and the second Fc domain together can bind to CD16 (FIG. 1). In some embodiments, the first immunoglobulin light chain can be identical to the second immunoglobulin light chain.

[0088] Another exemplary format includes a heterodimeric multispecific antibody comprising a first immunoglobulin heavy chain, a second immunoglobulin heavy chain, and an immunoglobulin light chain (Figure 2). The first immunoglobulin heavy chain comprises a first Fc (hinge-CH2-CH3) domain fused via a linker or antibody hinge to a single chain variable fragment (scFv) comprised of a heavy chain variable domain and a light chain variable domain that pair and bind to NKG2D or CD33. The second immunoglobulin heavy chain comprises a second Fc (hinge-CH2-CH3) domain, a second heavy chain variable domain, and optionally a CH1 heavy chain domain. The immunoglobulin light chain comprises a light chain variable domain and a constant light chain domain. The second immunoglobulin heavy chain pairs with the immunoglobulin light chain and binds to NKG2D or CD33. The first Fc domain and the second Fc domain together can bind to CD16 (Figure 2).

[0089] One or more additional binding motifs may be fused to the C-terminus of the constant region CH3 domain, optionally via a linker sequence. In certain embodiments, the antigen binding site may be a single chain or a disulfide-stabilized variable region (scFv), or may form a tetravalent or trivalent molecule.

[0090] In some embodiments, the multispecific binding protein is in the form of a Triomab, which is a trifunctional bispecific antibody that maintains an IgG-like shape. This chimera consists of two half antibodies, each with one light chain and one heavy chain, derived from two parent antibodies.

[0091] In some embodiments, the multispecific binding protein is a KiH Common Light Chain (LC) format, which encompasses knobs-into-holes (KIH) technology. KIH uses C to enhance heterodimerization. HThe concept behind "knob-into-hole (KiH)" Fc technology involves remodeling the three domains to create either a "knob" or a "hole" in each heavy chain. The concept behind "knob-into-hole (KiH)" Fc technology is to create a "knob" or a "hole" in each heavy chain by replacing small residues with bulky residues (e.g., T366W in the EU numbering). CH3A ), the aim was to introduce a "knob" into one of the CH3 domains (CH3A). To accommodate this "knob", the residues closest to the knob were replaced with smaller residues (e.g., T366S / L368A / Y407V CH3B), and a complementary "hole" surface was created on the other CH3 domain (CH3B). The "hole" mutations were optimized by structure-guided phage library screening (Atwell S, Ridgway JB, Wells JA, Carter P. Stable heterodimers from remodeling the domain interface of a homodimer using a phage display library. J. Mol. Biol. (1997) 270(1):26-35).X-ray crystal structure of the KiH Fc variant (Elliott JM, Ultsch M, Lee J, Tong R, Takeda K, Spiess C, "Antiparallel conformation of knoband hole aglycosylated half antibody homodimers is mediated by a CH2-CH3 hydrophobic interaction," J. Mol. Biol. (2014) 426(9):1947-57; Mimoto F, Kadono S, Katada H, Igawa T, Kamikawa T, Hattori K, "Crystal structure of a novel asymmetrically engineered Fc variant with improved affinity for Fc gamma Rs," Mol. Immunol. (2014) 58(1):132-8) demonstrated that heterodimerization is thermodynamically favored by hydrophobic interactions driven by steric complementarity at the core interface between the CH3 domains, whereas knob-knob and hole-hole interfaces disfavor homodimerization due to steric hindrance and disruption of favorable interactions, respectively.

[0092] In some embodiments, the multispecific binding protein is in the form of a dual variable domain immunoglobulin (DVD-Ig™), which combines the target binding domains of two monoclonal antibodies via a flexible natural linker to generate a tetravalent IgG-like molecule.

[0093] In some embodiments, the multispecific binding protein is in the form of an orthogonal Fab interface (Ortho-Fab). In the ortho-Fab IgG approach (Lewis SM, Wu X, Pustilnik A, Sereno A, Huang F, Rick HL, et al., "Generation of bispecific IgG antibodies by structure-based design of an orthogonal Fab interface," Nat. Biotechnol. (2014) 32(2):191-8), structure-based region design involves the LC and HC of only one Fab. VH-CH1 At the interface, complementary mutations are introduced without making any changes to the other Fab.

[0094] In some embodiments, the multispecific binding protein is in a 2-in-1 Ig format. In some embodiments, the multispecific binding protein is in an ES form, which is a heterodimeric construct containing two different Fabs that bind target 1 and target 2 fused to an Fc. Heterodimerization is ensured by electrostatic steering mutations in the Fc. In some embodiments, the multispecific binding protein is in a κλ-Body form, which is a heterodimeric construct with two different Fabs fused to an Fc stabilized by heterodimerization mutations: Fab1, which targets antigen 1, contains a kappa LC, while the second Fab, which targets antigen 2, contains a lambda LC. Figure 30A is an exemplary representation of one form of a κλ-Body; Figure 30B is an exemplary representation of another κλ-Body.

[0095] In some embodiments, the multispecific binding protein is in Fab Arm Exchange format (an antibody that exchanges Fab arms by swapping the heavy chain and the associated light chain (half molecule) with a heavy-light chain pair from another molecule, resulting in a bispecific antibody). In some embodiments, the multispecific binding protein is in SEED Body format. The Strand Exchange Remodeling Domain (SEED) platform was designed to generate asymmetric bispecific antibody-like molecules, a potential to expand the therapeutic application of natural antibodies. This protein remodeling platform is based on swapping structurally related sequences of immunoglobulins within the conserved CH3 domain. The SEED design allows for the efficient generation of AG / GA heterodimers, while avoiding homodimerization of the AG and GA SEED CH3 domains (Muda M. et al., Protein Eng. Des. Sel. (2011, 24(5):447-54)). In some embodiments, the multispecific binding protein is in the LuZ-Y format, in which a leucine zipper is used to induce heterodimerization of two different HCs (Wranik, BJ. et al., J. Biol. Chem. (2012), 287:43331-9).

[0096] In some embodiments, the multispecific binding protein is in the form of a Cov-X-Body. In a bispecific CovX-Body, two different peptides are brought together using a branched azetidinone linker and fused to a scaffold antibody in a site-specific manner under mild conditions. The pharmacophore is responsible for the functional activity, while the antibody scaffold confers a long half-life and Ig-like distribution. The pharmacophore can be chemically optimized or replaced with other pharmacophores to generate optimized or unique bispecific antibodies (Doppalapudi VR et al., PNAS (2010), 107(52);22611-22616).

[0097] In some embodiments, the multispecific binding protein is in the form of an Oasc-Fab heterodimer comprising a Fab that binds target 1 and a scFab that binds target 2 fused to an Fc. Heterodimerization is ensured by mutations in the Fc.

[0098] In some embodiments, the multispecific binding protein is in the DuetMab format, which is a heterodimeric construct containing two different Fabs that bind antigens 1 and 2, and an Fc stabilized by a heterodimerization mutation. Fabs 1 and 2 contain differential S-S bridges that ensure correct LC and HC pairing.

[0099] In some embodiments, the multispecific binding protein is in a CrossmAb format, which is a heterodimeric construct with two different Fabs that bind targets 1 and 2 fused to an Fc that is stabilized by heterodimerization. The CL and CH1 domains, and the VH and VL domains are switched, e.g., CH1 is fused in tandem with the VL, while CL is fused in tandem with the VH.

[0100] In some embodiments, the multispecific binding protein is in the Fit-Ig format, which is a homodimeric construct in which a Fab that binds antigen 2 is fused to the N-terminus of the HC of a Fab that binds antigen 1. This construct contains a wild-type Fc.

[0101] Table 1 lists peptide sequences of heavy and light chain variable domains that can bind to NKG2D in combination. The NKG2D binding domains can vary in their binding affinity to NKG2D, but nevertheless, they all activate human NKG2D and NK cells. [Table 1] TIFF2025061059000003.tif223164 TIFF2025061059000004.tif222164 TIFF2025061059000005.tif223164 TIFF2025061059000006.tif224164 TIFF2025061059000007.tif224164 TIFF2025061059000008.tif197164

[0102] Alternatively, as described in U.S. Pat. No. 9,273,136, a heavy chain variable domain defined by SEQ ID NO:49 can be paired with a light chain variable domain defined by SEQ ID NO:50 to form an antigen-binding site capable of binding to NKG2D. [ka]

[0103] Alternatively, as described in U.S. Pat. No. 7,879,985, a heavy chain variable domain defined by SEQ ID NO:51 can be paired with a light chain variable domain defined by SEQ ID NO:52 to form an antigen-binding site capable of binding to NKG2D. [ka]

[0104] Table 2 lists peptide sequences of heavy and light chain variable domains that in combination are capable of binding to CD33. [Table 2] TIFF2025061059000012.tif240164

[0105] Alternatively, novel antigen binding sites capable of binding to CD33 can be identified by screening for binding to the amino acid sequence defined by SEQ ID NO:53. [ka]

[0106] Within the Fc domain, CD16 binding is mediated by the hinge region and the CH2 domain. For example, within human IgG1, the interaction with CD16 is primarily focused on amino acid residues Asp 265-Glu 269, Asn 297-Thr 299, Ala 327-Ile 332, Leu 234-Ser 239 in the CH2 domain, and carbohydrate residue N-acetyl-D-glucosamine (see Sondermann et al., Nature, 406 (6793):267-273). Based on the known domains, mutations can be selected to enhance or reduce binding affinity to CD16, such as by using a phage display library or a yeast surface display cDNA library, or can be designed based on the known three-dimensional structure of the interaction.

[0107] The assembly of heterodimeric antibody heavy chains can be achieved by expressing two different antibody heavy chain sequences in the same cell, which can result in the assembly of homodimers of each antibody heavy chain as well as the assembly of heterodimers. Enhancing the preferential assembly of heterodimers can be achieved by incorporating different mutations in the CH3 domain of each antibody heavy chain constant region, as shown in US 13 / 494870, US 16 / 028850, US 11 / 533709, US 12 / 875015, US 13 / 289934, US 14 / 773418, US 12 / 811207, US 13 / 866756, US 14 / 647480, and US 14 / 830336. For example, mutations can be made in the CH3 domain based on human IgG1 by incorporating distinct pairs of amino acid substitutions in the first and second polypeptides that allow the two chains to selectively heterodimerize with one another. All amino acid substitution positions described below are numbered according to the EU index as in Kabat.

[0108] In one scenario, an amino acid substitution in a first polypeptide replaces an original amino acid with a larger amino acid selected from arginine (R), phenylalanine (F), tyrosine (Y), or tryptophan (W), and at least one amino acid substitution in a second polypeptide replaces an original amino acid with a smaller amino acid selected from alanine (A), serine (S), threonine (T), or valine (V), such that the larger amino acid substitution (protuberance) fits within the surface (cavity) of the smaller amino acid substitution. For example, one polypeptide can incorporate a T366W substitution and the other can incorporate three substitutions including T366S, L368A, and Y407V.

[0109] The antibody heavy chain variable domain of the present invention can optionally be coupled to an amino acid sequence that is at least 90% identical to an antibody constant region, such as an IgG constant region, with or without a CH1 domain, including the hinge, CH2 and CH3 domains. In some embodiments, the amino acid sequence of the constant region is at least 90% identical to a human antibody constant region, such as a human IgG1 constant region, IgG2 constant region, IgG3 constant region, or IgG4 constant region. In some other embodiments, the amino acid sequence of the constant region is at least 90% identical to an antibody constant region from another mammal, such as a rabbit, dog, cat, mouse, or horse. One or more mutations can be incorporated into the constant region relative to the human IgG1 constant region, for example at Q347, Y349, L351, S354, E356, E357, K360, Q362, S364, T366, L368, K370, N390, K392, T394, D399, S400, D401, F405, Y407, K409, T411 and / or K439. Exemplary substitutions include, for example, Q347E, Q347R, Y349S, Y349K, Y349T, Y349D, Y349E, Y349C, T350V, L351K, L351D, L351Y, S354C, E356K, E357Q, E357L, E357W, K360E, K360W, Q362E, S364K, S364E, S364H, S364D, T366V, T366I, T366L, T366M, T366K, T366W, T366S , L368E, L368A, L368D, K370S, N390D, N390E, K392L, K392M, K392V, K392F, K392D, K392E, T394F, T394W, D399R, D399K, D399V, S400K, S400R, D401K, F405A, F405T, Y407A, Y407I, Y407V, K409F, K409W, K409D, T411D, T411E, K439D, and K439E.

[0110] In certain embodiments, mutations that can be incorporated into the CH1 of the human IgG1 constant region can be at amino acids V125, F126, P127, T135, T139, A140, F170, P171, and / or V173. In certain embodiments, mutations that can be incorporated into the Cκ of the human IgG1 constant region can be at amino acids E123, F116, S176, V163, S174, and / or T164.

[0111] The amino acid substitutions may be selected from the following set of substitutions shown in Table 3. [Table 3]

[0112] Alternatively, the amino acid substitutions may be selected from the set of substitutions shown in Table 4 below. [Table 4]

[0113] Alternatively, the amino acid substitutions may be selected from the set of substitutions shown in Table 5 below. [Table 5]

[0114] Alternatively, at least one amino acid substitution in each polypeptide chain may be selected from Table 6. [Table 6]

[0115] Alternatively, at least one amino acid substitution may be selected from the set of following substitutions in Table 7, where the position shown in the column in the first polypeptide is replaced with any known negatively charged amino acid, and the position shown in the column in the second polypeptide is replaced with any known positively charged amino acid. [Table 7]

[0116] Alternatively, at least one amino acid substitution may be selected from the following set in Table 8, where the position shown in the column in the first polypeptide is replaced with any known positively charged amino acid and the position shown in the column in the second polypeptide is replaced with any known negatively charged amino acid. [Table 8]

[0117] Alternatively, the amino acid substitutions may be selected from the following set of substitutions shown in Table 9. [Table 9]

[0118] Alternatively, or in addition, the structural stability of a heteromultimeric protein may be increased by introducing S354C into either the first or second polypeptide chain and Y349C into the opposite polypeptide chain, which forms an artificial disulfide bridge within the interface of the two polypeptides.

[0119] The multispecific proteins described above can be produced using recombinant DNA techniques well known to those skilled in the art. For example, a first nucleic acid sequence encoding a first immunoglobulin heavy chain can be cloned into a first expression vector; a second nucleic acid sequence encoding a second immunoglobulin heavy chain can be cloned into a second expression vector; a third nucleic acid sequence encoding an immunoglobulin light chain can be cloned into a third expression vector; the first, second and third expression vectors can be stably transfected together into a host cell to produce a multimeric protein.

[0120] To achieve the highest yield of the multispecific protein, different ratios of the first, second, and third expression vectors can be explored to determine the optimal ratio for transfection into the host cells. After transfection, single clones can be isolated for cell bank generation using methods known in the art, such as limiting dilution, ELISA, FACS, microscopy, or Clonepix.

[0121] The clones can be cultured under conditions suitable for scale-up in bioreactors to maintain expression of the multispecific proteins. The multispecific proteins can be isolated and purified using methods known in the art, including centrifugation, depth filtration, cell lysis, homogenization, freeze-thaw, affinity purification, gel filtration, ion exchange chromatography, hydrophobic interaction exchange chromatography, and mixed-mode chromatography.

[0122] II. Characteristics of multispecific proteins In some embodiments, the multispecific proteins described herein, which comprise an NKG2D binding domain and a binding domain for CD33, bind to cells expressing human NKG2D.In some embodiments, the multispecific proteins bind to tumor-associated antigen CD33 at a level comparable to that of monoclonal antibodies with the same CD33 binding domain.However, the multispecific proteins described herein can be more effective than the corresponding CD33 monoclonal antibodies in reducing tumor growth and killing cancer cells expressing CD33.

[0123] In certain embodiments, the multispecific proteins described herein, comprising an NKG2D binding domain and a binding domain for CD33, can activate primary human NK cells when cultured with tumor cells expressing the antigen CD33. NK cell activation is characterized by CD107a degranulation and increased IFNγ cytokine production. Moreover, compared to monoclonal antibodies comprising the same CD33 binding domain, the multispecific proteins show superior activation of human NK cells in the presence of tumor cells expressing the antigen CD33.

[0124] In certain embodiments, a multispecific protein described herein comprising an NKG2D binding domain and a binding domain for CD33 can enhance the activity of resting and IL-2-activated human NK cells in the presence of tumor cells expressing the antigen CD33.

[0125] In certain embodiments, the multispecific proteins described herein comprising an NKG2D binding domain and a binding domain for the tumor-associated antigen CD33 can enhance the cytotoxic activity of resting and IL-2-activated human NK cells in the presence of tumor cells expressing the antigen CD33. In certain embodiments, the multispecific proteins can confer an advantage against tumor cells expressing moderate and low levels of CD33 compared to the corresponding monoclonal antibodies.

[0126] In some embodiments, compared to the corresponding CD33 monoclonal antibody, the multispecific proteins described herein may be advantageous in treating cancers with high expression of Fc receptors (FcR) or cancers in tumor microenvironments with high levels of FcR. Monoclonal antibodies exert their effects on tumor growth through multiple mechanisms, including ADCC, CDC, phagocytosis, and signal blocking, among others. Among FcγRs, CD16 has the lowest affinity for IgG Fc; FcγRI (CD64) is a high-affinity FcR that binds IgG Fc about 1000 times more strongly than CD16. CD64 is normally expressed in many hematopoietic lineages, such as myeloid lineages, and can be expressed on tumors derived from these cell types, such as acute myeloid leukemia (AML). Tumor-infiltrating immune cells, such as MDSCs and monocytes, are also known to express CD64 and infiltrate the tumor microenvironment. Expression of CD64 by tumors or in the tumor microenvironment can have a detrimental effect on monoclonal antibody therapy. Because antibodies preferentially bind to high affinity receptors, expression of CD64 in the tumor microenvironment makes it difficult for these antibodies to engage with CD16 on the surface of NK cells. Multispecific proteins can overcome the detrimental effect of CD64 expression (on tumors or in the tumor microenvironment) on monoclonal antibody therapy through targeting two activating receptors on the surface of NK cells. Multispecific proteins can mediate human NK cell responses against all tumor cells, since dual targeting two activating receptors on NK cells provides stronger specific binding to NK cells, regardless of CD64 expression on tumor cells.

[0127] In some embodiments, the multispecific proteins described herein may provide a better safety profile through reduced on-target off-tumor side effects. Although both natural killer cells and CD8 T cells can directly lyse tumor cells, the mechanisms by which NK cells and CD8 T cells recognize normal self from tumor cells are different. The activity of NK cells is regulated by the balance of signals from activating (NCR, NKG2D, CD16, etc.) and inhibitory (KIR, NKG2A, etc.) receptors. The balance of these activating and inhibitory signals allows NK cells to discriminate healthy self cells from stressed, virus-infected or transformed self cells. This "built-in" mechanism of self-tolerance will help protect normal healthy tissues from NK cell responses. To extend this principle, the self-tolerance of NK cells will allow TriNKET to target antigens expressed on both self and tumor without off-tumor side effects or with an increased therapeutic window. Unlike natural killer cells, T cells require recognition of specific peptides presented by MHC molecules for activation and effector function. T cells are the primary target of immunotherapy, and many strategies have been developed to redirect T cell responses against tumors. T cell bispecifics, checkpoint inhibitors, and CAR-T cells have all been approved by the FDA, but often suffer from dose-limiting toxicity. T cell bispecifics and CAR-T cells work around the TCR-MHC recognition system by using binding domains to target antigens on the surface of tumor cells and engineered signaling domains to transmit activation signals to effector cells. Although effective in eliciting antitumor immune responses, these therapies are often associated with cytokine release syndrome (CRS) and on-target-off-tumor side effects. Multispecific proteins are unique in this context because they do not "bypass" the natural system of NK cell activation and inhibition.Instead, multispecific proteins are designed to tip the balance and provide additional activation signals to NK cells while maintaining NK tolerance to healthy self.

[0128] In some embodiments, the multispecific proteins described herein can delay tumor progression more effectively than the corresponding CD33 monoclonal antibody that contains the same CD33 binding domain. In some embodiments, the multispecific proteins described herein can be more effective against cancer metastasis than the corresponding CD33 monoclonal antibody that contains the same CD33 binding domain.

[0129] III. Therapeutic applications The present invention provides methods of treating cancer using the multispecific binding proteins described herein and / or pharmaceutical compositions described herein, which may be used to treat a wide variety of cancers that express CD33 by administering to a patient in need thereof a therapeutically effective amount of the multispecific binding proteins described herein.

[0130] The method of treatment can be characterized according to the cancer to be treated. For example, in certain embodiments, the cancer is AML, myelodysplastic syndrome, chronic myelomonocytic leukemia, myeloid blast crisis of chronic myelogenous leukemia, and ALL.

[0131] In certain other embodiments, the cancer is brain cancer, breast cancer, cervical cancer, colon cancer, colorectal cancer, endometrial cancer, esophageal cancer, leukemia, lung cancer, liver cancer, melanoma, ovarian cancer, pancreatic cancer, rectal cancer, renal cancer, gastric cancer, testicular cancer, or uterine cancer.In yet another embodiment, the cancer is selected from the group consisting of squamous cell carcinoma, adenocarcinoma, small cell carcinoma, melanoma, neuroblastoma, sarcoma (e.g., angiosarcoma or chondrosarcoma), laryngeal cancer, parotid gland cancer, biliary system cancer, thyroid cancer, acral melanoma, actinic keratosis, acute lymphocytic leukemia, acute myeloid leukemia, adenoid cystic carcinoma, adenoma, adenosarcoma, adenosquamous cell carcinoma, anal canal cancer, anal cancer, anorectal cancer, astrocytic tumor, Bartholin's gland carcinoma, basal cell carcinoma, biliary tract cancer, bone cancer, bone marrow cancer, bronchial cancer, bronchial adenocarcinoma, carcinoid, cholangiocarcinoma, chondrosarcoma, choroid plexus papilloma / carcinoma, Chronic lymphocytic leukemia, chronic myeloid leukemia, clear cell carcinoma, connective tissue cancer, cystadenoma, gastrointestinal cancer, duodenal cancer, endocrine cancer, endodermal sinus tumor, endometrial hyperplasia, endometrial stromal sarcoma, endometrioid adenocarcinoma, endothelial cell carcinoma, ependymal carcinoma, epithelial cell carcinoma, Ewing's sarcoma, orbital cancer, female genital tract cancer, focal nodular hyperplasia, gallbladder cancer, gastric pyloric cancer, gastric fundus cancer, gastric adenoma, glioblastoma, glucagonoma, cardiac cancer, hemangioblastoma, hemangioendothelioma, hemangioma, hepatic adenoma, hepatic adenomatosis, hepatobiliary cancer, hepatocellular carcinoma, Hodgkin's disease, ileal cancer, insulinoma, intraepithelial neoplasia , intraepithelial squamous cell neoplasm, intrahepatic cholangiocarcinoma, invasive squamous cell carcinoma, jejunal cancer, joint cancer, Kaposi's sarcoma, pelvic cancer, large cell carcinoma, colon cancer, leiomyosarcoma, lentigo maligna melanoma, lymphoma, male genital cancer, malignant melanoma, malignant mesothelial tumor, medulloblastoma, medulloepithelioma, meningeal cancer, mesothelial carcinoma, metastatic cancer, oral cancer, mucoepidermoid carcinoma, multiple myeloma, muscle cancer, nasal passage cancer, nervous system cancer, neuroepithelial adenocarcinoma, nodular melanoma, non-epithelial skin cancer, non-Hodgkin's lymphoma, oat cell carcinoma, oligodendroglial carcinoma, oral cancer, osteosarcoma, papillary serous adenocarcinoma, penile cancer, pharyngeal cancer, pituitary tumor, plasmacytoma, pseudosarcoma, pulmonary blastoma, rectal cancer, renal cell carcinoma, respiratory system cancer, retinoblastoma, rhabdomyosarcoma, sarcoma, serous carcinoma, paranasal sinus cancer, skin cancer, small cell carcinoma, small intestine cancer, smooth muscle carcinoma, soft tissue cancer, somatostatin-secreting tumor, spinal cancer, squamous cell carcinoma, rhabdomyosarcoma, submesothelial carcinoma, superficial malignant melanoma, T-cell leukemia, tongue cancer, undifferentiated carcinoma, ureteral cancer, urethral cancer, bladder cancer, urinary system cancer, cervical cancer, uterine cancer, uveal melanoma, vaginal cancer, verrucous carcinoma, vipoma, vulvar cancer, well-differentiated carcinoma, or Wilms' tumor.

[0132] In some other embodiments, the cancer is a non-Hodgkin's lymphoma, such as a B-cell lymphoma or a T-cell lymphoma. In some embodiments, the non-Hodgkin's lymphoma is a B-cell lymphoma, such as diffuse large B-cell lymphoma, primary mediastinal B-cell lymphoma, follicular lymphoma, small lymphocytic lymphoma, mantle cell lymphoma, marginal zone B-cell lymphoma, extranodal marginal zone B-cell lymphoma, nodal marginal zone B-cell lymphoma, splenic marginal zone B-cell lymphoma, Burkitt's lymphoma, lymphoplasmacytic lymphoma, hairy cell leukemia, or primary central nervous system (CNS) lymphoma. In certain other embodiments, the non-Hodgkin's lymphoma is a T-cell lymphoma, such as precursor T-lymphoblastic lymphoma, peripheral T-cell lymphoma, cutaneous T-cell lymphoma, angioimmunoblastic T-cell lymphoma, extranodal natural killer / T-cell lymphoma, enteropathy-type T-cell lymphoma, subcutaneous panniculitis-like T-cell lymphoma, anaplastic large cell lymphoma, or peripheral T-cell lymphoma.

[0133] The cancer to be treated can be characterized according to the presence of specific antigens expressed on the surface of cancer cells. In some embodiments, the cancer cells can express one or more of the following in addition to CD33: CD2, CD19, CD20, CD30, CD38, CD40, CD52, CD70, EGFR / ERBB1, IGF1R, HER3 / ERBB3, HER4 / ERBB4, MUC1, cMET, SLAMF7, PSCA, MICA, MICB, TRAILR1, TRAILR2, MAGE-A3, B7.1, B7.2, CTLA4, and PD1.

[0134] IV. Combination Therapy Another aspect of the invention provides combination therapy. The multispecific binding proteins described herein can be used in combination with additional therapeutic agents to treat cancer.

[0135] Exemplary therapeutic agents that may be used as part of a combination therapy in the treatment of cancer include, for example, radiation, mitomycin, tretinoin, ribomustine, gemcitabine, vincristine, etoposide, cladribine, mitobronitol, methotrexate, doxorubicin, carboquone, pentostatin, nitracrine, zinostatin, cetrorelix, letrozole, ralquitrexed, daunorubicin, fadrozole, fotemustine, thymalfasin, sobuzoxane, nedaplatin, cytarabine, bicalutamide, vinorelbine, vesnarinone, aminoglutethimide, amsacrine, proglumide, elliptinium acetate, ketanserin, doxifluridine, etretinate, isotretinoin, streptozocin, nimustine, Included are vindesine, flutamide, drogenil, butosin, carmofur, razoxane, sizofiran, carboplatin, mitolactol, tegafur, ifosfamide, prednimustine, picibanil, levamisole, teniposide, improsulfan, enocitabine, lisuride, oxymetholone, tamoxifen, progesterone, mepitiostane, epitiostanol, formestane, interferon-α, interferon-2α, interferon-β, interferon-γ, colony stimulating factor-1, colony stimulating factor-2, denileukin diftitox, interleukin-2, luteinizing hormone releasing factor, and variations of the foregoing agents that may exhibit differential binding to their cognate receptors and increased or decreased serum half-lives.

[0136] Another class of agents that may be used as part of combination therapy in the treatment of cancer are immune checkpoint inhibitors.Exemplary immune checkpoint inhibitors include agents that inhibit one or more of: (i) cytotoxic T-lymphocyte-associated antigen 4 (CTLA4), (ii) programmed cell death protein 1 (PD1), (iii) PDL1, (iv) LAG3, (v) B7-H3, (vi) B7-H4, and (vii) TIM3.The CTLA4 inhibitor ipilimumab has been approved by the US Food and Drug Administration to treat melanoma.

[0137] Still other agents that may be used as part of a combination therapy in the treatment of cancer include monoclonal antibody agents that target non-checkpoint targets (e.g., Herceptin) and non-cytotoxic agents (e.g., tyrosine kinase inhibitors).

[0138] Further categories of anticancer drugs include, for example: (i) ALK inhibitors, ATR inhibitors, A2A antagonists, base excision repair inhibitors, Bcr-Abl tyrosine kinase inhibitors, Bruton's tyrosine kinase inhibitors, CDC7 inhibitors, CHK1 inhibitors, cyclin-dependent kinase inhibitors, DNA-PK inhibitors, inhibitors of both DNA-PK and mTOR, DNMT1 inhibitors, DNMT1 inhibitors plus 2-chlorodeoxyadenosine, HDAC inhibitors, hedgehog signaling pathway inhibitors, IDO inhibitors, JAK inhibitors, mTOR inhibitors, M (ii) an inhibitor selected from an EK inhibitor, a MELK inhibitor, an MTH1 inhibitor, a PARP inhibitor, a phosphoinositide 3 kinase inhibitor, an inhibitor of both PARP1 and DHODH, a proteasome inhibitor, a topoisomerase II inhibitor, a tyrosine kinase inhibitor, a VEGFR inhibitor, and a WEE1 inhibitor; (iii) an agonist of OX40, CD137, CD40, GITR, CD27, HVEM, TNFRSF25, or ICOS; and (iii) a cytokine selected from IL-12, IL-15, GM-CSF, and G-CSF.

[0139] The proteins of the invention may also be used as an adjunct to surgical removal of the primary lesion.

[0140] The amounts of the multispecific binding protein and additional therapeutic agent, and the relative timing of administration, may be selected to achieve a desired combined therapeutic effect. For example, when administering a combination therapy to a patient in need of such administration, the combined therapeutic agents, or pharmaceutical compositions or compositions comprising these therapeutic agents, may be administered in any order, e.g., sequentially, contemporaneously, together, simultaneously, etc. Further, for example, the multispecific binding protein may be administered during the time that the additional therapeutic agent exerts its prophylactic or therapeutic effect, or vice versa.

[0141] V. Pharmaceutical Compositions The present disclosure also features a pharmaceutical composition containing a therapeutically effective amount of the protein described herein. The composition can be formulated for use in a variety of drug delivery systems. One or more physiologically acceptable excipients or carriers can also be included in the composition for proper formulation. Suitable formulations for use in the present disclosure can be found in Remington's Pharmaceutical Sciences, Mack Publishing Company, Philadelphia, Pa., 17th Edition, 1985. For a brief review of methods of drug delivery, see, e.g., Langer (Science 249:1527-1533, 1990).

[0142] The intravenous drug delivery formulation of the present disclosure may be contained in a bag, pen, or syringe. In some embodiments, the bag may be connected to a channel containing a tube and / or a needle. In some embodiments, the formulation may be a lyophilized formulation or a liquid formulation. In some embodiments, the formulation may be freeze-dried (lyophilized) and packaged in about 12 to 60 vials. In some embodiments, the formulation may be freeze-dried and 45 mg of the freeze-dried formulation may be packaged in one vial. In some embodiments, about 40 mg to about 100 mg of the freeze-dried formulation may be packaged in one vial. In some embodiments, freeze-dried formulations from 12, 27, or 45 vials are combined to obtain a therapeutic dose of protein in an intravenous drug formulation. In some embodiments, the formulation may be a liquid formulation and may be stored at about 250 mg / vial to about 1000 mg / vial. In some embodiments, the formulation may be a liquid formulation and may be stored at about 600 mg / vial. In certain embodiments, the formulation may be a liquid formulation and may be stored as about 250 mg / vial.

[0143] The present disclosure may be present in a liquid, aqueous pharmaceutical formulation that includes a therapeutically effective amount of protein in a buffer solution that forms the formulation.

[0144] These compositions may be sterilized by conventional sterilization techniques or may be sterile filtered. The resulting aqueous solutions may be packaged for use as is or lyophilized, with the lyophilized preparation being combined with a sterile aqueous carrier prior to administration. The pH of the preparation will typically be between 3 and 11, more preferably between 5 and 9 or between 6 and 8, most preferably between 7 and 8, for example 7 to 7.5. The resulting solid form composition may be packaged into a number of single dose units, each containing a fixed amount of the agent(s) described above. The solid form composition may also be packaged in a container for mobile amounts.

[0145] In certain embodiments, the disclosure provides a formulation having an extended shelf life comprising a protein of the disclosure in combination with mannitol, citric acid monohydrate, sodium citrate, disodium phosphate dihydrate, sodium dihydrogen phosphate dihydrate, sodium chloride, polysorbate 80, water, and sodium hydroxide.

[0146] In some embodiments, aqueous formulations are prepared containing the protein of the present disclosure in a pH buffer solution. The buffer of the present invention may have a pH in the range of about 4 to about 8, for example, about 4.5 to about 6.0, or about 4.8 to about 5.5, or may have a pH of about 5.0 to about 5.2. Intermediate pH ranges are also intended to be part of the present disclosure. For example, a range of values ​​using any combination of the values ​​listed above as upper and / or lower limits is intended to be included. Examples of buffers that control the pH within this range include acetate (e.g., sodium acetate), succinate (such as sodium succinate), gluconate, histidine, citrate, and other organic acid buffers.

[0147] In some embodiments, the formulation includes a buffer system containing citrate and phosphate to maintain a pH in the range of about 4 to about 8. In some embodiments, the pH range may be about 4.5 to about 6.0, or about pH 4.8 to about 5.5, or may be in the pH range of about 5.0 to about 5.2. In some embodiments, the buffer system includes citric acid monohydrate, sodium citrate, disodium phosphate dihydrate, and / or sodium dihydrogen phosphate dihydrate. In some embodiments, the buffer system includes about 1.3 mg / ml citric acid (e.g., 1.305 mg / ml), about 0.3 mg / ml sodium citrate (e.g., 0.305 mg / ml), about 1.5 mg / ml disodium phosphate dihydrate (e.g., 1.53 mg / ml), about 0.9 mg / ml sodium dihydrogen phosphate dihydrate (e.g., 0.86), and about 6.2 mg / ml sodium chloride (e.g., 6.165 mg / ml). In some embodiments, the buffer system comprises 1-1.5 mg / ml citric acid, 0.25-0.5 mg / ml sodium citrate, 1.25-1.75 mg / ml disodium phosphate dihydrate, 0.7-1.1 mg / ml sodium dihydrogen phosphate dihydrate, and 6.0-6.4 mg / ml sodium chloride. In some embodiments, the pH of the formulation is adjusted with sodium hydroxide.

[0148] A polyol may also be included in the formulation, which may act as an isotonic agent and stabilize the antibody. The polyol is added to the formulation in an amount that may vary with respect to the desired isotonicity of the formulation. In certain embodiments, the aqueous formulation may be isotonic. The amount of polyol added may vary with respect to the molecular weight of the polyol. For example, a lesser amount of a monosaccharide (e.g., mannitol) may be added as compared to a disaccharide (such as trehalose). In certain embodiments, the polyol that may be used in the formulation as an isotonic agent is mannitol. In certain embodiments, the mannitol concentration may be about 5 to about 20 mg / ml. In certain embodiments, the mannitol concentration may be about 7.5 to 15 mg / ml. In certain embodiments, the mannitol concentration may be about 10 to 14 mg / ml. In certain embodiments, the mannitol concentration may be about 12 mg / ml. In certain embodiments, the polyol, sorbitol, may be included in the formulation.

[0149] Detergents or surfactants may also be added to the formulation. Exemplary detergents include non-ionic detergents such as polysorbates (e.g., polysorbate 20, 80, etc.) or poloxamers (e.g., poloxamer 188). The amount of detergent added is such that it reduces the aggregation of the formulated antibody and / or minimizes the formation of particles in the formulation and / or reduces adsorption. In some embodiments, the formulation may include a surfactant that is a polysorbate. In some embodiments, the formulation may contain a detergent, polysorbate 80 or Tween 80. Tween 80 is a term used to describe polyoxyethylene (20) sorbitan monooleate (see Fiedler, Lexikon der Hifsstoffe, Editio Cantor Verlag Aulendorf, 4th ed., 1996). In certain embodiments, the formulation may contain between about 0.1 mg / mL and about 10 mg / mL, or between about 0.5 mg / mL and about 5 mg / mL of polysorbate 80. In certain embodiments, about 0.1% polysorbate 80 may be added to the formulation.

[0150] In one embodiment, the protein product of the present disclosure is formulated as a liquid formulation. The liquid formulation may be presented at 10 mg / mL in a USP / Ph Eur type I 50R vial closed with a rubber stopper and sealed with an aluminum crimp seal closure. The stopper may be made of an elastomer conforming to USP and Ph Eur. In one embodiment, the vial may be filled with 61.2 mL of protein product solution to achieve an extractable volume of 60 mL. In one embodiment, the liquid formulation may be diluted with 0.9% saline solution.

[0151] In some embodiments, the liquid formulation of the present disclosure may be prepared as a 10 mg / mL solution in combination with a stabilizing level of sugar.In some embodiments, the liquid formulation may be prepared in an aqueous carrier.In some embodiments, a stabilizer may be added in an amount that does not exceed the amount that may result in an undesirable or unsuitable viscosity for intravenous administration.In some embodiments, the sugar may be a disaccharide, such as sucrose.In some embodiments, the liquid formulation may also include one or more buffers, surfactants, and preservatives.

[0152] In some embodiments, the pH of the liquid formulation may be set by the addition of a pharma- ceutically acceptable acid and / or base. In some embodiments, the pharma-ceutically acceptable acid may be hydrochloric acid. In some embodiments, the base may be sodium hydroxide.

[0153] In addition to aggregation, deamidation is a common product variant of peptides and proteins that can occur during fermentation, harvest / cell clarification, purification, drug substance / drug product storage, and during sample analysis. Deamidation is the release of NH from proteins forming a succinimide intermediate that can undergo hydrolysis. 3The succinimide intermediate results in a mass reduction of 17 Daltons from the parent peptide. Subsequent hydrolysis results in a mass increase of 18 Daltons. Isolation of the succinimide intermediate is difficult due to its instability under aqueous conditions. As such, deamidation is typically detectable as a mass increase of 1 Dalton. Deamidation of asparagine results in either aspartic acid or isoaspartic acid. Parameters that affect the rate of deamidation include pH, temperature, solvent dielectric constant, ionic strength, primary sequence, local polypeptide conformation, and tertiary structure. The amino acid residue adjacent to Asn in the peptide chain affects the deamidation rate. Gly and Ser following Asn in the protein sequence result in a higher susceptibility to deamidation.

[0154] In certain embodiments, the liquid formulations of the present disclosure may be stored under pH and humidity conditions to prevent deamination of the protein product.

[0155] Aqueous carriers of interest herein are those that are pharma- ceutically acceptable (safe and non-toxic for administration to humans) and useful for preparing liquid formulations. Illustrative carriers include sterile water for injection (SWFI), bacteriostatic water for injection (BWFI), a pH buffered solution (e.g., phosphate buffered saline), sterile saline solution, Ringer's solution, or dextrose solution.

[0156] Preservatives may optionally be added to the formulations herein to reduce bacterial action. The addition of a preservative may, for example, facilitate the manufacture of multi-use (multiple dose) formulations.

[0157] Intravenous (IV) formulations may be the preferred route of administration in certain cases, such as when a patient is in hospital after transplant and receives all medications via IV route.In some embodiments, the liquid formulation is diluted with 0.9% sodium chloride solution before administration.In some embodiments, the diluted drug product for injection is isotonic and suitable for administration by intravenous infusion.

[0158] In some embodiments, the salt or buffer component may be added in an amount of 10 mM to 200 mM. The salts and / or buffers are pharma- citrate, phosphate buffer, carbonate, citrate buffer, in which case sodium, potassium or ammonium ions serve as counterions.

[0159] Preservatives may optionally be added to the formulations herein to reduce bacterial action. The addition of a preservative may, for example, facilitate the manufacture of multi-use (multiple dose) formulations.

[0160] Aqueous carriers of interest herein are those that are pharma- ceutically acceptable (safe and non-toxic for administration to humans) and useful for preparing liquid formulations. Illustrative carriers include sterile water for injection (SWFI), bacteriostatic water for injection (BWFI), a pH buffered solution (e.g., phosphate buffered saline), sterile saline solution, Ringer's solution, or dextrose solution.

[0161] The present disclosure may be in a lyophilized formulation comprising a protein and a lyoprotectant. The lyoprotectant may be a sugar, such as a disaccharide. In some embodiments, the lyoprotectant may be sucrose or maltose. The lyophilized formulation may also include one or more of a buffer, a surfactant, a bulking agent, and / or a preservative.

[0162] The amount of sucrose or maltose useful for stabilizing the lyophilized drug product may be a weight ratio of protein to sucrose or maltose of at least 1:2. In some embodiments, the weight ratio of protein to sucrose or maltose may be between 1:2 and 1:5.

[0163] In some embodiments, the pH of the formulation may be set by the addition of a pharma- ceutically acceptable acid and / or base prior to lyophilization. In some embodiments, the pharma-ceutically acceptable acid may be hydrochloric acid. In some embodiments, the pharma-ceutically acceptable base may be sodium hydroxide.

[0164] Prior to lyophilization, the pH of the solution containing the protein of the present disclosure may be adjusted to between 6 and 8. In some embodiments, the pH range of the lyophilized drug product may be between 7 and 8.

[0165] In some embodiments, the salt or buffer component may be added in an amount of 10 mM to 200 mM. The salts and / or buffers are pharma- citrate, phosphate buffer, carbonate, citrate buffer, in which case sodium, potassium or ammonium ions serve as counterions.

[0166] In some embodiments, a "bulking agent" may be added. A "bulking agent" is a compound that adds mass to the lyophilization mixture and contributes to the physical structure of the lyophilized cake (e.g., facilitating the production of an essentially uniform lyophilized cake that maintains an open pore structure). Illustrative bulking agents include mannitol, glycine, polyethylene glycol, and sorbitol. The lyophilized formulations of the present invention may contain such bulking agents.

[0167] Preservatives may optionally be added to the formulations herein to reduce bacterial action. The addition of a preservative may, for example, facilitate the manufacture of multi-use (multiple dose) formulations.

[0168] In some embodiments, lyophilized drug product may be composed of aqueous carrier.Aqueous carrier of interest herein is medicamentously acceptable (safe and non-toxic for administration to humans) and useful for preparing liquid formulation after lyophilization.Illustrative diluents include sterile water for injection (SWFI), bacteriostatic water for injection (BWFI), pH buffer solution (e.g. phosphate buffered saline), sterile saline solution, Ringer's solution, or dextrose solution.

[0169] In certain embodiments, the lyophilized drug product of the current disclosure is reconstituted with either Sterile Water for Injection, USP (SWFI), or 0.9% Sodium Chloride Injection, USP. During reconstitution, the lyophilized powder dissolves into solution.

[0170] In certain embodiments, the lyophilized protein product of the present disclosure is constituted in about 4.5 mL of water for injection and diluted with 0.9% saline solution (sodium chloride solution).

[0171] Actual dosage levels of the active ingredients in pharmaceutical compositions of the present invention may be varied to obtain an amount of the active ingredient effective to achieve the desired therapeutic response for a particular patient, composition, and mode of administration, without toxicity to the patient.

[0172] The specific dose can be a uniform dose for each patient, for example, 50-5000 mg of protein. Alternatively, the patient's dose can be adapted to the approximate body weight or surface area of ​​the patient. Other factors in determining the appropriate dosing include the disease or condition to be treated or prevented, the severity of the disease, the route of administration, and the age, sex, and medical condition of the patient. Furthermore, refinement of the calculations required to determine the appropriate dosing for treatment is routinely performed by those skilled in the art, especially in light of the dosing information and assays disclosed herein. Dosing can also be determined through the use of known assays for determining dosing used together with appropriate dose-response data. Dosing of an individual patient can be adjusted as disease progression is monitored. Blood levels of the targetable construct or complex in the patient can be measured to see if dosing needs to be adjusted to reach or maintain an effective concentration. Pharmacogenomics may be used to determine which targetable constructs and / or complexes, and their dosages, are most likely to be effective for a given individual (Schmitz et al., Clinica Chimica Acta 308: 43-53, 2001; Steimer et al., Clinica Chimica Acta 308: 33-41, 2001).

[0173] Generally, the dosage based on body weight is about 0.01 μg to about 100 mg per kg of body weight, for example, about 0.01 μg to about 100 mg / kg body weight, about 0.01 μg to about 50 mg / kg body weight, about 0.01 μg to about 10 mg / kg body weight, about 0.01 μg to about 1 mg / kg body weight, about 0.01 μg to about 100 μg / kg body weight, about 0.01 μg to about 50 μg / kg body weight, about 0.01 μg to about 10 μg / kg body weight, about 0.01 μg to about 1 μg / kg body weight, about 0.01 μg to about 0.1 μg / kg body weight, about 0.1 μg to about 100 mg / kg body weight, about 0.1 μg to about 50 mg / kg body weight, about 0.1 μg to about 10 mg / kg body weight, about 0.1 μg to about 1 mg / kg body weight, about 0.1 μg to about 100 μg / kg body weight, about 0.1 μg to about 10 μg / kg body weight, about 0.1 μg to about 1 μg / kg body weight, about 1 μg to about 100 mg / kg body weight, about 1 μg to about 50 mg / kg body weight, about 1 μg to about 10 mg / kg body weight, about 1 μg to about 1 mg / kg body weight, about 1 μg to about 100 μg / kg body weight, about 1 μg to about 50 μg / kg body weight, about 1 μg to about 10 μg / kg body weight, about 10 μg to about 100 mg / kg body weight, about 10 μg to about 50 mg / kg body weight, about 10 μg to about 10 mg / kg body weight, about 10 μg to about 1 mg / kg body weight, about 10 μg to about 100 μg / kg body weight, about 10 μg to about 50 μg / kg body weight, about 50 μg to about 100 mg / kg body weight, about 50 μg to about 50 mg / kg body weight, about 50 μg to about 10 mg / kg body weight, about 50 μg to about 1 mg / kg body weight, about 50 μg to about 100 μg / kg body weight, about 100 μg to about 100 mg / kg body weight, about 100 μg to about 50 mg / kg body weight, about 100 μg to about 10 mg / kg body weight, about 100 μg to about 1 mg / kg body weight, about 1 mg to about 100 mg / kg body weight, about 1 mg to about 50 mg / kg body weight, about 1 mg to about 10 mg / kg body weight, about 10 mg to about 100 mg / kg body weight, about 10 mg to about 50 mg / kg body weight, about 50 mg to about 100 mg / kg body weight.

[0174] Doses may be given one or more times daily, weekly, monthly, or yearly, or once every 2-20 years. One of skill in the art can easily estimate dosing repetition rates based on measured residence time and concentration of the targetable construct or complex in bodily fluids or tissues. Administration of the present invention may be intravenous, intraarterial, intraperitoneal, intramuscular, subcutaneous, intrapleural, intrathecal, intracavity, by perfusion via a catheter, or by direct intralesional injection. It may be administered one or more times daily, one or more times weekly, one or more times monthly, and one or more times yearly.

[0175] The above description describes multiple aspects and embodiments of the present invention. This patent specification specifically contemplates all combinations and permutations of these aspects and embodiments. EXAMPLES

[0176] (Example) The invention, generally described herein, will be more readily understood by reference to the following examples, which are included solely for purposes of illustrating certain aspects and embodiments of the invention and are not intended to limit the invention.

[0177] Example 1: NKG2D-binding domain binds to NKG2D The NKG2D-binding domain binds to purified recombinant NKG2D Nucleic acid sequences of human, mouse or cynomolgus NKG2D ectodomains were fused to nucleic acid sequences encoding human IgG1 Fc domains and introduced into mammalian cells to be expressed. After purification, NKG2D-Fc fusion proteins were adsorbed to wells of microplates. After blocking the wells with bovine serum albumin to prevent non-specific binding, NKG2D binding domains were titrated and added to wells pre-adsorbed with NKG2D-Fc fusion proteins. Primary antibody binding was detected using a secondary antibody conjugated with horseradish peroxidase that specifically recognizes human kappa light chains and avoids Fc cross-reactivity. Binding signals were visualized by adding 3,3',5,5'-tetramethylbenzidine (TMB), a substrate for horseradish peroxidase, to the wells, and the absorbance was measured at 450 nM and corrected at 540 nM. NKG2D binding domain clones, isotype controls or positive controls (SEQ ID NOs: 45-48, or anti-mouse NKG2D clones MI-6 and CX-5 available at eBioscience) were added to each well.

[0178] The isotype control showed minimal binding to recombinant NKG2D-Fc protein, whereas the positive control bound most strongly to the recombinant antigen. The NKG2D binding domains produced by all clones were demonstrated to cross-bind human, mouse, and cynomolgus monkey recombinant NKG2D-Fc proteins, with varying affinities from clone to clone. In general, each anti-NKG2D clone bound human (Figure 3) and cynomolgus monkey (Figure 4) recombinant NKG2D-Fc with similar affinity, but bound mouse (Figure 5) recombinant NKG2D-Fc with lower affinity.

[0179] The NKG2D-binding domain binds to cells expressing NKG2D EL4 mouse lymphoma cell lines were engineered to express human or mouse NKG2D-CD3 zeta signaling domain chimeric antigen receptors. NKG2D-binding clones, isotype controls or positive controls were used at 100 nM concentration to stain extracellular NKG2D expressed on EL4 cells. Antibody binding was detected using a fluorophore-conjugated anti-human IgG secondary antibody. Cells were analyzed by flow cytometry and fold over background (FOB) was calculated using the mean fluorescence intensity (MFI) of NKG2D-expressing cells compared to parental EL4 cells.

[0180] The NKG2D binding domains produced by all clones bound to EL4 cells expressing human and mouse NKG2D. Positive control antibodies (selected from SEQ ID NOs: 45-48, or anti-mouse NKG2D clones MI-6 and CX-5 available at eBioscience) gave the best FOB binding signal. The NKG2D binding affinity for each clone was similar between cells expressing human NKG2D (Figure 6) and mouse NKG2D (Figure 7).

[0181] Example 2: NKG2D-binding domains block natural ligand binding to NKG2D Competition with ULBP-6 Recombinant human NKG2D-Fc protein was adsorbed to the wells of a microplate, and the wells were blocked with bovine serum albumin to reduce non-specific binding. A saturating concentration of ULBP-6-His-biotin was added to the wells, followed by the addition of the NKG2D-binding domain clone. After 2 hours of incubation, the wells were washed, and the remaining ULBP-6-His-biotin bound to the NKG2D-Fc-coated wells was detected by streptavidin-conjugated horseradish peroxidase and TMB substrate. Absorbance was measured at 450 nM and corrected at 540 nM. After background subtraction, the specific binding of the NKG2D-binding domain to the NKG2D-Fc protein was calculated from the percentage of ULBP-6-His-biotin blocked from binding to the NKG2D-Fc protein in the wells. A positive control antibody (selected from SEQ ID NOs: 45-48) and various NKG2D binding domains blocked ULBP-6 binding to NKG2D, whereas an isotype control showed little competition with ULBP-6 (Figure 8).

[0182] Conflict with MICA Recombinant human MICA-Fc protein was adsorbed to the wells of a microplate, and the wells were blocked with bovine serum albumin to reduce non-specific binding. NKG2D-Fc-biotin was added to the wells, followed by the NKG2D binding domain. After incubation and washing, the remaining NKG2D-Fc-biotin bound to the MICA-Fc-coated wells was detected using streptavidin-HRP and TMB substrate. Absorbance was measured at 450 nM and corrected at 540 nM. After background subtraction, the specific binding of the NKG2D binding domain to the NKG2D-Fc protein was calculated from the percentage of NKG2D-Fc-biotin blocked from binding to the MICA-Fc-coated wells. A positive control antibody (selected from SEQ ID NOs: 45-48) and various NKG2D binding domains blocked MICA binding to NKG2D, whereas an isotype control showed little competition with MICA (FIG. 9).

[0183] Conflict with Rae-1 Delta Recombinant mouse Rae-1 delta-Fc (purchased from R&D Systems) was adsorbed to the wells of a microplate, and the wells were blocked with bovine serum albumin to reduce non-specific binding. Mouse NKG2D-Fc-biotin was added to the wells, followed by the addition of the NKG2D binding domain. After incubation and washing, the remaining NKG2D-Fc-biotin bound to the Rae-1 delta-Fc coated wells was detected using streptavidin-HRP and TMB substrate. Absorbance was measured at 450 nM and corrected at 540 nM. After background subtraction, the specific binding of the NKG2D binding domain to the NKG2D-Fc protein was calculated from the percentage of NKG2D-Fc-biotin blocked from binding to the Rae-1 delta-Fc coated wells. Positive controls (selected from SEQ ID NOs: 45-48, or anti-mouse NKG2D clones MI-6 and CX-5 available at eBioscience) and various NKG2D binding domain clones blocked Rae-1 delta binding to mouse NKG2D, whereas isotype control antibodies showed little competition with Rae-1 delta (Figure 10).

[0184] Example 3: NKG2D-binding domain clones activate NKG2D The nucleic acid sequences of human and mouse NKG2D were fused to a nucleic acid sequence encoding the CD3 zeta signaling domain to obtain a chimeric antigen receptor (CAR) construct. The NKG2D-CAR construct was then cloned into a retroviral vector using Gibson assembly and transfected into expi293 cells for retroviral production. EL4 cells were infected with a virus containing NKG2D-CAR together with 8 μg / mL polybrene. 24 hours after infection, the expression level of NKG2D-CAR in EL4 cells was analyzed by flow cytometry, and clones expressing high levels of NKG2D-CAR on the cell surface were selected.

[0185] To determine whether the NKG2D binding domains activate NKG2D, they were adsorbed to microplate wells and NKG2D-CAR EL4 cells were cultured in the presence of brefeldin-A and monensin in wells coated with the antibody fragments for 4 hours. Intracellular TNF-α production, an indicator for NKG2D activation, was assayed by flow cytometry. The percentage of TNF-α positive cells was normalized to cells treated with a positive control. All NKG2D binding domains activated both human NKG2D (Figure 11) and mouse NKG2D (Figure 12).

[0186] Example 4: NKG2D binding domains activate NK cells Primary human NK cells Peripheral blood mononuclear cells (PBMCs) were isolated from human peripheral blood buffy coats using density gradient centrifugation. NK cells (CD3 - CD56 + ) were isolated from PBMCs using negative selection with magnetic beads, and the purity of isolated NK cells was typically >95%. Isolated NK cells were then cultured in medium containing 100 ng / mL IL-2 for 24-48 hours, after which they were transferred to wells of a microplate adsorbed with NKG2D binding domains and cultured in medium containing fluorophore-conjugated anti-CD107a antibody, brefeldin-A, and monensin. Following culture, NK cells were assayed by flow cytometry using fluorophore-conjugated antibodies against CD3, CD56, and IFN-γ. CD107a and IFN-γ staining was measured using CD3 - CD56 + NK cell activation was assessed by analyzing CD107a / IFN-γ double positive cells. An increase in CD107a / IFN-γ double positive cells is an indication of better NK cell activation through engagement of two activating receptors rather than one receptor. The NKG2D binding domain and positive control (selected from SEQ ID NOs: 45-48) showed that a higher percentage of NK cells express CD107a than the isotype control. + and IFN-γ+ (FIGS. 13 and 14 represent data from two independent experiments, each of which used PBMCs from a different donor for NK cell preparation).

[0187] Primary mouse NK cells Spleens were obtained from C57Bl / 6 mice and broken through a 70 μm cell strainer to obtain a single cell suspension. Cells were pelleted and resuspended in ACK lysis buffer (purchased from Thermo Fisher Scientific #A1049201; 155 mM ammonium chloride, 10 mM potassium bicarbonate, 0.01 mM EDTA) to remove red blood cells. The remaining cells were cultured with 100 ng / mL hIL-2 for 72 hours before being harvested and prepared for NK cell isolation. NK cells (CD3 - NK1.1 + ) were then isolated from spleen cells using a negative depletion technique with magnetic beads and were typically >90% pure. Purified NK cells were cultured for 48 hours in medium containing 100 ng / mL mIL-15, then transferred to wells of microplates adsorbed with NKG2D binding domains and cultured in medium containing fluorophore-conjugated anti-CD107a antibodies, brefeldin-A, and monensin. After culture in wells coated with NKG2D binding domains, NK cells were assayed by flow cytometry using fluorophore-conjugated antibodies against CD3, NK1.1, and IFN-γ. CD107a and IFN-γ staining was measured using CD3 - NK1.1 + NK cell activation was assessed by analysis of IFN-γ double positive cells. An increase in CD107a / IFN-γ double positive cells is an indication of better NK cell activation through engagement of two activating receptors rather than one. NKG2D binding domains and positive controls (selected from anti-mouse NKG2D clones MI-6 and CX-5 available at eBioscience) showed that a higher percentage of NK cells express CD107a than the isotype control. + and IFN-γ +(FIGS. 15 and 16 represent data from two independent experiments, each of which used different mice for NK cell preparation).

[0188] Example 5: NKG2D-binding domains enable cytotoxicity of target tumor cells Human and mouse primary NK cell activation assays demonstrate increased cytotoxicity markers on NK cells after incubation with the NKG2D binding domain. To examine whether this translates into increased tumor cell lysis, a cell-based assay was utilized in which each NKG2D binding domain was developed into a monospecific antibody. The Fc region was used as one targeting arm, while the Fab region (NKG2D binding domain) acted as another targeting arm to activate NK cells. THP-1 cells, which are of human origin and express high levels of Fc receptors, were used as tumor targets, using the Perkin Elmer DELFIA Cytotoxicity Kit. THP-1 cells were labeled with BATDA reagent and incubated with 10 5 The labeled THP-1 cells were then combined with NKG2D antibodies and isolated mouse NK cells in wells of a microtiter plate for 3 hours at 37°C. After incubation, 20 μl of culture supernatant was taken, mixed with 200 μl of europium solution and incubated for 15 minutes with shaking, protected from light. Fluorescence was measured over time by a PheraStar plate reader equipped with a time-resolved fluorescence module (excitation 337 nm, emission 620 nm) and specific lysis was calculated according to the kit instructions.

[0189] The positive control, ULBP-6, the natural ligand for NKG2D, showed increased specific lysis of THP-1 target cells by mouse NK cells. NKG2D antibodies also increased specific lysis of THP-1 target cells, whereas an isotype control antibody showed reduced specific lysis. The dotted line indicates specific lysis of THP-1 cells by mouse NK cells without added antibody (Figure 17).

[0190] Example 6: NKG2D antibodies exhibit high thermal stability The melting temperature of the NKG2D binding domain was assayed using differential scanning fluorimetry, and the extrapolated apparent melting temperature is relatively high for a typical IgG1 antibody (Figure 18).

[0191] Example 7: Synergistic activation of human NK cells by cross-linking NKG2D and CD16 Primary human NK cell activation assay Peripheral blood mononuclear cells (PBMCs) were isolated from human peripheral blood buffy coats using density gradient centrifugation. NK cells were purified from PBMCs using negative magnetic beads (StemCell # 17955). NK cells were >90% CD3 - CD56 + The cells were then expanded in medium containing 100 ng / mL hIL-2 (Peprotech #200-02) for 48 hours before being used for activation assays. Antibodies were coated onto 96-well flat-bottom plates at concentrations of 2 μg / ml (anti-CD16, Biolegend # 302013) and 5 μg / mL (anti-NKG2D, R&D #MAB139) in 100 μl sterile PBS overnight at 4°C, followed by extensive washing of wells to remove excess antibody. For assessment of degranulation, IL-2-activated NK cells were cultured at 5×10 5 Cells / ml were resuspended in medium supplemented with 100 ng / mL hIL2 and 1 μg / mL APC-conjugated anti-CD107a mAb (Biolegend # 328619). 5 Cells / well were then added to the antibody-coated plates. Protein transport inhibitors Brefeldin A (BFA, Biolegend # 420601) and Monensin (Biolegend # 420701) were added at final dilutions of 1:1000 and 1:270, respectively. Plated cells were incubated at 4°C for 24 hours at 37 °C for 1 h at 4 °C for 3 h at 4 °C. 2The cells were incubated in 5% CO at 37°C for 4 hours. For intracellular staining of IFN-γ, NK cells were labeled with anti-CD3 (Biolegend #300452) and anti-CD56 mAb (Biolegend #318328) and then fixed, permeabilized, and labeled with anti-IFN-γ mAb (Biolegend #506507). NK cells were stained with live CD56 + CD3 - After gating on cells, they were analyzed for CD107a and IFN-γ expression by flow cytometry.

[0192] To examine the relative potency of receptor combinations, cross-linking of NKG2D or CD16 by plate-bound stimulation, and co-cross-linking of both receptors were performed. As shown in Figure 19 (Figures 19A-19C), combined stimulation of CD16 and NKG2D resulted in highly elevated levels of CD107a (degranulation) (Figure 19A) and / or IFN-γ production (Figure 19B). The dotted lines represent the additive effect of stimulation of each receptor individually.

[0193] CD107a levels and intracellular IFN-γ production of IL-2-activated NK cells were analyzed after 4 hours of plate-bound stimulation with anti-CD16, anti-NKG2D, or a combination of both monoclonal antibodies. Graphs show the mean (n=2) ± SD. Figure 19A demonstrates CD107a levels; Figure 19B demonstrates IFNγ levels; Figure 19C demonstrates CD107a and IFNγ levels. Data shown in Figures 19A-19C are representative of five independent experiments using five different healthy donors.

[0194] Example 8: Multispecific binding proteins bind to NKG2D The EL4 mouse lymphoma cell line was engineered to express human NKG2D. Trispecific binding proteins (TriNKET), each containing an NKG2D-binding domain, a tumor-associated antigen-binding domain (CD33-binding domain), and an Fc domain that binds CD16 as shown in FIG. 1, were tested for their affinity to extracellular NKG2D expressed on EL4 cells. Binding of the multispecific binding proteins to NKG2D was detected using a fluorophore-conjugated anti-human IgG secondary antibody. Cells were analyzed by flow cytometry, and fold over background (FOB) was calculated using the mean fluorescence intensity (MFI) of NKG2D-expressing cells compared to parental EL4 cells.

[0195] The TriNKETs tested included CD33-TriNKET-C26 (ADI-28226 and CD33 binding domain), CD33-TriNKET-F04 (ADI-29404 and CD33 binding domain), CD33-TriNKET-A44 (ADI-27744 and CD33 binding domain), CD33-TriNKET-F47 (ADI-29447 and CD33 binding domain), CD33-TriNKET-A49 (ADI-27749 and CD33 binding domain) and CD33-TriNKET-F63 (ADI-27463 and CD33 binding domain). The CD33 binding domains used in the tested molecules were composed of the heavy and light chain variable domains listed below. CD33 Heavy Chain Variable Domain (SEQ ID NO:125): [ka] CD33 Light Chain Variable Domain (SEQ ID NO:126): [ka]

[0196] The data show that TriNKET, which contains a CD33-binding domain and an NKG2D-binding domain, binds to NKG2D (Figures 35A-35B). Figure 35A shows the binding of TriNKET in comparison to a monoclonal antibody containing the corresponding NKG2D-binding domain. Figure 35B shows the binding profile of TriNKET targeting CD33, which contains six different NKG2D-binding domains.

[0197] Example 9: Multispecific binding proteins bind to human tumor antigens Trispecific binding proteins bind to CD33 The CD33-expressing human AML cell lines MV4-11 and Molm-13 were used to assay the binding of TriNKET to the tumor-associated antigen CD33. TriNKET and, in some cases, parental anti-CD33 monoclonal antibodies were incubated with the cells, and binding was detected using a fluorophore-conjugated anti-human IgG secondary antibody. Cells were analyzed by flow cytometry, and fold over background (FOB) was calculated using the mean fluorescence intensity (MFI) from TriNKET and parental monoclonal anti-CD33 antibodies normalized to secondary antibody control. TriNKET, which targets CD33, shows comparable levels of binding to CD33 compared to parental anti-CD33 antibodies (Figure 36). TriNKET shows binding to cell surface CD33 on MV4-11 (Figures 36A, 36B, and 36D) and Molm-13 cells (Figure 36C). The overall binding signal is comparable between TriNKETs, as they contain the same CD33 binding domain.

[0198] Example 10: Multispecific binding proteins activate NK cells Peripheral blood mononuclear cells (PBMCs) were isolated from human peripheral blood buffy coats using density gradient centrifugation. NK cells (CD3 - CD56 +) were isolated from PBMCs using negative selection with magnetic beads, and the purity of isolated NK cells was typically >90%. Isolated NK cells were either cultured in medium containing 100 ng / mL IL-2 for activation or rested overnight without cytokines. IL-2-activated NK cells were used within 24–48 h after activation.

[0199] CD33-expressing MV4-11 cells were harvested and cultured at 2 × 10 6 CD33 monoclonal antibodies or TriNKET targeting CD33 were diluted in culture medium. Activated NK cells were harvested, washed, and resuspended at 2 × 10 6 The tumor cells were then resuspended at 100 / mL. The cancer cells were then mixed with monoclonal antibody / TriNKET and activated NK cells in the presence of IL-2. Brefeldin-A and monensin were also added to the mixed culture to block protein transport out of the cell for intracellular cytokine staining. Fluorophore-conjugated anti-CD107a was added to the mixed culture and the culture was incubated for 4 hours before preparing samples for FACS analysis using fluorophore-conjugated antibodies against CD3, CD56 and IFN-γ. CD107a and IFN-γ staining was performed using CD3 - CD56 + NK cell activation was assessed by analysis of CD107a / IFN-γ double positive cells, which is an indication of better NK cell activation through engagement of two activating receptors rather than one.

[0200] Co-culture of primary human NK cells with CD33-positive MV4-11 cells resulted in TriNKET-mediated activation of primary human NK cells. TriNKET targeting CD33 mediated activation of human NK cells co-cultured with MV4-11 cells, as shown by increased CD107a degranulation and IFNγ cytokine production (Figures 37A and 37B). NK cells alone, NK cells co-cultured with MV4-11 cells without TriNKET, and TriNKET targeting CD20 were used as controls. Compared to C33 monoclonal antibody, TriNKET targeting CD33 showed increased NK cell activity (Figures 37A and 37B).

[0201] Example 11: Trispecific binding proteins enable cytotoxicity of targeted cancer cells Peripheral blood mononuclear cells (PBMCs) were isolated from human peripheral blood buffy coats using density gradient centrifugation. NK cells (CD3 - CD56 + ) were isolated from PBMCs using negative selection with magnetic beads, and the purity of isolated NK cells was typically >90%. Isolated NK cells were either cultured in medium containing 100 ng / mL IL-2 for activation or rested overnight without cytokines. IL-2-activated or resting NK cells were used for cytotoxicity assays the following day.

[0202] DELFIA Cytotoxicity Assay: Human cancer cell lines expressing CD33 were harvested from culture, the cells were washed with PBS, and incubated for 10 min for labeling with BATDA reagent (Perkin Elmer AD0116). 6 The cells were resuspended in growth medium at 0.5–1.0 × 10 / mL. The manufacturer's instructions were followed for labeling of target cells. After labeling, cells were washed three times with PBS and resuspended at 0.5–1.0 × 10 5The cells were resuspended in medium at 10 / mL. To prepare background wells, an aliquot of the labeled cells was set aside and the cells were centrifuged from the medium. 100 μl of medium was added to triplicate wells, carefully avoiding disturbing the pelleted cells. 100 μl of BATDA-labeled cells were added to each well of a 96-well plate. Wells were reserved for accidental release from the target cells and wells were prepared for maximum lysis of the target cells by the addition of 1% Triton-X. Monoclonal antibodies against the tumor target of interest or TriNKET were diluted in medium and 50 μl of the diluted mAb or TriNKET was added to each well. Resting and / or activated NK cells were harvested from the culture, the cells were washed and 10 μl of the diluted mAb or TriNKET was added to each well depending on the desired effector to target cell (E:T) ratio. 5 ~2.0×10 6 NK cells were resuspended at 100 / mL. 50 μl of NK cells were added to each well of the plate, bringing the media volume to 200 μl in total. Plates were incubated at 37°C with 5% CO2 for 2-3 hours before the assay was developed.

[0203] After 2-3 hours of incubation, plates were removed from the incubator and cells were pelleted by centrifugation at 200g for 5 minutes. 20 μl of culture supernatant was transferred to a clean microplate provided by the manufacturer and 200 μl of room temperature europium solution was added to each well. Plates were incubated for 15 minutes on a plate shaker at 250 rpm, protected from light. Plates were read using either a Victor 3 or SpectraMax i3X instrument. % specific lysis was calculated as follows: % specific lysis = ((experimental release - accidental release) / (maximum release - accidental release)) * 100%.

[0204] TriNKET mediated the cytotoxicity of human NK cells against CD33-positive human cancer cell lines. Resting human NK cells were mixed with MV4-11 cancer cells (Figure 40A), resting human NK cells were mixed with Molm-13 cancer cells (Figure 40B), and resting human NK cells were mixed with THP-1 cancer cells (Figure 40C). TriNKET (e.g., CD33-TriNKET-C26 and CD33-TriNKET-F04) can enhance the cytotoxic activity of resting human NK cells in a dose-responsive manner against these cancer cells. The dotted line indicates the cytotoxic activity of resting NK cells without TriNKET. CD33 monoclonal antibodies showed reduced potency against MV4-11 cells, which express CD64 but at lower levels than THP-1 cells. CD33 monoclonal antibodies showed good potency against Molm-13 cells, which do not express CD64. CD33 monoclonal antibody had no effect on THP-1 cells, which have high CD64 levels.

[0205] TriNKET-mediated lysis of CD33-positive human cancer cells Molm-13 was assayed. Resting human NK cells were mixed with Molm-13 cancer cells at a ratio of 5:1 ( FIG. 41 ), and TriNKET (e.g., CD33-TriNKET-A49, CD33-TriNKET-A44, CD33-TriNKET-C26, and CD33-TriNKET-E79) was able to enhance the cytotoxic activity of resting human NK cells against these cancer cells in a dose-responsive manner. The dotted line indicates the cytotoxic activity of resting NK cells without TriNKET.

[0206] Example 12: Benefits of TriNKET in the treatment of cancers with high expression of FcR or in tumor microenvironments with high levels of FcR Monoclonal antibody therapy has been approved for the treatment of many cancer types, including both hematological and solid tumors. While the use of monoclonal antibodies in cancer therapy has improved patient outcomes, there are still limitations. Mechanistic studies have demonstrated that monoclonal antibodies exert their effects on tumor growth through multiple mechanisms, including ADCC, CDC, phagocytosis, and signal blockade, among others.

[0207] Most notably, ADCC is believed to be the primary mechanism by which monoclonal antibodies exert their effects. ADCC relies on antibody Fc engagement of the low affinity FcγRIII (CD16) on the surface of natural killer cells, which mediates direct lysis of tumor cells. Among the FcγRs, CD16 has the lowest affinity for IgG Fc, while FcγRI (CD64) is a high affinity FcR that binds IgG Fc approximately 1000-fold more tightly than CD16.

[0208] CD64 is normally expressed in many hematopoietic lineages, such as myeloid lineages, and can be expressed on tumors derived from these cell types, such as acute myeloid leukemia (AML). Tumor-infiltrating immune cells, such as MDSCs and monocytes, are also known to express CD64 and infiltrate the tumor microenvironment. Expression of CD64 by tumors or in the tumor microenvironment can have a detrimental effect on monoclonal antibody therapy. Because antibodies preferentially bind to high affinity receptors, expression of CD64 in the tumor microenvironment makes it difficult for these antibodies to engage CD16 on the surface of NK cells. TriNKET may be able to overcome the detrimental effect of CD64 expression on monoclonal antibody therapy through targeting two activating receptors on the surface of NK cells.

[0209] FcRγI (CD64) expression in three AML cell lines An in vitro culture system was developed to test the activity of TriNKET and monoclonal antibodies against tumors with high and low levels of CD64 surface expression. Molm-13 and THP-1 are two human AML cell lines with similar expression of surface CD33, but Molm-13 cells do not express CD64, whereas THP-1 cells express CD64 on their surface (Figures 38A-38C). Using monoclonal antibodies or TriNKET directed to target CD33, the effect of tumor CD64 expression on monoclonal antibody or TriNKET therapy was tested. Figures 38A-38C show the expression of high affinity FcRγI (CD64) on three human AML cell lines, Molm-13 cell line (Figure 38A), MV4-11 cell line (Figure 38B), and THP-1 cell line (Figure 38C). Molm-13 cells do not express CD64, whereas MV4-11 cells have low levels and THP-1 have high levels of cell surface CD64.

[0210] TriNKET has advantages in targeting tumor cells with high surface expression of FcR Figures 39A-39B show monoclonal antibody or TriNKET-mediated activation of human NK cells in co-culture with either Molm-13 (Figure 39B) or THP-1 (Figure 39A) cells. A monoclonal antibody against human CD33 demonstrated good activation of human NK cells in the Molm-13 co-culture system, as evidenced by increased CD107a degranulation and IFNγ production. This monoclonal antibody has no effect on the THP-1 co-culture system, where high levels of CD64 are present on the cancer cells. Interestingly, TriNKET was effective against both Molm-13 (Figure 39B) and THP-1 (Figure 39A) cells, indicating that TriNKET can overcome binding to CD64 on the tumor and effectively target NK cells for activation. Dual targeting of two activating receptors on NK cells provided stronger specific binding to NK cells. Monoclonal antibodies that target only CD16 on NK cells can be bound by other high affinity FcRs and prevent engagement of CD16 on NK cells. As shown in Figure 39C, TriNKET also effectively mediated activation of resting human NK cells in coculture with MV4-11 human AML cells.

[0211] TriNKET demonstrates efficacy against AML cell lines regardless of FcγRI expression Figures 40A-40C show human NK cytotoxicity assays using three human AML cell lines as targets. Monoclonal antibodies against CD33 showed good efficacy against Molm-13 cells, which do not express CD64 (Figure 40B). MV4-11 cells (Figure 40A), which express CD64 but at lower levels than THP-1, showed reduced efficacy with monoclonal anti-CD33. THP-1 cells (Figure 40C) showed no effect with monoclonal anti-CD33 alone. Regardless of CD64 expression on the tumor cells, TriNKET was able to mediate human NK cell responses against all tumor cells tested here.

[0212] Figures 40A-40C show that THP-1 cells were protected against monoclonal antibody therapy due to the high levels of high affinity FcR expression on their surface. TriNKET circumvented this protection by targeting two activating receptors on the surface of NK cells. The cytotoxicity data directly correlated with what was seen in the co-culture activation experiments. TriNKET circumvented the protection from mAb therapy seen for THP-1 cells and was able to induce NK cell-mediated lysis despite the high levels of FcR.

[0213] Example 13: Killing of normal bone marrow and normal B cells in PBMC cultures: TriNKET offers a better safety profile through fewer on-target off-tumor side effects Both natural killer cells and CD8 T cells can directly lyse tumor cells, but the mechanisms by which NK cells and CD8 T cells recognize normal self from tumor cells are different. NK cell activity is regulated by the balance of signals from activating (NCR, NKG2D, CD16, etc.) and inhibitory (KIR, NKG2A, etc.) receptors. The balance of these activating and inhibitory signals allows NK cells to determine healthy self cells from stressed, virus-infected or transformed self cells. This "built-in" mechanism of self-tolerance would help protect normal healthy tissues from NK cell responses. To extend this principle, NK cell self-tolerance would allow TriNKET to target antigens expressed on both self and tumor without off-tumor side effects or with an increased therapeutic window.

[0214] Unlike natural killer cells, T cells require recognition of specific peptides presented by MHC molecules for activation and effector function. T cells are the primary target of immunotherapy, and many strategies have been developed to redirect T cell responses against tumors. T cell bispecifics, checkpoint inhibitors, and CAR-T cells have all been approved by the FDA, but often suffer from dose-limiting toxicity. T cell bispecifics and CAR-T cells work around the TCR-MHC recognition system by using binding domains to target antigens on the surface of tumor cells and engineered signaling domains to transmit activation signals to effector cells. Although effective in eliciting antitumor immune responses, these therapies are often coupled with cytokine release syndrome (CRS) and on-target off-tumor side effects. TriNKET is unique in this context because it does not "bypass" the natural system of NK cell activation and inhibition. Instead, TriNKET is designed to tip the balance and provide additional activation signals to NK cells while maintaining NK tolerance to healthy self.

[0215] PBMCs were isolated from whole blood by density gradient centrifugation. Any contaminating red blood cells were lysed by incubation in ACK lysis buffer. PBMCs were washed three times in PBS and total PBMCs were counted. PBMCs were cultured at 10 6 The PBMCs were adjusted to 10 μg / mL. 1 mL of PBMCs was seeded into wells of a 24-well plate and the indicated TriNKET or mAbs were added to these PBMC cultures at 10 μg / mL. Cells were cultured overnight at 37°C with 5% CO2. The next day (24 hours later), PBMCs were harvested from the cultures and prepared for FACS analysis. The percentages of CD45+;CD19+ B cells and CD45+;CD33+;CD11b+ myeloid cells were analyzed across the different treatment groups.

[0216] Figure 42A shows that B cells (CD33 negative) from healthy donors were not affected by TriNKET targeting CD33. PBMCs treated with TriNKET targeting CD33 showed no effect on CD45+, CD3-, CD56- lymphocyte populations. Figure 42B shows that autologous CD33+ bone marrow cells were protected from CD33-targeting TriNKET-mediated NK cell responses and therefore resistant to TriNKET-mediated lysis. In these cultures, the frequency of CD45+, CD33+, CD11b+ bone marrow cells was not altered upon incubation with TriNKET targeting CD33.

[0217] Incorporation by Reference The entire disclosure of each of the patent documents and scientific articles cited herein is incorporated by reference for all purposes.

[0218] Equivalent The present invention may be embodied in other specific forms without departing from its spirit or essential characteristics. The foregoing embodiments are therefore to be considered in all respects as illustrative and not limiting of the invention described herein. The scope of the invention is therefore indicated by the appended claims rather than by the foregoing description, and all changes that come within the meaning and range of equivalency of the claims are intended to be embraced therein. The present application provides the following aspects of the invention. (Aspect 1) The following: (a) a first antigen-binding site that binds to NKG2D; (b) a second antigen-binding site that binds to CD33; and (c) an antibody Fc domain or a portion thereof sufficient to bind to CD16, or a third antigen-binding site that binds to CD16. Proteins including: (Aspect 2) 2. The protein of embodiment 1, wherein the first antigen-binding site binds to human, non-human primate, and rodent NKG2D. (Aspect 3) 3. The protein of embodiment 1 or 2, wherein the first antigen-binding site comprises a heavy chain variable domain and a light chain variable domain. (Aspect 4) 4. The protein of embodiment 3, wherein the heavy chain variable domain and the light chain variable domain are present in the same polypeptide. (Aspect 5) The protein according to any one of aspects 3 to 4, wherein the second antigen-binding site comprises a heavy chain variable domain and a light chain variable domain. (Aspect 6) 6. The protein of embodiment 5, wherein the heavy chain variable domain and the light chain variable domain of the second antigen-binding site are present in the same polypeptide. (Aspect 7) 7. The protein according to embodiment 5 or 6, wherein the light chain variable domain of the first antigen-binding site has an amino acid sequence identical to the amino acid sequence of the light chain variable domain of the second antigen-binding site. (Aspect 8) The protein according to any one of aspects 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO:1. (Aspect 9) 8. The protein according to any one of aspects 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO: 41 and a light chain variable domain that is at least 90% identical to SEQ ID NO: 42. (Aspect 10) 8. The protein according to any one of aspects 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO: 43 and a light chain variable domain that is at least 90% identical to SEQ ID NO: 44. (Aspect 11) 8. The protein according to any one of aspects 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO: 45 and a light chain variable domain that is at least 90% identical to SEQ ID NO: 46. (Aspect 12) 8. The protein according to any one of aspects 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO: 47 and a light chain variable domain that is at least 90% identical to SEQ ID NO: 48. (Aspect 13) 8. The protein according to any one of aspects 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO: 69 and a light chain variable domain that is at least 90% identical to SEQ ID NO: 70. (Aspect 14) 8. The protein according to any one of aspects 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO: 77 and a light chain variable domain that is at least 90% identical to SEQ ID NO: 78. (Aspect 15) 8. The protein according to any one of aspects 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO: 85 and a light chain variable domain that is at least 90% identical to SEQ ID NO: 86. (Aspect 16) 8. The protein according to any one of aspects 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO: 133 and a light chain variable domain that is at least 90% identical to SEQ ID NO: 134. (Aspect 17) 3. The protein of embodiment 1 or 2, wherein the first antigen-binding site is a single domain antibody. (Aspect 18) 18. The protein of embodiment 17, wherein the single domain antibody is a VHH fragment or a VNAR fragment. (Aspect 19) 19. The protein according to any one of aspects 1 to 2 or 17 to 18, wherein the second antigen-binding site comprises a heavy chain variable domain and a light chain variable domain. (Aspect 20) 20. The protein of embodiment 19, wherein the heavy chain variable domain and the light chain variable domain of the second antigen-binding site are present in the same polypeptide. (Aspect 21) 21. The protein according to any one of aspects 1 to 20, wherein the heavy chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO: 93, and the light chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO: 94. (Aspect 22) 21. The protein according to any one of aspects 1 to 20, wherein the heavy chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO: 101, and the light chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO: 102. (Aspect 23) 21. The protein according to any one of aspects 1 to 20, wherein the heavy chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO: 109, and the light chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO: 110. (Aspect 24) 21. The protein according to any one of aspects 1 to 20, wherein the heavy chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO: 117, and the light chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO: 118. (Aspect 25) 21. The protein according to any one of aspects 1 to 20, wherein the heavy chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO: 125, and the light chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO: 126. (Aspect 26) The protein according to any one of aspects 1 to 4 or 8 to 16, wherein the second antigen-binding site is a single domain antibody. (Aspect 27) 27. The protein of embodiment 26, wherein the second antigen-binding site is a VHH fragment or a VNAR fragment. (Aspect 28) 28. The protein of any one of aspects 1 to 27, wherein the protein comprises a sufficient portion of an antibody Fc domain to bind to CD16, the antibody Fc domain comprising a hinge and a CH2 domain. (Aspect 29) 30. The protein of embodiment 28, wherein the antibody Fc domain comprises the hinge and CH2 domains of a human IgG1 antibody. (Aspect 30) 30. The protein according to aspect 28 or 29, wherein the Fc domain comprises an amino acid sequence that is at least 90% identical to amino acids 234 to 332 of a human IgG1 antibody. (Aspect 31) 31. The protein of any one of aspects 28 to 30, wherein the Fc domain comprises an amino acid sequence that is at least 90% identical to an Fc domain of human IgG1 and differs at one or more positions selected from the group consisting of Q347, Y349, T350, L351, S354, E356, E357, K360, Q362, S364, T366, L368, K370, N390, K392, T394, D399, S400, D401, F405, Y407, K409, T411, K439. (Aspect 32) A formulation comprising the protein according to any one of aspects 1 to 31 and a pharma- ceutical acceptable carrier. (Aspect 33) A cell comprising one or more nucleic acids expressing the protein according to any one of aspects 1 to 31. (Aspect 34) 32. A method for directly and / or indirectly enhancing tumor cell death, the method comprising exposing tumor and natural killer cells to a protein according to any one of aspects 1 to 31. (Aspect 35) A method for treating cancer, the method comprising administering to a patient a protein according to any one of aspects 1 to 31 or a formulation according to aspect 32. (Aspect 36) The method of embodiment 35, wherein said cancer is selected from the group consisting of AML, myelodysplastic syndrome, chronic myelomonocytic leukemia, myeloid blast crisis of chronic myelogenous leukemia, and ALL.

Claims

1. The following: (a) a first antigen-binding site that binds to NKG2D; (b) a second antigen-binding site that binds to CD33; and (c) an antibody Fc domain or a portion thereof sufficient to bind to CD16, or a third antigen-binding site that binds to CD16. Proteins including:

2. The protein of claim 1, wherein the first antigen-binding site binds to human, non-human primate, and rodent NKG2D.

3. The protein of claim 1 or 2, wherein the first antigen-binding site comprises a heavy chain variable domain and a light chain variable domain.

4. The protein of claim 3 , wherein the heavy chain variable domain and the light chain variable domain are present in the same polypeptide.

5. The protein according to any one of claims 3 to 4, wherein the second antigen-binding site comprises a heavy chain variable domain and a light chain variable domain.

6. The protein of claim 5 , wherein the heavy chain variable domain and the light chain variable domain of the second antigen binding site are present in the same polypeptide.

7. 7. The protein according to claim 5 or 6, wherein the light chain variable domain of the first antigen-binding site has an amino acid sequence identical to the amino acid sequence of the light chain variable domain of the second antigen-binding site.

8. The protein according to any one of claims 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO:

1.

9. 8. The protein according to any one of claims 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO: 41 and a light chain variable domain that is at least 90% identical to SEQ ID NO:

42.

10. 8. The protein according to any one of claims 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO: 43 and a light chain variable domain that is at least 90% identical to SEQ ID NO:

44.

11. 8. The protein according to any one of claims 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO: 45 and a light chain variable domain that is at least 90% identical to SEQ ID NO:

46.

12. 8. The protein according to any one of claims 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO: 47 and a light chain variable domain that is at least 90% identical to SEQ ID NO:

48.

13. 8. The protein according to any one of claims 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO: 69 and a light chain variable domain that is at least 90% identical to SEQ ID NO:

70.

14. 8. The protein according to any one of claims 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO: 77 and a light chain variable domain that is at least 90% identical to SEQ ID NO:

78.

15. 8. The protein according to any one of claims 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO: 85 and a light chain variable domain that is at least 90% identical to SEQ ID NO:

86.

16. 8. The protein according to any one of claims 1 to 7, wherein the first antigen-binding site comprises a heavy chain variable domain that is at least 90% identical to SEQ ID NO: 133 and a light chain variable domain that is at least 90% identical to SEQ ID NO:

134.

17. The protein of claim 1 or 2, wherein the first antigen-binding site is a single domain antibody.

18. The single domain antibody is H H fragment or V NAR 18. The protein of claim 17, which is a fragment.

19. 19. The protein of any one of claims 1 to 2 or 17 to 18, wherein the second antigen-binding site comprises a heavy chain variable domain and a light chain variable domain.

20. 20. The protein of claim 19, wherein the heavy chain variable domain and the light chain variable domain of the second antigen binding site are present in the same polypeptide.

21. 21. The protein according to any one of claims 1 to 20, wherein the heavy chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO: 93 and the light chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO:

94.

22. 21. The protein according to any one of claims 1 to 20, wherein the heavy chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO: 101 and the light chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO:

102.

23. 21. The protein according to any one of claims 1 to 20, wherein the heavy chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO: 109 and the light chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO:

110.

24. 21. The protein according to any one of claims 1 to 20, wherein the heavy chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO: 117 and the light chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO:

118.

25. 21. The protein according to any one of claims 1 to 20, wherein the heavy chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO: 125 and the light chain variable domain of the second antigen-binding site comprises an amino acid sequence that is at least 90% identical to SEQ ID NO:

126.

26. The protein of any one of claims 1 to 4 or 8 to 16, wherein the second antigen-binding site is a single domain antibody.

27. The second antigen-binding site is V H H fragment or V NAR 27. The protein of claim 26, which is a fragment.

28. 28. The protein of any one of claims 1 to 27, wherein the protein comprises a sufficient portion of an antibody Fc domain to bind to CD16, the antibody Fc domain comprising a hinge and a CH2 domain.

29. 29. The protein of claim 28, wherein the antibody Fc domain comprises the hinge and CH2 domains of a human IgG1 antibody.

30. 30. The protein of claim 28 or 29, wherein the Fc domain comprises an amino acid sequence that is at least 90% identical to amino acids 234 to 332 of a human IgG1 antibody.

31. 31. The protein of any one of claims 28 to 30, wherein the Fc domain comprises an amino acid sequence that is at least 90% identical to the Fc domain of human IgG1 and differs at one or more positions selected from the group consisting of Q347, Y349, T350, L351, S354, E356, E357, K360, Q362, S364, T366, L368, K370, N390, K392, T394, D399, S400, D401, F405, Y407, K409, T411, K439.

32. A formulation comprising a protein according to any one of claims 1 to 31 and a pharma- ceutically acceptable carrier.

33. A cell comprising one or more nucleic acids expressing a protein according to any one of claims 1 to 31.

34. 32. A method for directly and / or indirectly enhancing tumor cell death, comprising exposing tumor and natural killer cells to a protein according to any one of claims 1 to 31.

35. A method for treating cancer, comprising administering to a patient a protein according to any one of claims 1 to 31 or a formulation according to claim 32.

36. 36. The method of claim 35, wherein the cancer is selected from the group consisting of AML, myelodysplastic syndrome, chronic myelomonocytic leukemia, myeloid blast crisis of chronic myelogenous leukemia, and ALL.

Images