Antibodies that bind to PD-L1, PD-L2 and / or CD28

Abstract

Provided herein are novel aPD-L1, aPD-L2, and aCD28 antibodies. In some embodiments, the antibodies are aPD-L1xaPD-L2xaCD28 antibodies. Such antibodies enhance anti-tumor activity by providing costimulatory signals for T cell activation against tumor cells, while also advantageously blocking inhibitory PD-L1:PD1 and / or PD-L2:PD1 pathway interactions.

Description

[Technical Field]

[0001] Priority claim This application claims priority to and the benefit of U.S. Provisional Application Nos. 63 / 330,769, filed April 13, 2022, 63 / 379,100, filed October 11, 2022, 63 / 382,592, filed November 7, 2022, 63 / 478,881, filed January 6, 2023, and 63 / 480,478, filed January 18, 2023, the contents of which are hereby incorporated by reference in their entireties.

[0002] Sequence Listing This application contains a Sequence Listing that has been submitted electronically in XML file format, which is hereby incorporated by reference in its entirety. The XML copy, created on April 11, 2023, is named 067461-5301-WO_SL.xml and is 3,420,050 bytes in size. [Background technology]

[0003] The innate immune response to tumors deploys immune effector cells, such as natural killer (NK) cells and T cells, to attack and destroy tumor cells. Tumor-infiltrating lymphocytes (TILs) often express multiple immune checkpoint receptors (e.g., PD-1, CTLA-4) and costimulatory receptors (e.g., ICOS, 4-1BB, OX40, GITR, and CD28). TILs lose their cytotoxicity over time due to the upregulation of inhibitory immune checkpoints. Although checkpoint blockade has demonstrated increased clinical response rates compared to other treatment options, many patients still fail to respond to checkpoint blockade. Engagement of costimulatory receptors on TILs could provide positive signals that can overcome the negative signals of immune checkpoints. Preclinical and clinical studies of agonistic costimulatory receptor antibodies have demonstrated that agonism of costimulatory receptors can result in superior antitumor responses and activate T cells to attack tumor cells.

[0004] Improving antitumor activity by specifically destroying tumor cells while minimizing peripheral toxicity is also important for cancer therapy. In this context, it is important to provide costimulatory signals only to T cells in the presence of target tumor cells. However, agonism of costimulatory receptors with monospecific full-length antibodies is likely not discriminatory with respect to TILs, peripheral T cells, or autoantigen-reactive T cells that contribute to autoimmune toxicity. For example, urelumab, a monospecific, non-discriminatory pan-4-1BB agonist antibody, demonstrated significant hepatotoxicity in early-phase clinical trials (Segal et al., 2016). Thus, there remains a need for novel immune response enhancing compositions for cancer treatment. Summary of the Invention

[0005] Provided herein are novel αPD-L1, αPD-L2, and αCD28 antibodies. In some embodiments, the antibodies are αPD-L1×αPD-L2×αCD28 trispecific antibodies. Such antibodies enhance anti-tumor activity by providing costimulatory signals for T cell activation against tumor cells, while also advantageously blocking inhibitory PD-L1:PD1 and / or PD-L2:PD1 pathway interactions (see Figure 94). In some embodiments, such trispecific antibodies are useful in the treatment of cancer when combined with an αCD3×α tumor target antigen (TTA) bispecific antibody.

[0006] In one aspect, provided herein is a multispecific antibody comprising a PD-L1 antigen-binding domain, a PD-L2 antigen-binding domain, and a CD28 binding domain, wherein each of the binding domains comprises a variable heavy chain domain and a variable light chain domain.

[0007] In one aspect, provided herein is a novel 1+1+1 stack Fab2-Fab-Fc format antibody (Figure 83D). The antibody comprises a) a first monomer, b) a second monomer, and c) a first, second, and third common light chain. The first monomer comprises, from N- to C-terminus, VH1-CH1-hinge-CH2-CH3, where VH1 is the first variable heavy chain domain and CH2-CH3 is the first Fc domain. The second monomer comprises, from N- to C-terminus, VH2-CH1-linker-VH3-CH1-hinge-CH2-CH3, where VH2 is the second variable heavy chain domain, VH3 is the third variable heavy chain domain, and CH2-CH3 is the second Fc domain. Each common light chain comprises, from N-terminus to C-terminus, VL-CL, where VL is a common variable light chain domain and CL is a constant light chain domain. In this format, the first variable heavy chain domain and common variable light chain domain of the first common light chain form a first antigen-binding domain, the second variable heavy chain domain and common variable light chain domain of the second common light chain form a second antigen-binding domain, and the third variable heavy chain domain and common variable light chain domain form a third antigen-binding domain of the third common light chain.

[0008] In some embodiments, the antibody is a trispecific antibody that binds to PD-L1, PD-L2, and CD28, and each of the first antigen-binding domain, the second antigen-binding domain, and the third antigen-binding domain is selected from the group consisting of a PD-L1 antigen-binding domain, a PD-L2 antigen-binding domain, and a CD28-binding domain.

[0009] In some embodiments, the first antigen-binding domain, the second antigen-binding domain, and the third antigen-binding domain are selected from: i) the first antigen-binding domain is a PD-L1-binding domain, the second antigen-binding domain is a PD-L2-binding domain, and the third antigen-binding domain is a CD28-binding domain; ii) the first antigen-binding domain is a PD-L1-binding domain, the second antigen-binding domain is a CD28-binding domain, and the third antigen-binding domain is a PD-L2-binding domain; or iii) the first antigen-binding domain is a PD-L2-binding domain and the second antigen-binding domain is a PD-L1-binding domain. and the third antigen-binding domain is a CD28-binding domain; or iv) the first antigen-binding domain is a PD-L2-binding domain, the second antigen-binding domain is a CD28-binding domain, and the third antigen-binding domain is a PD-L1-binding domain; or v) the first antigen-binding domain is a CD28-binding domain, the second antigen-binding domain is a PD-L1-binding domain, and the third antigen-binding domain is a PD-L2-binding domain; or vi) the first antigen-binding domain is a CD28-binding domain, the second antigen-binding domain is a PD-L2-binding domain, and the third antigen-binding domain is a PD-L1-binding domain.

[0010] In some embodiments, the CD28 binding domain comprises a variable heavy chain domain having an amino acid sequence selected from SEQ ID NOs: 3354-3389 or a variant thereof, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239 or a variant thereof.

[0011] In some embodiments, the PD-L1 antigen-binding domain comprises an amino acid sequence selected from SEQ ID NOs: 3235 and 3243-3260, or a variant thereof, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239, or a variant thereof.

[0012] In some embodiments, the PD-L2 antigen-binding domain comprises an amino acid sequence selected from SEQ ID NOs: 3267 and 3275-3347 or a variant thereof, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239 or a variant thereof.

[0013] In some embodiments, the first and second Fc domains are variant Fc domains. In certain embodiments, the first and second Fc domains comprise a set of heterodimerized scubariants selected from the group consisting of S364K / E357Q:L368D / K370S; S364K:L368D / K370S; S364K:L368E / K370S; D401K:T411E / K360E / Q362E; and T366W:T366S / L368A / Y407V, where the numbering is according to EU numbering. In some embodiments, the first and second Fc domains comprise the heterodimerized scubariant S364K / E357Q:L368D / K370S.

[0014] In some embodiments, the first and second Fc domains each comprise one or more reduced variants, hi some embodiments, the one or more reduced variants are E233P / L234V / L235A / G236del / S267K, where the numbering is according to EU numbering.

[0015] In certain embodiments, one of the first or second monomers comprises one or more pI variants, hi some embodiments, the CH1-hinge-CH2-CH3 of the first monomer comprises the pI variants N208D / Q295E / N384D / Q418E / N421D, where the numbering is according to EU numbering.

[0016] In some embodiments, the first Fc domain comprises the amino acid variants L368D / K370S / N208D / Q295E / N384D / Q418E / N421D / E233P / L234V / L235A / G236del / S267K and the CH1-hinge-CH2-CH3 of the second monomer comprises the amino acid variants S364K / E357Q / E233P / L234V / L235A / G236del / S267K, where the numbering is according to EU numbering.

[0017] In some embodiments, the first and second Fc domains each further comprise the amino acid variant 428 / 434S.

[0018] In some embodiments, a) the PD-L1 binding domain comprises a variable heavy chain domain having the amino acid sequence of SEQ ID NO: 3251 and a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239; b) the PD-L2 binding domain comprises a variable heavy chain domain having the amino acid sequence of SEQ ID NO: 3319 and a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239; and c) the CD28 binding domain comprises a variable heavy chain domain having the amino acid sequence of SEQ ID NO: 3380 and a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239.

[0019] In some embodiments, a) the first monomer has the amino acid sequence of SEQ ID NO: 3201, b) the second monomer has the amino acid sequence of SEQ ID NO: 3202, and c) the first, second, and third common light chains each have the amino acid sequence of SEQ ID NO: 3203.

[0020] In one aspect, provided herein is a novel 1+1+1 stack Fab2-scFv-Fc format antibody (Figure 83A). The 1+1+1 stack Fab2-scFv-Fc format comprises a) a first monomer, b) a second monomer, and c) a first and a second common light chain. The first monomer comprises, from N- to C-terminus, scFv-linker-CH2-CH3, where CH2-CH3 is a first Fc domain. The second monomer comprises, from N- to C-terminus, VH1-CH1-linker-VH2-CH1-hinge-CH2-CH3, where VH1 is a first variable heavy chain domain, VH2 is a second variable heavy chain domain, and CH2-CH3 is a second Fc domain. The first and second common light chains each comprise, from N- to C-terminus, VL1-CL, where VL1 is the first variable light domain and CL is the constant light domain. The scFv further comprises a third VH domain (VH3), an scFv linker, and a second variable light domain (VL2). In this format, the first variable heavy domain and the first variable light domain of the first common light chain form a first antigen-binding domain, the second variable heavy domain and the first variable light domain of the second common light chain form a second antigen-binding domain, and the third variable heavy domain and the second variable light domain form a third antigen-binding domain.

[0021] In some embodiments, the antibody is a trispecific antibody that binds to PD-L1, PD-L2, and CD28, and each of the first antigen-binding domain, the second antigen-binding domain, and the third antigen-binding domain is selected from the group consisting of a PD-L1 antigen-binding domain, a PD-L2 antigen-binding domain, and a CD28-binding domain.

[0022] In some embodiments, the first antigen-binding domain, the second antigen-binding domain, and the third antigen-binding domain are selected from the following:

[0023] i) the first antigen-binding domain is a PD-L1-binding domain, the second antigen-binding domain is a PD-L2-binding domain, and the third antigen-binding domain is a CD28-binding domain; ii) the first antigen-binding domain is a PD-L1-binding domain, the second antigen-binding domain is a CD28-binding domain, and the third antigen-binding domain is a PD-L2-binding domain; or iii) the first antigen-binding domain is a PD-L2-binding domain, the second antigen-binding domain is a PD-L1-binding domain, and the third antigen-binding domain is a CD28-binding domain; iv) the first antigen-binding domain is a PD-L2-binding domain, the second antigen-binding domain is a CD28-binding domain, and the third antigen-binding domain is a PD-L1-binding domain; v) the first antigen-binding domain is a CD28-binding domain, the second antigen-binding domain is a PD-L1-binding domain, and the third antigen-binding domain is a PD-L2-binding domain; or vi) the first antigen-binding domain is a CD28-binding domain, the second antigen-binding domain is a PD-L2-binding domain, and the third antigen-binding domain is a PD-L1-binding domain.

[0024] In some embodiments, the CD28 binding domain comprises a variable heavy chain domain and a variable light chain domain selected from the following: i) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 35, 36, 38, and a variable light chain domain selected from any of the variable light chain domains in Figures 35, 37, and 38; ii) the variable heavy chain domain and variable light chain domain of the CD28 binding domain in Figure 38; iii) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 41-74, and a variable light chain domain selected from any of the variable light chain domains in Figures 41-74; iv) 41 to 74, v) a variable heavy chain domain selected from any of the variable light chain domains of Figures 41 to 74, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 20 (Figure 24), vi) a variable heavy chain domain and a variable light chain domain of the CD28 binding domain of Figure 81, and vii) a variable heavy chain domain selected from any of the variable heavy chain domains of Figure 163, and a variable light chain domain selected from any of the variable light chain domains of Figures 35 and 37, or a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239.

[0025] In some embodiments, the PD-L1 antigen-binding domain comprises a variable heavy chain domain and a variable light chain domain selected from: i) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 17 and 19, and a variable light chain domain selected from any of the variable light chain domains in Figures 17 and 20; ii) the variable heavy chain domain and variable light chain domain of the PD-L1 binding domain in Figures 17 and 21; iii) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 25 and 26, and one of the variable light chain domains in Figures 25 and 26. 25 and 26; iv) a variable heavy chain domain and a variable light chain domain of the PD-L1 binding domain in Figures 25 and 26; v) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 25 and 26, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 20 (Figure 24); vi) a variable heavy chain domain and a variable light chain domain of the PD-L1 binding domain in Figure 28, and vii) a variable heavy chain domain selected from any of the variable heavy chain domains in Figure 161, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239.

[0026] In some embodiments, the PD-L2 antigen-binding domain comprises a variable heavy chain domain and a variable light chain domain selected from the following: i) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 29-32, and a variable light chain domain selected from any of the variable light chain domains in Figures 29-32; ii) the variable heavy chain domain and variable light chain domain of the PD-L2 binding domain in Figures 29-32; iii) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 29-32, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 20 (Figure 24); iv) the variable heavy chain domain and variable light chain domain of the PD-L2 binding domain in Figure 34; and v) a variable heavy chain domain selected from any of the variable heavy chain domains in Figure 162, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239.

[0027] In some embodiments, the first and second Fc domains are variant Fc domains. In certain embodiments, the first and second Fc domains comprise a set of heterodimerized scubariants selected from the group consisting of S364K / E357Q:L368D / K370S; S364K:L368D / K370S; S364K:L368E / K370S; D401K:T411E / K360E / Q362E; and T366W:T366S / L368A / Y407V, where the numbering is according to EU numbering. In some embodiments, the first and second Fc domains comprise the heterodimerized scubariant S364K / E357Q:L368D / K370S.

[0028] In some embodiments, the first and second Fc domains each comprise one or more reduced variants, hi some embodiments, the one or more reduced variants are E233P / L234V / L235A / G236del / S267K, where the numbering is according to EU numbering.

[0029] In some embodiments, one of the first or second monomers comprises one or more pI variants, in some embodiments, the CH1-hinge-CH2-CH3 of the second monomer comprises the pI variants N208D / Q295E / N384D / Q418E / N421D, where the numbering is according to EU numbering.

[0030] In some embodiments, the first Fc domain comprises the amino acid variants S364K / E357Q / E233P / L234V / L235A / G236del / S267K and the CH1-hinge-CH2-CH3 of the second monomer comprises the amino acid variants L368D / K370S / N208D / Q295E / N384D / Q418E / N421D / E233P / L234V / L235A / G236del / S267K, where numbering is according to EU numbering.

[0031] In some embodiments, the scFv linker is a charged scFv linker having the amino acid sequence (GKPGS) 4. In some embodiments, the first and second Fc domains each further comprise the amino acid variant 428 / 434S.

[0032] In one aspect, provided herein is a novel 1+1+1 Fab-(Fab-scFv)-Fc format antibody (Figure 83B). The 1+1+1 Fab-(Fab-scFv)-Fc format comprises a) a first monomer, b) a second monomer, and c) a first and a second common light chain. The first monomer comprises, from N- to C-terminus, VH1-CH1-linker-scFv-linker-CH2-CH3, where VH1 is the first variable heavy chain domain and CH2-CH3 is the first Fc domain. The second monomer comprises, from N- to C-terminus, VH2-CH1-hinge-CH2-CH3, where VH2 is the second variable heavy chain domain and CH2-CH3 is the second Fc domain. The first and second common light chains each comprise, from N- to C-terminus, VL1-CL, where VL1 is the first variable light domain and CL is the constant light domain. The scFv further comprises a third VH domain (VH3), an scFv linker, and a second variable light domain (VL2). In this format, the first variable heavy domain and the first variable light domain of the first common light chain form a first antigen-binding domain, the second variable heavy domain and the first variable light domain of the second common light chain form a second antigen-binding domain, and the third variable heavy domain and the second variable light domain form a third antigen-binding domain.

[0033] In some embodiments, the antibody is a trispecific antibody that binds to PD-L1, PD-L2, and CD28, and each of the first antigen-binding domain, the second antigen-binding domain, and the third antigen-binding domain is selected from the group consisting of a PD-L1 antigen-binding domain, a PD-L2 antigen-binding domain, and a CD28-binding domain.

[0034] In some embodiments, the first antigen-binding domain, the second antigen-binding domain, and the third antigen-binding domain are selected from the following:

[0035] i) the first antigen-binding domain is a PD-L1-binding domain, the second antigen-binding domain is a PD-L2-binding domain, and the third antigen-binding domain is a CD28-binding domain; ii) the first antigen-binding domain is a PD-L1-binding domain, the second antigen-binding domain is a CD28-binding domain, and the third antigen-binding domain is a PD-L2-binding domain; or iii) the first antigen-binding domain is a PD-L2-binding domain, the second antigen-binding domain is a PD-L1-binding domain, and the third antigen-binding domain is a CD28-binding domain; iv) the first antigen-binding domain is a PD-L2-binding domain, the second antigen-binding domain is a CD28-binding domain, and the third antigen-binding domain is a PD-L1-binding domain; v) the first antigen-binding domain is a CD28-binding domain, the second antigen-binding domain is a PD-L1-binding domain, and the third antigen-binding domain is a PD-L2-binding domain; or vi) the first antigen-binding domain is a CD28-binding domain, the second antigen-binding domain is a PD-L2-binding domain, and the third antigen-binding domain is a PD-L1-binding domain.

[0036] In some embodiments, the CD28 binding domain comprises a variable heavy chain domain and a variable light chain domain selected from the following: i) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 35, 36, 38, and a variable light chain domain selected from any of the variable light chain domains in Figures 35, 37, and 38; ii) the variable heavy chain domain and variable light chain domain of the CD28 binding domain in Figure 38; iii) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 41-74, and a variable light chain domain selected from any of the variable light chain domains in Figures 41-74; iv) 41 to 74, v) a variable heavy chain domain selected from any of the variable light chain domains of Figures 41 to 74, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 20 (Figure 24), vi) a variable heavy chain domain and a variable light chain domain of the CD28 binding domain of Figure 81, and vii) a variable heavy chain domain selected from any of the variable heavy chain domains of Figure 163, and a variable light chain domain selected from any of the variable light chain domains of Figures 35 and 37, or a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239.

[0037] In some embodiments, the PD-L1 antigen-binding domain comprises a variable heavy chain domain and a variable light chain domain selected from: i) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 17 and 19, and a variable light chain domain selected from any of the variable light chain domains in Figures 17 and 20; ii) the variable heavy chain domain and variable light chain domain of the PD-L1 binding domain in Figures 17 and 21; iii) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 25 and 26, and one of the variable light chain domains in Figures 25 and 26. 25 and 26; iv) a variable heavy chain domain and a variable light chain domain of the PD-L1 binding domain in Figures 25 and 26; v) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 25 and 26, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 20 (Figure 24); vi) a variable heavy chain domain and a variable light chain domain of the PD-L1 binding domain in Figure 28, and vii) a variable heavy chain domain selected from any of the variable heavy chain domains in Figure 161, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239.

[0038] In some embodiments, the PD-L2 antigen-binding domain comprises a variable heavy chain domain and a variable light chain domain selected from the following: i) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 29-32, and a variable light chain domain selected from any of the variable light chain domains in Figures 29-32; ii) the variable heavy chain domain and variable light chain domain of the PD-L2 binding domain in Figures 29-32; iii) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 29-32, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 20 (Figure 24); iv) the variable heavy chain domain and variable light chain domain of the PD-L2 binding domain in Figure 34; and v) a variable heavy chain domain selected from any of the variable heavy chain domains in Figure 162, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239.

[0039] In some embodiments, the first and second Fc domains are variant Fc domains. In certain embodiments, the first and second Fc domains comprise a set of heterodimerized scubariants selected from the group consisting of S364K / E357Q:L368D / K370S; S364K:L368D / K370S; S364K:L368E / K370S; D401K:T411E / K360E / Q362E; and T366W:T366S / L368A / Y407V, where the numbering is according to EU numbering. In some embodiments, the first and second Fc domains comprise the heterodimerized scubariant S364K / E357Q:L368D / K370S.

[0040] In some embodiments, the first and second Fc domains each comprise one or more reduced variants, hi some embodiments, the one or more reduced variants are E233P / L234V / L235A / G236del / S267K, where the numbering is according to EU numbering.

[0041] In some embodiments, one of the first or second monomers comprises one or more pI variants, in some embodiments, the CH1-hinge-CH2-CH3 of the second monomer comprises the pI variants N208D / Q295E / N384D / Q418E / N421D, where the numbering is according to EU numbering.

[0042] In some embodiments, the first Fc domain comprises the amino acid variants S364K / E357Q / E233P / L234V / L235A / G236del / S267K and the CH1-hinge-CH2-CH3 of the second monomer comprises the amino acid variants L368D / K370S / N208D / Q295E / N384D / Q418E / N421D / E233P / L234V / L235A / G236del / S267K, where numbering is according to EU numbering.

[0043] In some embodiments, the scFv linker is a charged scFv linker having the amino acid sequence (GKPGS) 4. In some embodiments, the first and second Fc domains each further comprise the amino acid variant 428 / 434S.

[0044] In one aspect, provided herein is a novel 1+1+1 mAb-scFv format antibody (Figure 83C). A 1+1+1 mAb-scFv format antibody generally comprises a) a first monomer, b) a second monomer, and c) a first and a second common light chain. The first monomer comprises, from N- to C-terminus, VH1-CH1-hinge-CH2-CH3-domain linker-scFv, where VH1 is the first variable heavy chain domain and CH2-CH3 is the first Fc domain. The second monomer comprises, from N- to C-terminus, VH2-CH1-hinge-CH2-CH3, where VH2 is the second variable heavy chain domain and CH2-CH3 is the second Fc domain. The first and second common light chains each comprise, from N- to C-terminus, VL1-CL, where VL1 is the first variable light domain and CL is the constant light domain. The scFv further comprises a third VH domain (VH3), an scFv linker, and a second variable light domain (VL2). In this format, the first variable heavy domain and the first variable light domain of the first common light chain form a first antigen-binding domain, the second variable heavy domain and the first variable light domain of the second common light chain form a second antigen-binding domain, and the third variable heavy domain and the second variable light domain form a third antigen-binding domain.

[0045] In some embodiments, the antibody is a trispecific antibody that binds to PD-L1, PD-L2, and CD28, and each of the first antigen-binding domain, the second antigen-binding domain, and the third antigen-binding domain is selected from the group consisting of a PD-L1 antigen-binding domain, a PD-L2 antigen-binding domain, and a CD28-binding domain.

[0046] In some embodiments, the first antigen-binding domain, the second antigen-binding domain, and the third antigen-binding domain are selected from the following:

[0047] i) the first antigen-binding domain is a PD-L1-binding domain, the second antigen-binding domain is a PD-L2-binding domain, and the third antigen-binding domain is a CD28-binding domain; ii) the first antigen-binding domain is a PD-L1-binding domain, the second antigen-binding domain is a CD28-binding domain, and the third antigen-binding domain is a PD-L2-binding domain; or iii) the first antigen-binding domain is a PD-L2-binding domain, the second antigen-binding domain is a PD-L1-binding domain, and the third antigen-binding domain is a CD28-binding domain; iv) the first antigen-binding domain is a PD-L2-binding domain, the second antigen-binding domain is a CD28-binding domain, and the third antigen-binding domain is a PD-L1-binding domain; v) the first antigen-binding domain is a CD28-binding domain, the second antigen-binding domain is a PD-L1-binding domain, and the third antigen-binding domain is a PD-L2-binding domain; or vi) the first antigen-binding domain is a CD28-binding domain, the second antigen-binding domain is a PD-L2-binding domain, and the third antigen-binding domain is a PD-L1-binding domain.

[0048] In some embodiments, the CD28 binding domain comprises a variable heavy chain domain and a variable light chain domain selected from the following: i) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 35, 36, 38, and a variable light chain domain selected from any of the variable light chain domains in Figures 35, 37, and 38; ii) the variable heavy chain domain and variable light chain domain of the CD28 binding domain in Figure 38; iii) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 41-74, and a variable light chain domain selected from any of the variable light chain domains in Figures 41-74; iv) 41 to 74, v) a variable heavy chain domain selected from any of the variable light chain domains of Figures 41 to 74, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 20 (Figure 24), vi) a variable heavy chain domain and a variable light chain domain of the CD28 binding domain of Figure 81, and vii) a variable heavy chain domain selected from any of the variable heavy chain domains of Figure 163, and a variable light chain domain selected from any of the variable light chain domains of Figures 35 and 37, or a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239.

[0049] In some embodiments, the PD-L1 antigen-binding domain comprises a variable heavy chain domain and a variable light chain domain selected from: i) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 17 and 19, and a variable light chain domain selected from any of the variable light chain domains in Figures 17 and 20; ii) the variable heavy chain domain and variable light chain domain of the PD-L1 binding domain in Figures 17 and 21; iii) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 25 and 26, and one of the variable light chain domains in Figures 25 and 26. 25 and 26; iv) a variable heavy chain domain and a variable light chain domain of the PD-L1 binding domain in Figures 25 and 26; v) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 25 and 26, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 20 (Figure 24); vi) a variable heavy chain domain and a variable light chain domain of the PD-L1 binding domain in Figure 28, and vii) a variable heavy chain domain selected from any of the variable heavy chain domains in Figure 161, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239.

[0050] In some embodiments, the PD-L2 antigen-binding domain comprises a variable heavy chain domain and a variable light chain domain selected from the following: i) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 29-32, and a variable light chain domain selected from any of the variable light chain domains in Figures 29-32; ii) the variable heavy chain domain and variable light chain domain of the PD-L2 binding domain in Figures 29-32; iii) a variable heavy chain domain selected from any of the variable heavy chain domains in Figures 29-32, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 20 (Figure 24); iv) the variable heavy chain domain and variable light chain domain of the PD-L2 binding domain in Figure 34; and v) a variable heavy chain domain selected from any of the variable heavy chain domains in Figure 162, and a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239.

[0051] In some embodiments, the first and second Fc domains are variant Fc domains. In certain embodiments, the first and second Fc domains comprise a set of heterodimerized scubariants selected from the group consisting of S364K / E357Q:L368D / K370S; S364K:L368D / K370S; S364K:L368E / K370S; D401K:T411E / K360E / Q362E; and T366W:T366S / L368A / Y407V, where the numbering is according to EU numbering. In some embodiments, the first and second Fc domains comprise the heterodimerized scubariant S364K / E357Q:L368D / K370S.

[0052] In some embodiments, the first and second Fc domains each comprise one or more reduced variants, hi some embodiments, the one or more reduced variants are E233P / L234V / L235A / G236del / S267K, where the numbering is according to EU numbering.

[0053] In some embodiments, one of the first or second monomers comprises one or more pI variants, in some embodiments, the CH1-hinge-CH2-CH3 of the second monomer comprises the pI variants N208D / Q295E / N384D / Q418E / N421D, where the numbering is according to EU numbering.

[0054] In some embodiments, the first Fc domain comprises the amino acid variants S364K / E357Q / E233P / L234V / L235A / G236del / S267K and the CH1-hinge-CH2-CH3 of the second monomer comprises the amino acid variants L368D / K370S / N208D / Q295E / N384D / Q418E / N421D / E233P / L234V / L235A / G236del / S267K, where numbering is according to EU numbering.

[0055] In some embodiments, the scFv linker is a charged scFv linker having the amino acid sequence (GKPGS) 4. In some embodiments, the first and second Fc domains each further comprise the amino acid variant 428 / 434S.

[0056] In another aspect, provided herein are methods of treating cancer, comprising administering to a patient in need of cancer treatment a subject antibody disclosed herein or a pharmaceutical composition comprising a subject antibody and a pharmaceutically acceptable carrier. In some embodiments, the method further comprises administering to the patient an anti-CD3 x tumor target antigen (TTA) bispecific antibody.

[0057] Also provided herein are nucleic acid compositions encoding the compositions and antibodies provided herein, expression vectors containing such nucleic acids, and host cells containing the nucleic acids and expression vectors.

[0058] In another aspect, provided herein are methods of treating cancer, comprising administering one of the subject antibodies provided herein to a patient in need of cancer treatment. In some embodiments, the method further comprises administering to the patient an anti-CD3 x tumor target antigen bispecific antibody.

[0059] In another aspect, provided herein is a composition comprising a CD28 antigen binding domain, wherein the CD28 binding domain comprises: a) a variable heavy chain domain having vhCDR1-3 of the variable heavy chain domain of Figure 163; and b) a variable light chain domain having vlCDR1-3 of the variable light chain domain of Figures 35 and 37, or the variable light chain domain having the amino acid sequence of SEQ ID NO: 3239.

[0060] In another aspect, provided herein is a CD28 antigen binding domain, the CD28 binding domain comprising: a) a variable heavy chain domain having at least 85%, 90%, 95%, or 99% sequence identity to the variable heavy chain domain of Figure 163; and b) a variable light chain domain having at least 85%, 90%, 95%, or 99% sequence identity to the variable light chain domain of Figures 35 and 37, or a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239. In some embodiments, the variable heavy chain domain has the amino acid sequence of the variable heavy chain domain of Figure 163, and the variable light chain domain has the amino acid sequence of the variable light chain domain of Figures 35 and 37, or a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239.

[0061] In another aspect, provided herein is a composition comprising a PD-L1 antigen-binding domain, wherein the PD-L1-binding domain comprises: a) a variable heavy chain domain having vhCDR1-3 of the variable heavy chain domain in Figure 161; and b) a variable light chain domain having vlCDR1-3 of the variable light chain domain having the amino acid sequence of SEQ ID NO: 3239.

[0062] In another aspect, provided herein are compositions comprising a PD-L1 antigen-binding domain, wherein the PD-L1 binding domain comprises: a) a variable heavy chain domain having at least 85%, 90%, 95%, or 99% sequence identity to the variable heavy chain domain of Figure 161; and b) a variable light chain domain having at least 85%, 90%, 95%, or 99% sequence identity to a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239. In some embodiments, the variable heavy chain domain has the amino acid sequence of the variable heavy chain domain of Figure 161, and the variable light chain domain has the amino acid sequence of the variable light chain domain having the amino acid sequence of SEQ ID NO: 3239.

[0063] In another aspect, provided herein is a composition comprising a PD-L2 antigen-binding domain, wherein the PD-L1-binding domain comprises: a) a variable heavy chain domain having vhCDR1-3 of the variable heavy chain domain in Figure 162; and b) a variable light chain domain having vlCDR1-3 of the variable light chain domain having the amino acid sequence of SEQ ID NO: 3239.

[0064] In another aspect, provided herein is a composition comprising a PD-L2 antigen-binding domain, wherein the PD-L1 binding domain comprises a) a variable heavy chain domain having at least 85% sequence identity to the variable heavy chain domain of Figure 162, and b) a variable light chain domain having at least 85% sequence identity to a variable light chain domain having the amino acid sequence of SEQ ID NO: 3239. In some embodiments, the variable heavy chain domain has the amino acid sequence of the variable heavy chain domain of Figure 162, and the variable light chain domain has the amino acid sequence of the variable light chain domain having the amino acid sequence of SEQ ID NO: 3239. [Brief explanation of the drawings]

[0065] [Figure 1] The sequences of human, mouse, and cynomolgus monkey PDL1 are shown, which are useful for the development of cross-reactive PDL1 antigen-binding domains to facilitate clinical development. [Figure 2] The sequences of human, mouse, and cynomolgus PDL2 are shown. Such PDL2 is useful for the development of cross-reactive PDL2 antigen-binding domains to facilitate clinical development. [Figure 3A] The sequences of human, mouse, and cynomolgus CD28 are shown. Such CD28s are useful for the development of cross-reactive CD28 antigen-binding domains to facilitate clinical development. [Figure 3B] The sequences of human, mouse, and cynomolgus CD28 are shown. Such CD28s are useful for the development of cross-reactive CD28 antigen-binding domains to facilitate clinical development. [Figure 4A]Useful pairs of heterodimerization variant sets (including skew and pI variants) are shown. In Figure 4F, variants are present for which no corresponding "monomer 2" variant exists. Such variants can be used alone on either the monomers of a bispecific antibody (e.g., αPD-L1×αCD28 bsAb) or trispecific antibody (e.g., PDL1×PDL2×CD28 triAb), or are pI variants that can be included on the non-scFv side of a format that utilizes an scFv with a suitable charged scFv linker as a component on the second monomer (suitable charged linkers are shown in Figure 7). The heterodimer yield (%) and CH3 Tm (°C) of preferred Fc heterodimerization variants have been previously described (see, e.g., Figure 8 of U.S. Patent Application Publication No. 2019 / 0248898). [Figure 4B] Useful pairs of heterodimerization variant sets (including skew and pI variants) are shown. In Figure 4F, variants are present for which no corresponding "monomer 2" variant exists. Such variants can be used alone on either the monomers of a bispecific antibody (e.g., αPD-L1×αCD28 bsAb) or trispecific antibody (e.g., PDL1×PDL2×CD28 triAb), or are pI variants that can be included on the non-scFv side of a format that utilizes an scFv with a suitable charged scFv linker as a component on the second monomer (suitable charged linkers are shown in Figure 7). The heterodimer yield (%) and CH3 Tm (°C) of preferred Fc heterodimerization variants have been previously described (see, e.g., Figure 8 of U.S. Patent Application Publication No. 2019 / 0248898). [Figure 4C]Useful pairs of heterodimerization variant sets (including skew and pI variants) are shown. In Figure 4F, variants are present for which no corresponding "monomer 2" variant exists. Such variants can be used alone on either the monomers of a bispecific antibody (e.g., αPD-L1×αCD28 bsAb) or trispecific antibody (e.g., PDL1×PDL2×CD28 triAb), or are pI variants that can be included on the non-scFv side of a format that utilizes an scFv with a suitable charged scFv linker as a component on the second monomer (suitable charged linkers are shown in Figure 7). The heterodimer yield (%) and CH3 Tm (°C) of preferred Fc heterodimerization variants have been previously described (see, e.g., Figure 8 of U.S. Patent Application Publication No. 2019 / 0248898). [Figure 4D] Useful pairs of heterodimerization variant sets (including skew and pI variants) are shown. In Figure 4F, variants are present for which no corresponding "monomer 2" variant exists. Such variants can be used alone on either the monomers of a bispecific antibody (e.g., αPD-L1×αCD28 bsAb) or trispecific antibody (e.g., PDL1×PDL2×CD28 triAb), or are pI variants that can be included on the non-scFv side of a format that utilizes an scFv with a suitable charged scFv linker as a component on the second monomer (suitable charged linkers are shown in Figure 7). The heterodimer yield (%) and CH3 Tm (°C) of preferred Fc heterodimerization variants have been previously described (see, e.g., Figure 8 of U.S. Patent Application Publication No. 2019 / 0248898). [Figure 4E]Useful pairs of heterodimerization variant sets (including skew and pI variants) are shown. In Figure 4F, variants are present for which no corresponding "monomer 2" variant exists. Such variants can be used alone on either the monomers of a bispecific antibody (e.g., αPD-L1×αCD28 bsAb) or trispecific antibody (e.g., PDL1×PDL2×CD28 triAb), or are pI variants that can be included on the non-scFv side of a format that utilizes an scFv with a suitable charged scFv linker as a component on the second monomer (suitable charged linkers are shown in Figure 7). The heterodimer yield (%) and CH3 Tm (°C) of preferred Fc heterodimerization variants have been previously described (see, e.g., Figure 8 of U.S. Patent Application Publication No. 2019 / 0248898). [Figure 4F] Useful pairs of heterodimerization variant sets (including skew and pI variants) are shown. In Figure 4F, variants are present for which no corresponding "monomer 2" variant exists. Such variants can be used alone on either the monomers of a bispecific antibody (e.g., αPD-L1×αCD28 bsAb) or trispecific antibody (e.g., PDL1×PDL2×CD28 triAb), or are pI variants that can be included on the non-scFv side of a format that utilizes an scFv with a suitable charged scFv linker as a component on the second monomer (suitable charged linkers are shown in Figure 7). The heterodimer yield (%) and CH3 Tm (°C) of preferred Fc heterodimerization variants have been previously described (see, e.g., Figure 8 of U.S. Patent Application Publication No. 2019 / 0248898). [Figure 5] A list of isosteric variant antibody constant regions and their respective substitutions is shown. pI_(-) indicates a lower pI variant, while pI_(+) indicates a higher pI variant. These variants can optionally and independently be combined with other variants, including heterodimerization variants, as outlined herein. [Figure 6]Useful reduced variants (also referred to as "knockout" or "KO" variants) that reduce FcγR binding are shown. In some embodiments, such reduced variants are contained in the Fc domains of both monomers of the subject antibodies described herein. In other embodiments, the reduced variant is contained in only one variant Fc domain. [Figure 7] As described herein, numerous charged scFv linkers are shown for use in increasing or decreasing the pI of the subject multimeric bispecific and trispecific antibodies (e.g., PDL1xCD28 bsAb) that utilize one or more scFv components. (+H) Positive linkers are particularly utilized herein, particularly with the anti-CD28 VL and VH sequences shown herein. A single prior art scFv linker bearing a single charge is referred to as "Whitlow" after Whitlow et al., Protein Engineering 6(8):989-995 (1993). It should be noted that this linker was used in scFvs to reduce aggregation and improve proteolytic stability. Such charged scFv linkers can be used in any of the subject antibody formats disclosed herein that include scFvs (e.g., 1+1 Fab-scFv-Fc, 2+1 Fab2-scFv-Fc formats, etc.). [Figure 8A] A number of exemplary domain linkers are shown. In some embodiments, these linkers are used to link a single-chain Fv to an Fc chain. In some embodiments, these linkers can be combined in any orientation. For example, a GGGGS linker can be combined with a "lower half-hinge" linker at the N-terminus or C-terminus. [Figure 8B] A number of exemplary domain linkers are shown. In some embodiments, these linkers are used to link a single-chain Fv to an Fc chain. In some embodiments, these linkers can be combined in any orientation. For example, a GGGGS linker can be combined with a "lower half-hinge" linker at the N-terminus or C-terminus. [Figure 9]Particularly useful embodiments of the heterodimeric Fc domains (i.e., CH2-CH3 in this embodiment) of the PDL1xCD28 and PDL2xCD28 bsAbs, and PDL1xPDL2xCD28 triAb of the present invention are shown. [Figure 10]

[0039] Figure 1 shows the sequences of several useful multimeric PDL1xCD28 and PDL2xCD28 bispecific antibody (bsAb) or PDL1xPDL2xCD28 trispecific antibody (triAb) scaffolds based on human IgG1, lacking cytokine sequences. Heterodimeric Fc scaffold 1 is based on human IgG1 (356E / 358M allotype) and contains the L368D / K370S scuba variant and Q295E / N384D / Q418E / N421D pI variants on the first heterodimeric Fc chain, the S364K / E357Q scuba variant on the second heterodimeric Fc chain, and the E233P / L234V / L235A / G236del / S267K reduced variants on both chains. Heterodimeric Fc scaffold 2 is based on human IgG1 (356E / 358M allotype) and comprises the L368D / K370S scuba variant and the Q295E / N384D / Q418E / N421D pI variants on the first heterodimeric Fc chain, the S364K scuba variant on the second heterodimeric Fc chain, and the E233P / L234V / L235A / G236del / S267K reduced variants on both chains. Heterodimeric Fc scaffold 3 is based on human IgG1 (356E / 358M allotype) and contains the L368E / K370S scuba variant and Q295E / N384D / Q418E / N421D pI variants on the first heterodimeric Fc chain, the S364K scuba variant on the second heterodimeric Fc chain, and the E233P / L234V / L235A / G236del / S267K reduced variants on both chains. Heterodimeric Fc scaffold 4 is based on human IgG1 (356E / 358M allotype) and contains K360E / Q362E / T411E scubariant and Q295E / N384D / Q418E / N421D pI variants on the first heterodimeric Fc chain, a D401K scubariant on the second heterodimeric Fc chain, and E233P / L234V / L235A / G236del / S267K reduced variants on both chains.Heterodimeric Fc scaffold 5 is based on human IgG1 (356D / 358L allotype) and comprises the L368D / K370S scuba variant and the Q295E / N384D / Q418E / N421D pI variants on the first heterodimeric Fc chain, the S364K / E357Q scuba variant on the second heterodimeric Fc chain, and the E233P / L234V / L235A / G236del / S267K reduced variants on both chains. Heterodimeric Fc scaffold 6 is based on human IgG1 (356E / 358M allotype) and comprises an L368D / K370S scuba variant and a Q295E / N384D / Q418E / N421D pI variant on the first heterodimeric Fc chain, an S364K / E357Q scuba variant on the second heterodimeric Fc chain, and E233P / L234V / L235A / G236del / S267K reduced variants and an N297A variant that eliminates glycosylation on both chains. Heterodimeric Fc scaffold 7 is based on human IgG1 (356E / 358M allotype) and comprises an L368D / K370S scuba variant and a Q295E / N384D / Q418E / N421D pI variant on the first heterodimeric Fc chain, an S364K / E357Q scuba variant on the second heterodimeric Fc chain, and E233P / L234V / L235A / G236del / S267K reduced variants and an N297S variant that eliminates glycosylation on both chains. Heterodimeric Fc scaffold 8 is based on human IgG4 and comprises the L368D / K370S scuba variant and the Q295E / N384D / Q418E / N421D pI variant on the first heterodimeric Fc chain, the S364K / E357Q scuba variant on the second heterodimeric Fc chain, and the S228P (S241P according to EU numbering, Kabat) variant, which reduces Fab arm exchange (as known in the art), on both chains. Heterodimeric Fc scaffold 9 is based on human IgG2 and comprises the L368D / K370S scuba variant and the Q295E / N384D / Q418E / N421D pI variant on the first heterodimeric Fc chain, and the S364K / E357Q scuba variant on the second heterodimeric Fc chain.The heterodimeric Fc scaffold 10 is based on human IgG2 and comprises an L368D / K370S scuba variant and a Q295E / N384D / Q418E / N421D pI variant on the first heterodimeric Fc chain, an S364K / E357Q scuba variant on the second heterodimeric Fc chain, and an S267K reduced variant on both chains. Heterodimeric Fc scaffold 11 is based on human IgG1 (356E / 358M allotype) and comprises an L368D / K370S scuba variant and a Q295E / N384D / Q418E / N421D pI variant on the first heterodimeric Fc chain, an S364K / E357Q scuba variant on the second heterodimeric Fc chain, and E233P / L234V / L235A / G236del / S267K reduced variants and an M428L / N434S extended variant on both chains. Heterodimeric Fc scaffold 12 is based on human IgG1 (356E / 358M allotype) and comprises an L368D / K370S variant on the first heterodimeric Fc chain, an S364K / E357Q variant and a P217R / P229R / N276K pI variant on the second heterodimeric Fc chain, and E233P / L234V / L235A / G236del / S267K reduced variants on both chains. Heterodimeric Fc scaffold 13 is based on human IgG1 (356D / 358L allotype) and comprises an L368D / K370S scuba variant and a Q295E / N384D / Q418E / N421D pI variant on the first heterodimeric Fc chain, an S364K / E357Q scuba variant on the second heterodimeric Fc chain, and E233P / L234V / L235A / G236del / S267K reduced variants and an M428L / N434S extended variant on both chains. Heterodimeric Fc scaffold 14 is based on human IgG1 (356E / 358M allotype) and comprises an L368D / K370S scuba variant and a Q295E / N384D / Q418E / N421D pI variant on the first heterodimeric Fc chain, an S364K / E357Q scuba variant on the second heterodimeric Fc chain, and E233P / L234V / L235A / G236del / S267K reduced variants and an M428L / N434A extended variant on both chains.Heterodimeric Fc scaffold 15 is based on human IgG1 (356D / 358L allotype) and comprises an L368D / K370S scuba variant and a Q295E / N384D / Q418E / N421D pI variant on the first heterodimeric Fc chain, an S364K / E357Q scuba variant on the second heterodimeric Fc chain, and E233P / L234V / L235A / G236del / S267K reduced variants and an M428L / N434A Xtend variant on both chains. Included in each of these scaffolds are sequences that are 90, 95, 98, and 99% identical (as defined herein) to the depicted sequences and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more additional amino acid substitutions (as compared to the parent human IgG1 (or IgG2 or IgG4, depending on the scaffold), which, as will be understood by those skilled in the art, already contain multiple amino acid modifications when compared to the "parent" in the diagram). That is, the depicted scaffolds can contain additional amino acid modifications (usually amino acid substitutions) in addition to or as an alternative to the skew, pI, and reduced variants contained within the scaffolds in this diagram. The scaffolds shown herein can also include a C-terminal glycine (K446_) and / or lysine (K447_) deletion. C-terminal glycine and / or lysine deletions can be engineered to reduce heterogeneity or in the context of a given bispecific format, such as a mAb-scFv format. C-terminal glycine and / or lysine deletions can also occur naturally, for example, during production and storage. Additionally, these sequences may contain the H435R / Y436F variant in either monomer 1 or monomer 2 to facilitate purification. [Figure 11]Exemplary sequences of scaffolds for use in 2+1 mAb-scFv and 1+1+1 mAb-scFv formats are shown. The formats shown here are based on heterodimeric Fc scaffold 1 shown in Figure X, except that it further includes G446 on monomer 1 (-) and G446 / K447 on monomer 2 (+). It should be noted that any of the additional scaffolds shown in Figure X can be adapted for use in 2+1 mAb-scFv formats with or without K447 on one or both chains. It should be noted that these sequences may further include the M428L / N434S variant. Additionally, these sequences may include the H435R / Y436F variant in either monomer 1 or monomer 2 to facilitate purification. [Figure 12] 1 shows the sequence of "CH1" utilized in bsAb and triAb embodiments of the present invention. [Figure 13] 1 shows the sequence of the "hinge" utilized in bsAb and triAb embodiments of the present invention. [Figure 14] The constant domains of the cognate light chains utilized in the subject PDL1xCD28 or PDL2xCD28 bsAb and PDL1xPDL2xCD28 triAb utilizing Fab binding domains are shown. [Figure 15A]

[0033] Figure 6 shows the sequence of an exemplary anti-CD3 binding domain suitable for use in a CD3 bispecific antibody that may be combined with a CD28 bispecific or trispecific antibody of the present invention. The CDRs are underlined, the scFv linker is double underlined (in the sequence, the scFv linker is a positively charged scFv(GKPGS)4 linker (SEQ ID NO: 892), but as will be understood by those skilled in the art, this linker can be replaced with other linkers, including uncharged or negatively charged linkers, some of which are shown in Figure 6), and the slash indicates the boundary(s) of the variable domain. Additionally, the nomenclature indicates the orientation of the scFv from N-terminus to C-terminus. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within the VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either an scFv format or in a Fab format. [Figure 15B]

[0033] Figure 6 shows the sequence of an exemplary anti-CD3 binding domain suitable for use in a CD3 bispecific antibody that may be combined with a CD28 bispecific or trispecific antibody of the present invention. The CDRs are underlined, the scFv linker is double underlined (in the sequence, the scFv linker is a positively charged scFv(GKPGS)4 linker (SEQ ID NO: 892), but as will be understood by those skilled in the art, this linker can be replaced with other linkers, including uncharged or negatively charged linkers, some of which are shown in Figure 6), and the slash indicates the boundary(s) of the variable domain. Additionally, the nomenclature indicates the orientation of the scFv from N-terminus to C-terminus. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within the VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either an scFv format or in a Fab format. [Figure 15C]

[0033] Figure 6 shows the sequence of an exemplary anti-CD3 binding domain suitable for use in a CD3 bispecific antibody that may be combined with a CD28 bispecific or trispecific antibody of the present invention. The CDRs are underlined, the scFv linker is double underlined (in the sequence, the scFv linker is a positively charged scFv(GKPGS)4 linker (SEQ ID NO: 892), but as will be understood by those skilled in the art, this linker can be replaced with other linkers, including uncharged or negatively charged linkers, some of which are shown in Figure 6), and the slash indicates the boundary(s) of the variable domain. Additionally, the nomenclature indicates the orientation of the scFv from N-terminus to C-terminus. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within the VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either an scFv format or in a Fab format. [Figure 15D]

[0033] Figure 6 shows the sequence of an exemplary anti-CD3 binding domain suitable for use in a CD3 bispecific antibody that may be combined with a CD28 bispecific or trispecific antibody of the present invention. The CDRs are underlined, the scFv linker is double underlined (in the sequence, the scFv linker is a positively charged scFv(GKPGS)4 linker (SEQ ID NO: 892), but as will be understood by those skilled in the art, this linker can be replaced with other linkers, including uncharged or negatively charged linkers, some of which are shown in Figure 6), and the slash indicates the boundary(s) of the variable domain. Additionally, the nomenclature indicates the orientation of the scFv from N-terminus to C-terminus. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within the VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either an scFv format or in a Fab format. [Figure 15E]

[0033] Figure 6 shows the sequence of an exemplary anti-CD3 binding domain suitable for use in a CD3 bispecific antibody that may be combined with a CD28 bispecific or trispecific antibody of the present invention. The CDRs are underlined, the scFv linker is double underlined (in the sequence, the scFv linker is a positively charged scFv(GKPGS)4 linker (SEQ ID NO: 892), but as will be understood by those skilled in the art, this linker can be replaced with other linkers, including uncharged or negatively charged linkers, some of which are shown in Figure 6), and the slash indicates the boundary(s) of the variable domain. Additionally, the nomenclature indicates the orientation of the scFv from N-terminus to C-terminus. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within the VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either an scFv format or in a Fab format. [Figure 15F]

[0033] Figure 6 shows the sequence of an exemplary anti-CD3 binding domain suitable for use in a CD3 bispecific antibody that may be combined with a CD28 bispecific or trispecific antibody of the present invention. The CDRs are underlined, the scFv linker is double underlined (in the sequence, the scFv linker is a positively charged scFv(GKPGS)4 linker (SEQ ID NO: 892), but as will be understood by those skilled in the art, this linker can be replaced with other linkers, including uncharged or negatively charged linkers, some of which are shown in Figure 6), and the slash indicates the boundary(s) of the variable domain. Additionally, the nomenclature indicates the orientation of the scFv from N-terminus to C-terminus. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within the VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either an scFv format or in a Fab format. [Figure 16] The sequences of exemplary αPSMA×αCD3 bsAbs in a 2+1 Fab2-scFv-Fc format are shown, each containing the H1.30_L1.47 anti-CD3 scFv (also known as CD3 High [VHVL]) or L1.47_H1.32 anti-CD3 scFv (also known as CD3 High-Int#1 [VLVH]). CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other chain components (e.g., constant region and domain linker). It should be noted that αPSMA×αCD3 bsAbs can utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each of the sequences outlined herein may include or exclude the M428L / N434S variant in one or preferably both Fc domains, resulting in an increased serum half-life. [Figure 17]The variable heavy and variable light chain sequences of an exemplary humanized hybridoma-derived PDL1-binding domain, 2G4, are shown, as well as the sequences of XENP25859, an anti-PDL1 mAb based on an IgG1 backbone with 2G4 and the E233P / L234V / L235A / G236del / S267K attenuation variant, and XENP36627, a monovalent anti-PDL1 mAb based on 2G4. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and the constant domain. As is true for all sequences described herein that contain CDRs, the exact location of the CDRs varies slightly depending on the numbering used, as shown in Table 2; therefore, not only the underlined CDRs but also CDRs contained within the VH and VL domains using other numbering systems may be included herein. Furthermore, for all of the sequences in the figures, these VH and VL sequences can be used in either the scFv or Fab format. [Figure 18] PD1:PDL1 blockade (binding of PDL1-mFc fusions to PD1-transfected HEK293T cells) with anti-PDL1 clone 2G4 (XENP25859), partial blocking anti-PDL1 (XENP25853), non-blocking anti-PDL1 (XENP25858), and XENP24118 (avelumab-based benchmark anti-PDL1 mAb) is shown. [Figure 19]

[0023] Figure 1 shows the sequence of the affinity-optimized variable heavy chain domain from anti-PDL1 clone 2G4. Note that the variable heavy chain domain can be paired with any of the other 2G4 variable light chain domains shown herein, including SEQ ID NOS: 1467-1528 (e.g., 2G4_H1.12_L1.14, as utilized in XENP40706). [Figure 20]

[0023] Figure 1 shows the sequence of the affinity-optimized variable light chain domain from anti-PDL1 clone 2G4. Note that the variable heavy chain domain can be paired with any of the other 2G4 variable heavy chain domains shown herein, including SEQ ID NOS: 1529-1599 (e.g., 2G4_H1.12_L1.14, as utilized in XENP40706). [Figure 21]The sequences of exemplary affinity-optimized 2G4 VH / VL pairs are shown. Note that these pairs may be formatted as Fabs or as scFvs. [Figure 22] The consensus framework regions (FR) and complementarity determining regions (CDR) (as in Kabat) of the anti-PDL1 clone 2G4 variable heavy and variable light domain variants are shown. [Figure 23] 1 shows an exemplary affinity engineered 2G4 VH / VL pair and its binding affinity in the context of scFv (in the context of a 1+1 Fab-scFv-Fc bsAb format). [Figure 24] A) The sequence of the common variable light chain domain used in the humanized mouse that was the basis for the PDL1, PDL2, and CD28 common light chain designs (designated 6B1_L1) is shown. This variable light chain domain sequence may be paired with a variable heavy chain domain as shown in Figures 25, 26, 29, 30, 31, 32, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, and 74. [Figure 25] The novel PDL1-binding domain clone 13G1 is shown. The CDRs are underlined, and the slash indicates the boundary(s) between the variable and constant domains. As is true for all sequences described herein that contain CDRs, the exact identification of the CDR positions varies slightly depending on the numbering used, as shown in Table 2; therefore, not only the underlined CDRs but also CDRs contained within the VH and VL domains using other numbering systems may be included herein. Furthermore, for all sequences in the figure, these VH and VL sequences may be used in either scFv or Fab formats. [Figure 26A]Figure 2 shows the novel PDL1-binding domain clone 13G7, the alternative VH 13G7_H2, the alternative VL 13G7_L2, and an exemplary affinity-engineered VH variant (furthermore, it should be noted that the CDRs of the affinity-engineered VH may be grafted onto an alternative parent framework as described in Example 1B). It should be noted that both 13G7_H1 and 13G7_H2, as well as the affinity-engineered 13G7 VH variant, may be further paired with 6B1_L1 as shown in Figure 24. Additionally, the consensus framework regions (FR) and complementarity-determining regions (CDRs) of the 13G7 VH are shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable and constant domains. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions will vary slightly depending on the numbering used as shown in Table 2, and thus not only the underlined CDRs but also CDRs contained within the VH and VL domains using other numbering systems may be included herein. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either an scFv format or in a Fab format. [Figure 26B]Figure 2 shows the novel PDL1-binding domain clone 13G7, the alternative VH 13G7_H2, the alternative VL 13G7_L2, and an exemplary affinity-engineered VH variant (furthermore, it should be noted that the CDRs of the affinity-engineered VH may be grafted onto an alternative parent framework as described in Example 1B). It should be noted that both 13G7_H1 and 13G7_H2, as well as the affinity-engineered 13G7 VH variant, may be further paired with 6B1_L1 as shown in Figure 24. Additionally, the consensus framework regions (FR) and complementarity-determining regions (CDRs) of the 13G7 VH are shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable and constant domains. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions will vary slightly depending on the numbering used as shown in Table 2, and thus not only the underlined CDRs but also CDRs contained within the VH and VL domains using other numbering systems may be included herein. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either an scFv format or in a Fab format. [Figure 26C]Figure 2 shows the novel PDL1-binding domain clone 13G7, the alternative VH 13G7_H2, the alternative VL 13G7_L2, and an exemplary affinity-engineered VH variant (furthermore, it should be noted that the CDRs of the affinity-engineered VH may be grafted onto an alternative parent framework as described in Example 1B). It should be noted that both 13G7_H1 and 13G7_H2, as well as the affinity-engineered 13G7 VH variant, may be further paired with 6B1_L1 as shown in Figure 24. Additionally, the consensus framework regions (FR) and complementarity-determining regions (CDRs) of the 13G7 VH are shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable and constant domains. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions will vary slightly depending on the numbering used as shown in Table 2, and thus not only the underlined CDRs but also CDRs contained within the VH and VL domains using other numbering systems may be included herein. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either an scFv format or in a Fab format. [Figure 26D]Figure 2 shows the novel PDL1-binding domain clone 13G7, the alternative VH 13G7_H2, the alternative VL 13G7_L2, and an exemplary affinity-engineered VH variant (furthermore, it should be noted that the CDRs of the affinity-engineered VH may be grafted onto an alternative parent framework as described in Example 1B). It should be noted that both 13G7_H1 and 13G7_H2, as well as the affinity-engineered 13G7 VH variant, may be further paired with 6B1_L1 as shown in Figure 24. Additionally, the consensus framework regions (FR) and complementarity-determining regions (CDRs) of the 13G7 VH are shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable and constant domains. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions will vary slightly depending on the numbering used as shown in Table 2, and thus not only the underlined CDRs but also CDRs contained within the VH and VL domains using other numbering systems may be included herein. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either an scFv format or in a Fab format. [Figure 27] Figure 1 shows the dissociation constant (KD), association rate (ka), and dissociation rate (kd) of PDL1 antibodies produced by single-cell technology in mice genetically engineered with a fully human heavy chain variable domain combined with a human consensus light chain substitution. [Figure 28]

[0023] Figures 1A-1C show the variable heavy and variable light chain sequences of additional PDL1-binding domains utilized in the PDL1xCD28 bsAb and PDL1xPDL2xCD28 triAb of the present invention. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the present specification includes not only the underlined CDRs, but also CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 29] In addition to the novel PDL2-binding domain clone 5C11, an alternative VH 5C11_H2 is shown. It should be noted that both 5C11_H1 and 5C11_H2 may be further paired with 6B1_L1 as shown in Figure 24. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the present specification includes not only the underlined CDRs, but also CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either scFv or Fab formats. [Figure 30] In addition to the novel PDL2-binding domain clone 8G2, an alternative VH 8G2_H2 is shown. It should be noted that both 8G2_H1 and 8G2_H2 may be further paired with 6B1_L1 as shown in Figure 24. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either scFv or Fab formats. [Figure 31]In addition to the novel PDL2-binding domain clone 8G5, an alternative VH 8G5_H3 is shown. It should be noted that both 8G5_H1 and 8G5_H3 may be further paired with 6B1_L1 as shown in Figure 24. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the present specification includes not only the underlined CDRs, but also CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either scFv or Fab formats. [Figure 32A] In addition to the novel PDL2-binding domain clone 16G11, alternative VHs 16G11_H2 and 16G11_H3, as well as exemplary affinity-engineered VH variants, are shown (furthermore, it should be noted that the CDRs of the affinity-engineered VH may be grafted onto alternative parent frameworks as described in Example 1B). It should be noted that 16G11_H1, 16G11_H2, and 16G11_H3, as well as the affinity-engineered 16G11 VH variants, may be further paired with 6B1_L1 as shown in Figure 24. Additionally, the consensus framework regions (FRs) and complementarity-determining regions (CDRs) of the 16G11 VH are shown. As is true for all sequences herein described and containing CDRs, the precise identification of CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the present specification includes not only the underlined CDRs, but also CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either an scFv format or in a Fab format. [Figure 32B]In addition to the novel PDL2-binding domain clone 16G11, alternative VHs 16G11_H2 and 16G11_H3, as well as exemplary affinity-engineered VH variants, are shown (furthermore, it should be noted that the CDRs of the affinity-engineered VH may be grafted onto alternative parent frameworks as described in Example 1B). It should be noted that 16G11_H1, 16G11_H2, and 16G11_H3, as well as the affinity-engineered 16G11 VH variants, may be further paired with 6B1_L1 as shown in Figure 24. Additionally, the consensus framework regions (FRs) and complementarity-determining regions (CDRs) of the 16G11 VH are shown. As is true for all sequences herein described and containing CDRs, the precise identification of CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the present specification includes not only the underlined CDRs, but also CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either an scFv format or in a Fab format. [Figure 32C]In addition to the novel PDL2-binding domain clone 16G11, alternative VHs 16G11_H2 and 16G11_H3, as well as exemplary affinity-engineered VH variants, are shown (furthermore, it should be noted that the CDRs of the affinity-engineered VH may be grafted onto alternative parent frameworks as described in Example 1B). It should be noted that 16G11_H1, 16G11_H2, and 16G11_H3, as well as the affinity-engineered 16G11 VH variants, may be further paired with 6B1_L1 as shown in Figure 24. Additionally, the consensus framework regions (FRs) and complementarity-determining regions (CDRs) of the 16G11 VH are shown. As is true for all sequences herein described and containing CDRs, the precise identification of CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the present specification includes not only the underlined CDRs, but also CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either an scFv format or in a Fab format. [Figure 32D]In addition to the novel PDL2-binding domain clone 16G11, alternative VHs 16G11_H2 and 16G11_H3, as well as exemplary affinity-engineered VH variants, are shown (furthermore, it should be noted that the CDRs of the affinity-engineered VH may be grafted onto alternative parent frameworks as described in Example 1B). It should be noted that 16G11_H1, 16G11_H2, and 16G11_H3, as well as the affinity-engineered 16G11 VH variants, may be further paired with 6B1_L1 as shown in Figure 24. Additionally, the consensus framework regions (FRs) and complementarity-determining regions (CDRs) of the 16G11 VH are shown. As is true for all sequences herein described and containing CDRs, the precise identification of CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the present specification includes not only the underlined CDRs, but also CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either an scFv format or in a Fab format. [Figure 33] Figure 1 shows the dissociation constant (KD), association rate (ka), and dissociation rate (kd) of PDL2 antibodies produced by single cell technology in mice genetically engineered with a fully human heavy chain variable domain combined with a human consensus light chain substitution. [Figure 34]

[0023] Figures 1A-1C show the variable heavy and variable light chain sequences of additional PDL2-binding domains utilized in the PDL2xCD28 bsAb and PDL1xPDL2xCD28 triAb of the present invention. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the present specification includes not only the underlined CDRs, but also CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 35]The variable heavy and variable light chain sequences of 1A7, an exemplary non-superagonist phage-derived CD28 binding domain, are shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences may be used in either scFv or Fab formats. [Figure 36]

[0033] Figure 3 shows the sequence of the affinity-optimized variable heavy chain domain from anti-CD28 clone 1A7. It should be noted that the variable heavy chain domain can be paired with any of the other variable light chain domains shown in Figures 35 and 37, including SEQ ID NOs: 2456-2524. [Figure 37]

[0033] Figure 3 shows the sequence of the affinity-optimized variable light chain domain from anti-CD28 clone 1A7. It should be noted that the variable light chain domain can be paired with any of the other variable light chain domains shown in Figures 35 and 36, including SEQ ID NOs: 2525-2630. [Figure 38A] The sequences of exemplary affinity-optimized 1A7 VH / VL pairs are shown. Note that these pairs may be formatted as Fabs or as scFvs. In the scFv format, these pairs may also be formatted in a VHVL or VLVH orientation. [Figure 38B] The sequences of exemplary affinity-optimized 1A7 VH / VL pairs are shown. Note that these pairs may be formatted as Fabs or as scFvs. In the scFv format, these pairs may also be formatted in a VHVL or VLVH orientation. [Figure 38C] The sequences of exemplary affinity-optimized 1A7 VH / VL pairs are shown. Note that these pairs may be formatted as Fabs or as scFvs. In the scFv format, these pairs may also be formatted in a VHVL or VLVH orientation. [Figure 39A] The consensus framework regions (FR) and complementarity determining regions (CDR) (according to Kabat) of the anti-CD28 clone 1A7 variable heavy and variable light domain variants are shown. [Figure 39B] The consensus framework regions (FR) and complementarity determining regions (CDR) (according to Kabat) of the anti-CD28 clone 1A7 variable heavy and variable light domain variants are shown. [Figure 40] 1 shows an exemplary affinity engineered 1A7 VH / VL pair and its binding affinity in the context of scFv (in the context of a 1+1 Fab-scFv-Fc bsAb format). [Figure 41] In addition to the novel CD28-binding domain clone 1B1-5.88488, the sequences of XENP41791, an anti-CD28 mAb based on 1B1-5.88488, and XENP42418, an anti-CD28 mAb utilizing the VH of 1B1-5.88488 and the 6B1 common VL, are shown. As is true for all sequences herein described and containing CDRs, the precise identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the present specification includes not only the underlined CDRs, but also CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 42]In addition to the novel CD28-binding domain clone 1B1-4.88488, the sequences of XENP41792, an anti-CD28 mAb based on 1B1-4.88488, and XENP42419, an anti-CD28 mAb utilizing the VH of 1B1-4.88488 and the 6B1 common VL, are shown. As is true for all sequences herein described and containing CDRs, the precise identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 43] In addition to the novel CD28-binding domain clone 1D8.88474, the sequences of XENP41834, an anti-CD28 mAb based on 1D8.88474, and XENP42420, an anti-CD28 mAb utilizing the VH of 1D8.88474 and the 6B1 consensus VL, are shown. As is true for all sequences herein described and containing CDRs, the precise identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the present specification includes not only the underlined CDRs, but also CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 44]In addition to the novel CD28-binding domain clone 1G2-2.88474, the sequences of XENP41846, an anti-CD28 mAb based on 1G2-2.88474, and XENP42421, an anti-CD28 mAb utilizing the VH of 1G2-2.88474 and the 6B1 consensus VL, are shown. As is true for all sequences herein described and containing CDRs, the precise identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the present specification includes not only the underlined CDRs, but also CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 45] In addition to the novel CD28-binding domain clone 2G5.88497, the sequences of XENP41864, an anti-CD28 mAb based on 2G5.88497, and XENP42423, an anti-CD28 mAb utilizing the VH of 2G5.88497 and the 6B1 consensus VL, are shown. As is true for all sequences herein described and containing CDRs, the precise identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the present specification includes not only the underlined CDRs, but also CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 46]In addition to the novel CD28-binding domain clone 2A3.88497, the sequences of XENP41882, an anti-CD28 mAb based on 2A3.88497, and XENP42425, an anti-CD28 mAb utilizing the VH of 2A3.88497 and the 6B1 consensus VL, are shown. As is true for all sequences herein that contain CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 47] In addition to the novel CD28-binding domain clone 1D9-3.83967, the sequences of XENP41907, an anti-CD28 mAb based on 1D9-3.83967, and XENP42426, an anti-CD28 mAb utilizing the VH of 1D9-3.83967 and the 6B1 common VL, are shown. As is true for all sequences herein described and containing CDRs, the precise identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the present specification includes not only the underlined CDRs, but also CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 48]In addition to the novel CD28-binding domain clone 1C9-1.83967, the sequences of XENP41927, an anti-CD28 mAb based on 1C9-1.83967, and XENP42429, an anti-CD28 mAb utilizing the VH of 1C9-1.83967 and the 6B1 consensus VL, are shown. As is true for all sequences herein described and containing CDRs, the precise identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the present specification includes not only the underlined CDRs, but also CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 49] In addition to the novel CD28-binding domain clone 1D10-4.83967, the sequences of XENP41936, an anti-CD28 mAb based on 1D10-4.83967, and XENP42430, an anti-CD28 mAb utilizing the VH of 1D10-4.83967 and the 6B1 consensus VL, are shown. As is true for all sequences herein that contain CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 50]The sequence of the novel CD28-binding domain clone 1A12.83967, as well as XENP41957, an anti-CD28 mAb based on 1A12.83967, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 51] The sequence of the novel CD28 binding domain clone 1B11.83967, as well as XENP41949, an anti-CD28 mAb based on 1B11.83967, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 52] The sequence of the novel CD28-binding domain clone 1D10-2.83967, as well as XENP41935, an anti-CD28 mAb based on 1D10-2.83967, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 53]The sequence of the novel CD28-binding domain clone 1D7.83967, as well as XENP41904, an anti-CD28 mAb based on 1D7.83967, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 54] The sequence of the novel CD28-binding domain clone 1D3.83967, as well as XENP41901, an anti-CD28 mAb based on 1D3.83967, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 55] The sequence of the novel CD28-binding domain clone 2B10.88497, as well as XENP41891, an anti-CD28 mAb based on 2B10.88497, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 56]The sequence of the novel CD28-binding domain clone 2B9.88497, as well as XENP41890, an anti-CD28 mAb based on 2B9.88497, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 57] The sequence of the novel CD28-binding domain clone 2B8.88497, as well as XENP41889, an anti-CD28 mAb based on 2B8.88497, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 58] The sequence of the novel CD28-binding domain clone 1G6-1.83967, as well as XENP41877, an anti-CD28 mAb based on 1G6-1.83967, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 59]The sequence of the novel CD28-binding domain clone 1C7.88474, as well as XENP41874, an anti-CD28 mAb based on 1C7.88474, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 60] The sequence of the novel CD28-binding domain clone 1A5-2.88474, as well as XENP41869, an anti-CD28 mAb based on 1A5-2.88474, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 61] The sequence of the novel CD28-binding domain clone 1A2.88474, as well as XENP41868, an anti-CD28 mAb based on 1A2.88474, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 62]The sequence of the novel CD28-binding domain clone 2F5.88497, as well as XENP41860, an anti-CD28 mAb based on 2F5.88497, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 63] The sequence of the novel CD28-binding domain clone 2E9.88497, as well as XENP41858, an anti-CD28 mAb based on 2E9.88497, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 64A] The sequences of the novel CD28-binding domain clone 1A3.88474, as well as XENP41849, an anti-CD28 mAb based on 1A3.88474, and exemplary affinity-engineered VH variants are shown. Additionally, the consensus framework regions (FRs) and complementarity-determining regions (CDRs) of the 1A3.88474 VH are shown. As is true for all sequences described herein that contain CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 64B]The sequences of the novel CD28-binding domain clone 1A3.88474, as well as XENP41849, an anti-CD28 mAb based on 1A3.88474, and exemplary affinity-engineered VH variants are shown. Additionally, the consensus framework regions (FRs) and complementarity-determining regions (CDRs) of the 1A3.88474 VH are shown. As is true for all sequences described herein that contain CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 64C] The sequences of the novel CD28-binding domain clone 1A3.88474, as well as XENP41849, an anti-CD28 mAb based on 1A3.88474, and exemplary affinity-engineered VH variants are shown. Additionally, the consensus framework regions (FRs) and complementarity-determining regions (CDRs) of the 1A3.88474 VH are shown. As is true for all sequences described herein that contain CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 65]The sequence of the novel CD28-binding domain clone 1H11.88474, as well as XENP41816, an anti-CD28 mAb based on 1H11.88474, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 66] The sequence of the novel CD28-binding domain clone 1G1-1.88474, as well as XENP41807, an anti-CD28 mAb based on 1G1-1.88474, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 67] The sequence of the novel CD28-binding domain clone 1E6-1.88474, as well as XENP41802, an anti-CD28 mAb based on 1E6-1.88474, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 68]The sequence of the novel CD28-binding domain clone 1A1-5.88488, as well as XENP41797, an anti-CD28 mAb based on 1A1-5.88488, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 69] The sequence of the novel CD28-binding domain clone 1E2-5.88488, as well as XENP41781, an anti-CD28 mAb based on 1E2-5.88488, is shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figure, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 70] In addition to the novel CD28-binding domain clone 2F6.88497, the sequences of XENP41861, an anti-CD28 mAb based on 2F6.88497, and XENP42422, an anti-CD28 mAb utilizing the VH of 2F6.88497 and the 6B1 common VL, are shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 71]In addition to the novel CD28-binding domain clone 1H2.83967, the sequences of XENP41880, an anti-CD28 mAb based on 1H2.83967, and XENP42424, an anti-CD28 mAb utilizing the VH of 1H2.83967 and the 6B1 consensus VL, are shown. As is true for all sequences herein that contain CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 72] In addition to the novel CD28-binding domain clone 1E4-3.83967, the sequences of XENP41909, an anti-CD28 mAb based on 1E4-3.83967, and XENP42427, an anti-CD28 mAb utilizing the VH of 1E4-3.83967 and the 6B1 consensus VL, are shown. As is true for all sequences herein described and containing CDRs, the precise identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 73]In addition to the novel CD28-binding domain clone 1C2-2.83967, the sequences of XENP41918, an anti-CD28 mAb based on 1C2-2.83967, and XENP42428, an anti-CD28 mAb utilizing the VH of 1C2-2.83967 and the 6B1 consensus VL, are shown. As is true for all sequences herein described and containing CDRs, the precise identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 74] In addition to the novel CD28-binding domain clone 1A11.83967, the sequences of XENP41956, an anti-CD28 mAb based on 1A11.83967, and XENP42431, an anti-CD28 mAb utilizing the VH of 1A11.83967 and the 6B1 consensus VL, are shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences can be used in either scFv or Fab formats. [Figure 75] Figure 1 shows the maximum BLI response of binding by CD28 mAbs based on single-cell technology-derived binding domains (WT sequence compared to VH paired with 6B1_L1) to the CD28 antigen. The data show that many of the binding domains retain CD28 binding when paired with the 6B1_L1 VL, but VHs from some of the clones (e.g., 1B1-4.88488, 2G5.88497, and 1D10-4.83967) demonstrated reduced CD28 binding when paired with 6B1_L1. [Figure 76]Figure 1 shows the maximum BLI response of binding by a CD28 mAb based on an additional single cell technology-derived binding domain against the CD28 antigen. [Figure 77] 1 shows binding to Jurkat cells by a CD28 mAb based on a binding domain derived from single cell technology. [Figure 78] Figure 1 shows binding to Jurkat cells by CD28 mAbs based on binding domains (WT sequence compared to VH paired with 6B1_L1 consensus light chain) derived from single cell technology. [Figure 79] Figure 1 shows the induction of IL2 secretion from CD3-stimulated (100 ng / mL OKT3) purified T cells by CD28 mAb based on the binding domain derived from single cell technology. [Figure 80] Figure 1 shows the induction of IL2 secretion from CD3-stimulated (100 ng / mL OKT3) purified T cells by a CD28 mAb based on a binding domain (WT sequence compared to VH paired with the 6B1_L1 common light chain) derived from single cell technology. [Figure 81]

[0023] Figures 1A-1C show the variable heavy and variable light chain sequences of additional CD28-binding domains utilized in the PDL1xCD28 and PDL2xCD28 bsAbs, and PDL1xPDL2xCD28 triAb of the present invention. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the present specification includes not only the underlined CDRs, but also CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either scFv or Fab formats. [Figure 82A]

[0023] Figure 1 shows a bispecific format of the invention. The "1+1 Fab-scFv-Fc" format has a first Fab arm that binds a first antigen and a second scFv arm that binds a second antigen. The 1+1 Fab-scFv-Fc format includes a first monomer comprising a first heavy chain variable region (VH1) covalently attached (optionally via a linker) to the N-terminus of a first heterodimeric Fc scaffold, a second monomer comprising a single-chain Fv covalently attached (optionally via a linker) to the N-terminus of a second corresponding heterodimeric Fc scaffold, and a third monomer comprising a light chain variable region covalently attached to a light chain constant domain, where the light chain variable region is complementary to VH1. [Figure 82B]

[0023] Figure 1 shows a bispecific format of the invention. This shows a "2+1 Fab2-scFv-Fc" format having a first Fab arm and a second Fab-scFv arm, where the Fab binds a first antigen and the scFv binds a second antigen. The 2+1 Fab2-scFv-Fc format includes a first monomer comprising a first heavy chain variable region (VH1) covalently linked (optionally via a linker) to the N-terminus of a first heterodimeric Fc scaffold, a second monomer comprising a VH1 covalently linked (optionally via a linker) to a single-chain Fv that is covalently linked (optionally via a linker) to the N-terminus of a second, corresponding heterodimeric Fc scaffold, and a third monomer comprising a light chain variable region covalently linked to a light chain constant domain, where the light chain variable region is complementary to VH1. [Figure 82C]

[0023] Figure 1 illustrates a bispecific format of the invention. This shows a "1+1 common light chain" or "1+1 CLC" format, with a first Fc comprising a first Fab arm that binds a first antigen and a second Fc comprising a second Fab arm that binds a second antigen. The 1+1 CLC format includes a first monomer comprising VH1-CH1-hinge-CH2-CH3, a second monomer comprising VH2-CH1-hinge-CH2-CH3, and a third monomer comprising VL-CL. The VL pairs with the VH1 to form a binding domain with a first antigen-binding specificity; the VL pairs with the VH2 to form a binding domain with a second antigen-binding specificity. [Figure 82D]

[0023] Figure 1 illustrates a bispecific format of the invention. This shows a "2+1 common light chain" or "2+1 CLC" format, with a first Fc comprising two Fab arms that each bind a first antigen, and a second Fc comprising one Fab arm that binds a second antigen. The 2+1 CLC format includes a first monomer comprising VH1-CH1-hinge-VH1-CH1-hinge-CH2-CH3, a second monomer comprising VH2-CH1-hinge-CH2-CH3, and a third monomer comprising VL-CL. The VL pairs with the first and second VH1 to form a binding domain with a first antigen-binding specificity; the VL pairs with the VH2 to form a binding domain with a second antigen-binding specificity. [Figure 82E]

[0023] Figure 1 shows a bispecific format of the invention. It shows a "2+1 mAb-scFv" format having a first Fc comprising an N-terminal Fab arm that binds a first antigen and a second Fc comprising an N-terminal Fab arm that binds the first antigen and a C-terminal scFv that binds a second antigen. The 2+1 mAb-scFv format includes a first monomer comprising a VH1-CH1-hinge-CH2-CH3, a second monomer comprising a VH1-CH1-hinge-CH2-CH3-scFv, and a third monomer comprising a VL-CL. The VL pairs with the first and second VH1 to form a binding domain with binding specificity for the first antigen. [Figure 82F] 1 shows bispecific formats of the invention. Additional bispecific formats include F) bi-scFv, G) one-arm scFv-mAb, H) scFv-mAb, I) bispecific mAb, J) one-arm central-scFv, K) mAb-Fv, L) central-Fv, and M) trident. [Figure 82G] 1 shows bispecific formats of the invention. Additional bispecific formats include F) bi-scFv, G) one-arm scFv-mAb, H) scFv-mAb, I) bispecific mAb, J) one-arm central-scFv, K) mAb-Fv, L) central-Fv, and M) trident. [Figure 82H]1 shows bispecific formats of the invention. Additional bispecific formats include F) bi-scFv, G) one-arm scFv-mAb, H) scFv-mAb, I) bispecific mAb, J) one-arm central-scFv, K) mAb-Fv, L) central-Fv, and M) trident. [Figure 82I] 1 shows bispecific formats of the invention. Additional bispecific formats include F) bi-scFv, G) one-arm scFv-mAb, H) scFv-mAb, I) bispecific mAb, J) one-arm central-scFv, K) mAb-Fv, L) central-Fv, and M) trident. [Figure 82J] 1 shows bispecific formats of the invention. Additional bispecific formats include F) bi-scFv, G) one-arm scFv-mAb, H) scFv-mAb, I) bispecific mAb, J) one-arm central-scFv, K) mAb-Fv, L) central-Fv, and M) trident. [Figure 82K] 1 shows bispecific formats of the invention. Additional bispecific formats include F) bi-scFv, G) one-arm scFv-mAb, H) scFv-mAb, I) bispecific mAb, J) one-arm central-scFv, K) mAb-Fv, L) central-Fv, and M) trident. [Figure 82L] 1 shows bispecific formats of the invention. Additional bispecific formats include F) bi-scFv, G) one-arm scFv-mAb, H) scFv-mAb, I) bispecific mAb, J) one-arm central-scFv, K) mAb-Fv, L) central-Fv, and M) trident. [Figure 82M] 1 shows bispecific formats of the invention. Additional bispecific formats include F) bi-scFv, G) one-arm scFv-mAb, H) scFv-mAb, I) bispecific mAb, J) one-arm central-scFv, K) mAb-Fv, L) central-Fv, and M) trident. [Figure 83]Figure 1 shows a trispecific format of the invention. A shows a "1+1+1 stack Fab2-scFv-Fc" format, which comprises, from N- to C-terminus, a first monomer comprising VH1-CH1-linker-VH2-CH1-hinge-CH2-CH3 (where CH2-CH3 is a first heterodimeric Fc domain), a second monomer comprising, from N- to C-terminus, scFv-linker-CH2-CH3 (where CH2-CH3 is a second heterodimeric Fc domain complementary to the first heterodimeric Fc domain, and the scFv has a first antigen specificity), and a third monomer which is a common light chain comprising, from N- to C-terminus, VL-CL (where VL pairs with VH1 to form an antigen-binding domain with a second antigen specificity and pairs with VH2 to form an antigen-binding domain with a third antigen specificity). B indicates a "1+1+1 Fab-(Fab-scFv)-Fc format," which comprises, from N- to C-terminus, a first monomer that is a heavy chain comprising VH1-CH1-hinge-CH2-CH3 (where CH2-CH3 is a first heterodimeric Fc domain), a second monomer that comprises, from N- to C-terminus, VH2-CH1-linker-scFv-linker-CH2-CH3 (where CH2-CH3 is a second heterodimeric Fc domain complementary to the first heterodimeric Fc domain, and the scFv has a first antigen specificity), and a third monomer that is a common light chain comprising, from N- to C-terminus, VL-CL (where VL pairs with VH1 to form an antigen-binding domain with a second antigen specificity and pairs with VH2 to form an antigen-binding domain with a third antigen specificity). C indicates a "1+1+1 mAb-scFv" format, which comprises, from N- to C-terminus, a first monomer that is a heavy chain comprising VH1-CH1-hinge-CH2-CH3 (where CH2-CH3 is a first heterodimeric Fc domain), a second monomer that comprises, from N- to C-terminus, VH2-CH1-hinge-CH2-CH3-linker-scFv (where CH2-CH3 is a second heterodimeric Fc domain complementary to the first heterodimeric Fc domain, wherein the scFv has a first antigen specificity), and a third monomer that is a common light chain comprising, from N- to C-terminus, VL-CL (where VL pairs with VH1 to form an antigen-binding domain with a second antigen specificity and pairs with VH2 to form an antigen-binding domain with a third antigen specificity).D indicates a "1+1+1 stacked Fab2-Fab-Fc" format, which comprises, from N- to C-terminus, a first monomer that is a heavy chain comprising VH1-CH1-hinge-CH2-CH3 (where CH2-CH3 is a first heterodimeric Fc domain), a second monomer that comprises, from N- to C-terminus, VH2-CH1-linker-VH3-CH1-hinge-CH2-CH3 (where CH2-CH3 is a second heterodimeric Fc domain complementary to the first heterodimeric Fc domain), and a third monomer that is a common light chain comprising, from N- to C-terminus, VL-CL (where VL pairs with VH1 to form an antigen-binding domain having a first antigen specificity, pairs with VH2 to form an antigen-binding domain having a second antigen specificity, and pairs with VH3 to form an antigen-binding domain having a third antigen specificity). [Figure 84A] The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84B]The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84C] The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84D]The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84E] The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84F]The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84G] The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84H]The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84I] The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84J]The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84K] The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84L]The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84M] The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84N]The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84O] The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84P]The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84Q] The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84R-1]The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84R-2] The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84S]The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 84T] The sequence of an exemplary PDL1xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that αPD-L1xαCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, which results in extended serum half-life. [Figure 85]Figure 7 shows the sequence of an exemplary αPD-L1 x αCD28 bsAb in a 2+1 Fab2-scFv-Fc format. The CDRs are underlined, and a slash indicates the boundary(s) between the variable region, Fc region, and constant domain. The scFv domain has a VH-scFv linker-VL orientation (N-terminus to C-terminus), although this can also be reversed. It should be noted that the scFv domain sequence contains the sequence GKPGSGKPGSGKPGSGKPGS (SEQ ID NO: XX) as the scFv linker between the variable heavy and variable light regions, although this linker can be substituted with any of the scFv linkers in Figure 6. It should also be noted that the chain 2 sequence contains the sequence GGGGSGGGGSKTHTCPPCP (SEQ ID NO: XX) as the domain linker between the C-terminus of the scFv and the N-terminus of the CH2 domain, which is a "flexible half-hinge" domain linker, although this linker can be substituted with any of the "useful domain linkers" in Figure 7. It should be noted that αPD-L1×αCD28 bsAbs may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each sequence outlined herein may include or exclude the M428L / N434S variant within one or preferably both Fc domains, resulting in extended serum half-life. [Figure 86]The sequence of an exemplary PDL1xCD28 bsAb in a 2+1 mAb-scFv format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region, linker, Fc region, and constant domain. The scFv domains have a VH-scFv linker-VL orientation (N-terminus to C-terminus), although this can also be reversed. Note that the chain 2 sequence contains the sequence GKPGSGKPGSGKPGSGKPGS (SEQ ID NO: XX) as the domain linker, but this linker can be replaced with any domain linker, including any of the "useful domain linkers" in Figure 6. Note that αPDL1xαCD28 bsAbs can utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid substitutions. In addition, each of the sequences outlined herein may include or exclude the M428L / N434S variant in one or preferably both Fc domains, resulting in an increased serum half-life. [Figure 87A] The sequence of an exemplary PDL2xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that PDL2xCD28 bsAb may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant in one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 87B]The sequence of an exemplary PDL2xCD28 bsAb in a 1+1 Fab-scFv-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region, Fc region, and constant domain. It should be noted that PDL2xCD28 bsAb may utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant in one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 88A] The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stack Fab2-scFv-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAb utilizes variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 88B]The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stack Fab2-scFv-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAb utilizes variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 88C] The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stack Fab2-scFv-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAb utilizes variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 88D]The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stack Fab2-scFv-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAb utilizes variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 88E] The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stack Fab2-scFv-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAb utilizes variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 88F]The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stack Fab2-scFv-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAb utilizes variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 88G] The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stack Fab2-scFv-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAb utilizes variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 88H]The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stack Fab2-scFv-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAb utilizes variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 88I] The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stack Fab2-scFv-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAb utilizes variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 88J]The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stack Fab2-scFv-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAb utilizes variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 88K] The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stack Fab2-scFv-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAb utilizes variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 88L]The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stack Fab2-scFv-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAb utilizes variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 88M] The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stack Fab2-scFv-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAb utilizes variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 89A]The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 Fab-(Fab-scFv)-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAb utilizes variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 89B] The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 Fab-(Fab-scFv)-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAb utilizes variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 90]The sequence of an exemplary PDL1xPDL2xCD28 triAb in a 1+1+1 mAb-scFv format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1xPDL2xCD28 triAbs utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 91A] The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stacked Fab2-Fab-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAbs utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 91B]The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stacked Fab2-Fab-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAbs utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 91C] The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stacked Fab2-Fab-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAbs utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 91D]The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stacked Fab2-Fab-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAbs utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 91E] The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stacked Fab2-Fab-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAbs utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 91F]The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stacked Fab2-Fab-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAbs utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 91G] The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stacked Fab2-Fab-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAbs utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 91H]The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stacked Fab2-Fab-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAbs utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 91I] The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stacked Fab2-Fab-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAbs utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 91J]The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stacked Fab2-Fab-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAbs utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 91K] The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stacked Fab2-Fab-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAbs utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 91L]The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stacked Fab2-Fab-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAbs utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 91M] The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stacked Fab2-Fab-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAbs utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 91N]The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stacked Fab2-Fab-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAbs utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 91O] The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stacked Fab2-Fab-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAbs utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 91P]The sequence of an exemplary PDL1×PDL2×CD28 triAb in a 1+1+1 stacked Fab2-Fab-Fc format is shown. CDRs are underlined, and slashes indicate the boundary(s) between the variable region and other domains (e.g., constant domains). It should be noted that PDL1×PDL2×CD28 triAbs utilize variable region, Fc region, and constant domain sequences that are 90, 95, 98, and 99% identical (as defined herein) and / or may contain 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 or more amino acid substitutions. Additionally, each sequence outlined herein may include or exclude the M428L / N434S variant within one or, preferably, both Fc domains, resulting in extended serum half-life. [Figure 92] A) Classical T cell / APC interaction and B) replication of the classical T cell / APC interaction by combining a CD3 bispecific antibody with a CD28 bispecific antibody. In classical T cell / APC interactions, there is a first signal (signal 1) provided by TCR reactivity with peptide-MHC and a second signal (signal 2) provided by CD28 cross-linking by CD80 / CD86 expressed on the APC, which together are sufficient to activate the T cell. In contrast, treatment with a CD3 bispecific provides only the first signal. A CD28 signal may be provided by a CD28 bispecific with the goal of promoting activation and proliferation via CD28 costimulation. [Figure 93] 1 shows induction of CD28 signaling by a CD28 bispecific antibody and relief of any checkpoint-mediated suppression of the additional CD28 signal by checkpoint blockade (e.g., PD-1 blockade). [Figure 94] We show that the PDL1xCD28 bispecific antibody provides signal 2 while also advantageously allowing for blockade of the PDL1:PD1 interaction. Although not shown, the PDL2xCD28 bispecific antibody may similarly provide signal 2 while also advantageously allowing for blockade of the PDL2:PD1 interaction. [Figure 95]1 shows the sequences of avelumab and XENP24118, a PDL1 mAb based on an IgG1 backbone with the E233P / L234V / L235A / G236del / S267K attenuated variants. CDRs are underlined, and slashes indicate the boundary(s) between the variable and constant domains. [Figure 96] Figure 1 shows the proliferation of CMV+ T cells after incubation of NLV-loaded MDA-MB-231 cancer cells with purified CD3+ T cells from a CMV+ donor and either the αPD-L1 antibody XENP24118 or the αPD-L1×αCD28 bsAb XENP34963. The αPD-L1×αCD28 bsAb XENP34963 significantly enhanced T cell proliferation compared to PD-L1 blockade alone. [Figure 97] Figure 1 shows the induction of A) IL-2 secretion, B) IFNγ secretion, and C) CD3+ T cell proliferation by combining 1μg / ml of αB7H3xαCD3 bsAb with αPD-L1 mAb XENP24118, or αPD-L1xαCD28 bsAb with XENP34963. αPD-L1xαCD28 bsAb enhances the activity of CD3 bsAb T cell engagers. [Figure 98] Figure 1 shows cell killing over time after incubation of LNCaP cancer cells (PSMA+) with CD3+ T cells at a 1:1 effector:target ratio and with an exemplary CD3 bispecific (αPSMA×αCD3 XENP32220) alone or in combination with XENP36233. The data show that XENP32220 enhanced cell killing compared to incubation of cancer cells and T cells alone, but that the addition of αPD-L1×αCD28 overcomes cancer cell resistance to the CD3 bispecific and further enhances cell killing. [Figure 99]Shown are group median changes in tumor volume (baseline corrected, as determined by caliper) over time (days following first dose) and at day 20 (post-first dose) in MC38 (engineered to stably express human PD-L1) and huPBMC-engrafted human CD28 knock-in mice dosed with 5 mg / kg αPD-L1 mAb XENP24118, 8.3 mg / kg αPD-L1×αCD28 bsAb XENP34963, 6 mg / kg αPD-L1×αCD28 XENP34961, or PBS control. [Figure 100] Figure 1 shows the induction of A) IL-2 and B) IFNγ release by anti-PDL1 clone 2G4 (XENP25859), partial blocking anti-PDL1 (XENP25853), non-blocking anti-PDL1 (XENP25858), and XENP24118 (avelumab-based benchmark anti-PDL1 mAb). The data show that the partial blocking and non-blocking anti-PDL1 clones induced less cytokine release compared to anti-PDL1 clone 2G4. [Figure 101] Figure 1 shows blockade of PD1:PDL1 interactions during T cell:cancer cell interactions (as modeled by Jurkat-PD1 cells incubated with CHO-PDL1-CD80-αCD3 and CHO-PDL1-αCD3 cells) by αPDL1 × αCD28 bsAb with anti-PDL1 clone 2G4 (XENP36233), partial blocking anti-PDL1 (XENP36232), non-blocking anti-PDL1 (XENP26783), and XENP34963 (a benchmark bsAb with an avelumab-based anti-PDL1 arm). The data show that partial blocking and non-blocking anti-PDL1 clones induced less activity compared to bsAbs with anti-PDL1 clone 2G4. [Figure 102] Binding of αPDL1×αCD28 (XENP36233) to parental PDL1 null MC38 cells or MC38 cells transfected to express PDL1 with low or intermediate-high antigen density is shown. [Figure 103]1 shows the induction of IL-2 release by αPDL1×αCD28 (XENP36233) in the presence of parental PDL1 null HEK293T cells or HEK293T cells transfected to express PDL1 with intermediate or high antigen density. [Figure 104] Induction of cell killing by αPSMA×αCD3 alone or in combination with αPDL1×αCD28 XENP36233 in the presence of CD3+ T cells and the PDL1 null 22Rv1 cell line is shown at A) an E:T ratio of 10:1 and B) an E:T ratio of 1:1. The data show that the αPDL1×αCD28 bsAb does not synergize with the CD3 bsAb on PDL1-negative cell lines such as 22Rv1. [Figure 105] Serum concentrations of XENP36764 over time in cynomolgus monkeys. The αPDL1 × αCD28 bsAb demonstrated favorable pharmacokinetics. [Figure 106] FIG. 1 shows a diagram of the assumptions used in the mechanism-based PK / PD computer model. [Figure 107] Predictions from mechanism-based modeling are presented, suggesting A) linear PK at dose levels consistent with typical checkpoint inhibitor regimens, B) trimer formation within the tumor indicative of costimulation, and C) uniform PDL1 blockade. [Figure 108] Figure 1 shows induction of IL-2 release by αPDL1×αCD28 bsAb with affinity-engineered CD28 binding domain in the presence of αB7H3×αCD3 bsAb, CD3+ T cells, and A) MDA-MB-231 cells or B) DU145-NLR cells. The data demonstrate that increased affinity for CD28 results in more potent and effective IL-2 secretion by αPDL1×αCD28 bsAb. [Figure 109] Figure 1 shows induction of cell killing by αPDL1×αCD28 bsAb with an affinity-engineered CD28 binding domain in the presence of αB7H3×αCD3 bsAb, CD3+ T cells, and PDL1lowLnCAP cancer cells. The data demonstrate that increased affinity for CD28 increases targeting of PDL1low cancer cells even at E:T ratios as low as 1:1. [Figure 110] Figure 1 shows induction of cell killing by αPDL1×αCD28 bsAb with an affinity-engineered CD28-binding domain in the presence of αB7H3×αCD3 bsAb, CD3+ T cells, and PDL1med DU145 cells. The data demonstrate that increased affinity for CD28 increases targeting of PDL1med cancer cells even at E:T ratios as low as 0.1:1. [Figure 111] Changes in tumor volume (baseline corrected, as determined by caliper measurement) over time and in individual mice at D) day 28 are shown in hPDL1-MC38-implanted hCD28 knock-in mice dosed with A) PBS control, B) monovalent αPDL1 mAb XENP36627, and C) XENP37261, which has enhanced CD28 binding affinity. [Figure 112] Figure 1 shows induction of IL-2 release by αPDL1×αCD28 bsAb with affinity-engineered PDL1-binding domain in the presence of αB7H3×αCD3 bsAb, CD3+ T cells, and DU145-NLR cells. The data demonstrate that increased PDL1 affinity promotes IL-2 secretion. [Figure 113] Figure 1 shows the induction of A) IL2 and B) IFNγ release by αPDL1 mAb XENP24118 and αPDL1×αCD28 bsAb XENP38514, which enhanced PDL1 binding, in a DC:T cell MLR. The data demonstrate that αPDL1×αCD28 enhanced T cell / APC interactions. [Figure 114] Figure 1 shows PD1:PDL1 blockade (binding of PDL1-mFc fusion to PD1-transfected HEK293T cells) by αPDL1 × αCD28 bsAb with affinity-engineered PDL1-binding domain. The data demonstrate that αPDL1 × αCD28 bsAb can block the interaction between PD1 and PDL1. [Figure 115]Figure 1 shows induction of IL-2 release by αPDL1xαCD28 bsAb with affinity-engineered CD28-binding domain and affinity-engineered PDL1-binding domain in the presence of SEB-stimulated PBMC. The data demonstrate that increased PDL1 affinity promotes IL-2 secretion. [Figure 116] Figure 1 shows induction of IL-2 release by CD3+ enriched T cells, MDA-MB0231 transfected to express αCD3 scFv (functioning as signal 1), and αPDL1xαCD28 bsAb with affinity-engineered CD28 and PDL1 binding domains in the presence of 1 μg / mL of an exemplary B7H3xCD3 bsAb. The data show that XENP40409 (a non-Xtend analog of XENP40706) with 2G4_H1.12_L1.14 most potently induced IL2 production. [Figure 117] Figure 1 shows CD28 receptor occupancy on cynomolgus monkey T cells (as indicated by reduced binding by a secondary CD28 mAb) after dosing with A) XENP36803 (1x, 4x, and 10x doses) or B) XENP36764 (4x, 10x, and 20x doses). Data show CD28 receptor occupancy on T cells by day 14. [Figure 118] Figure 1 shows PDL1 receptor occupancy (reduction in free receptors as indicated by reduced binding by 2G4-based one-arm PDL1 mAb) on cynomolgus monkey T cells after dosing with A) XENP36803 (1x, 4x, and 10x doses) or B) XENP36764 (4x, 10x, and 20x doses). [Figure 119-1] Figure 1 shows proliferation (as indicated by increased Ki67 expression) of cynomolgus T cells after dosing with A) XENP36803 (1x, 4x, and 10x doses) or B) XENP36764 (4x, 10x, and 20x doses). Notably, PDL1xCD28 bsAb selectively induces proliferation of effector CD4+ and CD8+ T cells (i.e., CD45RA-). [Figure 119-2]Figure 1 shows proliferation (as indicated by increased Ki67 expression) of cynomolgus T cells after dosing with A) XENP36803 (1x, 4x, and 10x doses) or B) XENP36764 (4x, 10x, and 20x doses). Notably, PDL1xCD28 bsAb selectively induces proliferation of effector CD4+ and CD8+ T cells (i.e., CD45RA-). [Figure 120] Figure 1 shows the induction of A) IL-2 secretion and B) IFNγ secretion from T cells by combining 1 μg / ml of an exemplary B7H3xCD3 bsAb with XENP37261, XENP40409, and an additional PDL1xCD28 bsAb utilizing an additional PDL1-binding domain derived from a common light chain approach in the presence of MDA-MB-231 PDL1+PDL2+ cancer cells (1:1 E:T). XENP42047 and XENP42048, utilizing the 13G1 and 13G7 PDL1 CLC-binding domains, respectively, demonstrated similar potency and efficacy to XENP37261, while additional bsAbs utilizing other PDL1 CLC-binding domains were less potent and / or effective. [Figure 121] 1 shows the induction of A) IL-2 secretion and B) IFNγ secretion from T cells by combining 1 μg / ml of an exemplary B7H3xCD3 bsAb with XENP37261, XENP40409, and a PDL2xCD28 bsAb utilizing the PDL2 binding domain derived from the common light chain vector in the presence of MDA-MB-231 PDL1+PDL2+ cancer cells (1:1 E:T). Many PDL2xCD28 bsAbs (XENP42051, XENP42052, XENP42053, and XENP42054, which utilize the 5C11, 8G2, 8G5, and 16G11 PDL2 CLC-binding domains, respectively), but not all (i.e., XENP42050, which uses another high-affinity PDL2 CLC-binding domain), were comparable in potency and / or efficacy to or greater than PDL1xCD28 bsAbs in inducing cytokine release. [Figure 122]We present a scenario in which tumor cells have different expression levels of PDL1 and PDL2, and the relative binding capacities of the PDL1 x CD28 bsAb, the combination of PDL1 x CD28 and PDL2 x CD28 bsAb, and the PDL1 x PDL2 x CD28 triAb (a 1 + 1 + 1 Fab-Fab-scFv format with PDL1 and PDL2 CLC-binding domains and CD28-binding domains), demonstrating that the triAb targets PDL1-PDL2+ tumors, assuming saturation and, if possible, 100% avidity (without switching to monovalent interactions, which can occur at high concentrations). [Figure 123] Exemplary murine B7H3xCD3 bsAbs at 1 μg / ml in the presence of CHO-PDL1 cells (1:1 E:T) were tested using the following antibodies: XENP40409; PDL1xCD28 bsAbs utilizing the CLC PDL1 binding domain (XENP42047 and XENP42049); PDL2xCD28 bsAbs utilizing the CLC PDL2 binding domain (XENP42051, XENP42052, XENP42053, and XENP42054); and PDL1xPDL2xCD28 bsAbs in a 1+1+1 stack Fab2-scFv-Fc format with PDL2 underneath. triAb (XENP42324, XENP42325, XENP42326, XENP42327, XENP42328, XENP42329, XENP42330, and XENP42331), PDL1 × PDL2 × CD28 triAb in a 1 + 1 + 1 stacked Fab2-scFv-Fc format with PDL1 at the bottom (XENP42332, XENP42333, XENP42334, XENP42335, XENP42336, XENP42337, XENP42338, and XENP42339), and PDL1 × PDL2 × CD28 in a 1 + 1 + 1 mAb-scFv format. 1 shows the induction of IL-2 secretion by T cells by combination with triAbs (XENP42340, XENP42341, XENP42342, XENP42343, XENP42344, XENP42345, XENP42346, and XENP42347). [Figure 124]Exemplary murine B7H3xCD3 bsAbs at 1 μg / ml in the presence of CHO-PDL2 cells (1:1 E:T) were tested using the following antibodies: XENP40409, PDL1xCD28 bsAbs utilizing the CLC PDL1 binding domain (XENP42047 and XENP42049), PDL2xCD28 bsAbs utilizing the CLC PDL2 binding domain (XENP42051, XENP42052, XENP42053, and XENP42054), and PDL1xPDL2xCD28 bsAbs in a 1+1+1 stack Fab2-scFv-Fc format with PDL2 underneath. triAb (XENP42324, XENP42325, XENP42326, XENP42327, XENP42328, XENP42329, XENP42330, and XENP42331), PDL1 × PDL2 × CD28 triAb in a 1 + 1 + 1 stacked Fab2-scFv-Fc format with PDL1 at the bottom (XENP42332, XENP42333, XENP42334, XENP42335, XENP42336, XENP42337, XENP42338, and XENP42339), and PDL1 × PDL2 × CD28 in a 1 + 1 + 1 mAb-scFv format. 1 shows the induction of IL-2 secretion by T cells by combination with triAbs (XENP42340, XENP42341, XENP42342, XENP42343, XENP42344, XENP42345, XENP42346, and XENP42347). [Figure 125] 1 shows the relative expression levels of PDL1, PDL2, and B7H3 on LCLC-103H, SNU-423, and NCI-H460 cancer cells. [Figure 126]Figure 1 shows the induction of IL-2 secretion from T cells by combining 1 μg / ml of an exemplary B7H3xCD3 bsAb with XENP40409, PDL1xCD28 bsAbs (XENP42047 and XENP42049) utilizing a PDL1 binding domain derived from a common light chain vector, and PDL2xCD28 bsAbs (XENP42051, XENP42052, XENP42053, and XENP42054) utilizing a PDL2 binding domain derived from a common light chain vector in the presence of A) LCLC-103H, B) SNU-423, and C) NCI-H460 cancer cells (1:1 E:T). Each of the bsAbs showed activity against LCLC-103H and SNU-423, with XPL1-13G1 being the most potent PDL1 binding domain (as in XENP42047) and 8G2 being the most potent PDL2 binding domain (as in XENP42052). [Figure 127A] Exemplary B7H3xCD3 bsAb at 1 μg / ml was tested in the presence of A) LCLC-103H, B) SNU-423, and C) NCI-H460 cancer cells (1:1 E:T) with PDL1xPDL2xCD28 triAb (XENP42324, XENP42325, XENP42326, XENP42327, XENP42328, XENP42329, XENP42330, and XENP42331) in a 1+1+1 stack Fab2-scFv-Fc format with PDL2 at the bottom, and PDL1xPDL2xCD28 triAb (XENP42324, XENP42325, XENP42326, XENP42327, XENP42328, XENP42329, XENP42330, and XENP42331) in a 1+1+1 stack Fab2-scFv-Fc format with PDL1 at the bottom. Induction of IL-2 secretion by T cells by combination with triAb (XENP42332, XENP42333, XENP42334, XENP42335, XENP42336, XENP42337, XENP42338, and XENP42339). D) Legend. [Figure 127B]Exemplary B7H3xCD3 bsAb at 1 μg / ml was tested in the presence of A) LCLC-103H, B) SNU-423, and C) NCI-H460 cancer cells (1:1 E:T) with PDL1xPDL2xCD28 triAb (XENP42324, XENP42325, XENP42326, XENP42327, XENP42328, XENP42329, XENP42330, and XENP42331) in a 1+1+1 stack Fab2-scFv-Fc format with PDL2 at the bottom, and PDL1xPDL2xCD28 triAb (XENP42324, XENP42325, XENP42326, XENP42327, XENP42328, XENP42329, XENP42330, and XENP42331) in a 1+1+1 stack Fab2-scFv-Fc format with PDL1 at the bottom. Induction of IL-2 secretion by T cells by combination with triAb (XENP42332, XENP42333, XENP42334, XENP42335, XENP42336, XENP42337, XENP42338, and XENP42339). D) Legend. [Figure 127C] Exemplary B7H3xCD3 bsAb at 1 μg / ml was tested in the presence of A) LCLC-103H, B) SNU-423, and C) NCI-H460 cancer cells (1:1 E:T) with PDL1xPDL2xCD28 triAb (XENP42324, XENP42325, XENP42326, XENP42327, XENP42328, XENP42329, XENP42330, and XENP42331) in a 1+1+1 stack Fab2-scFv-Fc format with PDL2 at the bottom, and PDL1xPDL2xCD28 triAb (XENP42324, XENP42325, XENP42326, XENP42327, XENP42328, XENP42329, XENP42330, and XENP42331) in a 1+1+1 stack Fab2-scFv-Fc format with PDL1 at the bottom. Induction of IL-2 secretion by T cells by combination with triAb (XENP42332, XENP42333, XENP42334, XENP42335, XENP42336, XENP42337, XENP42338, and XENP42339). D) Legend. [Figure 128] Figure 1 shows PD-1 blockade by PDL1xCD28 and PDL2xCD28 bsAbs based on various PDL1 and PDL2 binding domains on CHO-PDL1 cells. [Figure 129]Figure 1 shows PD-1 blockade by PDL1xCD28 and PDL2xCD28 bsAbs based on various PDL1 and PDL2 binding domains on CHO-PDL2 cells. [Figure 130] Figure 1 shows the induction of IL-2 secretion by T cells by combining an exemplary murine B7H3xCD3 bsAb with A) XENP42049 PDL1xCD28, B) XENP42054 PDL2xCD28, and C) XENP43461 PDL1xPDL2xCD28 in the presence of the LCLC103H-NLR cell line and PDL1 or PDL2 blockade. [Figure 131] A control antibody utilizing the RSV binding domain is shown. [Figure 132] Figure 1 shows PDL1 expression on PC3 cells co-cultured with T cells after treatment with an exemplary B7H3xCD3 bsAb alone or in combination with an IFNγ-neutralizing mAb. CD3 bsAb induces PDL1 and PDL2 expression by promoting IFNγ release. [Figure 133A] Figure 1 shows IFNγ secretion by T cells co-cultured with PC3 cells and then treated with the indicated antibodies (exemplary B7H3xCD3 bsAb alone or in combination with XENP40409) for A) 1 day, B) 2 days, and C) 5 days. CD3 bsAb shows synergy with PDL1xCD28 over time as PDL1 expression increases. [Figure 133B] Figure 1 shows IFNγ secretion by T cells co-cultured with PC3 cells and then treated with the indicated antibodies (exemplary B7H3xCD3 bsAb alone or in combination with XENP40409) for A) 1 day, B) 2 days, and C) 5 days. CD3 bsAb shows synergy with PDL1xCD28 over time as PDL1 expression increases. [Figure 133C] Figure 1 shows IFNγ secretion by T cells co-cultured with PC3 cells and then treated with the indicated antibodies (exemplary B7H3xCD3 bsAb alone or in combination with XENP40409) for A) 1 day, B) 2 days, and C) 5 days. CD3 bsAb shows synergy with PDL1xCD28 over time as PDL1 expression increases. [Figure 134] Figure 1 shows the induction of IFNγ release (5 days post-treatment) in mixed lymphocyte reactions (n=14) after incubation with PBS, XENP43456 (PDL1 x PDL2 x RSV), or anti-PD1 mAb. PDL1 and PDL2 blockade is functionally equivalent to PD1 blockade. [Figure 135] IFNγ release (1 day after treatment) by PBMCs treated with the indicated air-dried αCD28 bivalent antibodies (derived from TGN1412 or 1A7) is shown. The anti-CD28 epitope derived from 1A7 lacks superagonistic properties. [Figure 136A] Figure 1 shows induction of IL-2 release by T cells co-cultured with A) PDL1high MDA-MB-231 cancer cells (1:1 effector:target ratio, approximately 130,000 PDL1 antigens) and B) PDL1low LNCaP (10:1 effector:target ratio, approximately 13,000 PDL1 antigens) and treated with the indicated PDL1xPDL2xCD28 triAb at the indicated concentrations in combination with an exemplary B7H3xCD3 bsAb. The PDL1xPDL2xCD28 triAb enhanced IL-2 release against both PDL1high and PDL1low cancer cells. [Figure 136B] Figure 1 shows induction of IL-2 release by T cells co-cultured with A) PDL1high MDA-MB-231 cancer cells (1:1 effector:target ratio, approximately 130,000 PDL1 antigens) and B) PDL1low LNCaP (10:1 effector:target ratio, approximately 13,000 PDL1 antigens) and treated with the indicated PDL1xPDL2xCD28 triAb at the indicated concentrations in combination with an exemplary B7H3xCD3 bsAb. The PDL1xPDL2xCD28 triAb enhanced IL-2 release against both PDL1high and PDL1low cancer cells. [Figure 137A]Figure 1 shows induction of redirected T cell cytotoxicity (as determined by RTCC, luminescence 5 days post-treatment) by T cells cocultured with A) PDL1high MDA-MB-231 cancer cells (effector:target ratio of 1:50, approximately 130,000 PDL1 antigens) and B) PDL1low LNCaP (effector:target ratio of 10:1, approximately 13,000 PDL1 antigens) and treated with the indicated concentrations of exemplary B7H3xCD3 bsAb alone or in combination with 1 μg / ml of the indicated PDL1xPDL2xCD28 triAb. PDL1xPDL2xCD28 triAb enhances redirected T cell cytotoxicity at low effector-to-target ratios as well as on PDL1low cancer cells. [Figure 137B] Figure 1 shows induction of redirected T cell cytotoxicity (as determined by RTCC, luminescence 5 days post-treatment) by T cells cocultured with A) PDL1high MDA-MB-231 cancer cells (effector:target ratio of 1:50, approximately 130,000 PDL1 antigens) and B) PDL1low LNCaP (effector:target ratio of 10:1, approximately 13,000 PDL1 antigens) and treated with the indicated concentrations of exemplary B7H3xCD3 bsAb alone or in combination with 1 μg / ml of the indicated PDL1xPDL2xCD28 triAb. PDL1xPDL2xCD28 triAb enhances redirected T cell cytotoxicity at low effector-to-target ratios as well as on PDL1low cancer cells. [Figure 138A] Induction of IL2 release (24 hours after treatment) by T cells from A) the first and B) the second CMV+ PBMC donors cocultured with A431-β2M-null cells or A431-β2M-null cells stably expressing a fusion of HLA-A2, β2M, and NLV peptides and treated with PDL1×PDL2×CD28, PDL1×PDL2×RSV, or RSV×CD28 is shown. [Figure 138B]Induction of IL2 release (24 hours after treatment) by T cells from A) the first and B) the second CMV+ PBMC donors cocultured with A431-β2M-null cells or A431-β2M-null cells stably expressing a fusion of HLA-A2, β2M, and NLV peptides and treated with PDL1×PDL2×CD28, PDL1×PDL2×RSV, or RSV×CD28 is shown. [Figure 139A] The sequences of XENP43734 and XENP43735, mouse surrogate PDL1 x PDL2 x CD28 and PDL1 x PDL2 x RSV triAbs, respectively, are shown. Note that XENP43735 can be considered a mouse surrogate for XENP43461 (without Xtend). [Figure 139B] The sequences of XENP43734 and XENP43735, mouse surrogate PDL1 x PDL2 x CD28 and PDL1 x PDL2 x RSV triAbs, respectively, are shown. Note that XENP43735 can be considered a mouse surrogate for XENP43461 (without Xtend). [Figure 140-1] Figure 1 shows the change in tumor volume (as determined by caliper measurement) over time for individual mice in hPDL1(high)-MC38-implanted hCD28 knock-in mice treated with A) PBS control, B) anti-mouse PD1 mAb, and C) XENP43735 (mouse surrogate PDL1 x PDL2 x RSV triAb control), and D) XENP43734 (mouse surrogate PDL1 x PDL2 x CD28 triAb similar to XENP43461 (no Xtend)). Arrows indicate the day of treatment. [Figure 140-2] Figure 1 shows the change in tumor volume (as determined by caliper measurement) over time for individual mice in hPDL1(high)-MC38-implanted hCD28 knock-in mice treated with A) PBS control, B) anti-mouse PD1 mAb, and C) XENP43735 (mouse surrogate PDL1 x PDL2 x RSV triAb control), and D) XENP43734 (mouse surrogate PDL1 x PDL2 x CD28 triAb similar to XENP43461 (no Xtend)). Arrows indicate the day of treatment. [Figure 141]1 shows blockade of PD1:PDL1 and PD1:PDL2 by bivalent PD-1 mAb or XENP43456 PDL1xPDL2xRSV triAb on Jurkat-PD1 cells incubated with A431-αCD3. [Figure 142A] Shown is the change in tumor volume (as determined by caliper measurement) over time for individual mice in MDA-MB-231 and huPBMC-engrafted NSG-DKO mice dosed with A) PBS control, B) exemplary B7H3xCD3 bsAb (0.5 mg / kg), C) PDL1xPDL2xCD28 XENP44676 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively), D) control PDL1xPDL2xRSV XENP44796 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively), or E) control RSVxRSVxCD28 XENP44797 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively). [Figure 142B] Shown is the change in tumor volume (as determined by caliper measurement) over time for individual mice in MDA-MB-231 and huPBMC-engrafted NSG-DKO mice dosed with A) PBS control, B) exemplary B7H3xCD3 bsAb (0.5 mg / kg), C) PDL1xPDL2xCD28 XENP44676 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively), D) control PDL1xPDL2xRSV XENP44796 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively), or E) control RSVxRSVxCD28 XENP44797 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively). [Figure 142C]Shown is the change in tumor volume (as determined by caliper measurement) over time for individual mice in MDA-MB-231 and huPBMC-engrafted NSG-DKO mice dosed with A) PBS control, B) exemplary B7H3xCD3 bsAb (0.5 mg / kg), C) PDL1xPDL2xCD28 XENP44676 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively), D) control PDL1xPDL2xRSV XENP44796 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively), or E) control RSVxRSVxCD28 XENP44797 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively). [Figure 142D] Shown is the change in tumor volume (as determined by caliper measurement) over time for individual mice in MDA-MB-231 and huPBMC-engrafted NSG-DKO mice dosed with A) PBS control, B) exemplary B7H3xCD3 bsAb (0.5 mg / kg), C) PDL1xPDL2xCD28 XENP44676 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively), D) control PDL1xPDL2xRSV XENP44796 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively), or E) control RSVxRSVxCD28 XENP44797 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively). [Figure 142E]Shown is the change in tumor volume (as determined by caliper measurement) over time for individual mice in MDA-MB-231 and huPBMC-engrafted NSG-DKO mice dosed with A) PBS control, B) exemplary B7H3xCD3 bsAb (0.5 mg / kg), C) PDL1xPDL2xCD28 XENP44676 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively), D) control PDL1xPDL2xRSV XENP44796 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively), or E) control RSVxRSVxCD28 XENP44797 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively). [Figure 143] Tumor volumes (as determined by caliper measurement) of individual MDA-MB-231 and huPBMC-engrafted NSG-DKO mice are shown 39 days after dosing with A) PBS control, B) exemplary B7H3xCD3 bsAb (0.5 mg / kg), C) PDL1xPDL2xCD28 XENP44676 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively), D) control PDL1xPDL2xRSV XENP44796 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively), or E) control RSVxRSVxCD28 XENP44797 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively). [Figure 144]Shown are CD3+ cell counts in the blood of MDA-MB-231 and huPBMC-engrafted NSG-DKO mice dosed with A) PBS control, B) exemplary B7H3xCD3 bsAb (0.5 mg / kg), C) PDL1xPDL2xCD28 XENP44676 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively), D) control PDL1xPDL2xRSV XENP44796 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively), or E) control RSVxRSVxCD28 XENP44797 in combination with B7H3xCD3 bsAb (5 mg / kg and 0.5 mg / kg, respectively) on day 7 after the first dose. [Figure 145] Figure 1 shows the KD binding constant, association constant (ka), and dissociation constant (kd) of affinity engineered anti-PDL1 clone 13G7 against human PDL1. [Figure 146] 1 shows the KD binding constant, association constant (ka), and dissociation constant (kd) of affinity engineered anti-PDL1 clone 13G7 to cynomolgus monkey PDL1. [Figure 147] Figure 1 shows the KD binding constant, association constant (ka), and dissociation constant (kd) of affinity engineered anti-PDL2 clone 16G11 against human PDL2. [Figure 148] Figure 1 shows the KD binding constant, association constant (ka), and dissociation constant (kd) of affinity engineered anti-PDL2 clone 16G11 to cynomolgus monkey PDL2. [Figure 149] Figure 1 shows the KD binding constant, association constant (ka), and dissociation constant (kd) of affinity engineered anti-PDL1 clone 1A3A4.248 against human PDL1. [Figure 150] 1 shows the KD binding constant, association constant (ka), and dissociation constant (kd) of affinity engineered anti-PDL1 clone 1A3A4.248 to cynomolgus monkey PDL1. [Figure 151] Figure 1 shows the KD binding constant, association constant (ka), and dissociation constant (kd) of affinity engineered anti-PDL2 clone 1F12A4.249 (round 1) against human PDL2. [Figure 152]Figure 1 shows the KD binding constant, association constant (ka), and dissociation constant (kd) of affinity engineered anti-PDL2 clone 1F12A4.249 (round 1) to cynomolgus monkey PDL2. [Figure 153A] Figure 1 shows the KD binding constant, association constant (ka), and dissociation constant (kd) of affinity engineered anti-PDL2 clone 1F12A4.249 (round 2) against human PDL2. [Figure 153B] Figure 1 shows the KD binding constant, association constant (ka), and dissociation constant (kd) of affinity engineered anti-PDL2 clone 1F12A4.249 (round 2) against human PDL2. [Figure 154A] Figure 1 shows the KD binding constant, association constant (ka), and dissociation constant (kd) of affinity engineered anti-PDL2 clone 1F12A4.249 (round 2) to cynomolgus monkey PDL2. [Figure 154B] Figure 1 shows the KD binding constant, association constant (ka), and dissociation constant (kd) of affinity engineered anti-PDL2 clone 1F12A4.249 (round 2) to cynomolgus monkey PDL2. [Figure 155] Figure 1 shows the induction of IL-2 secretion by combining 1 μg / mL of an exemplary B7H3xCD3 bsAb with 1+1+1 Fab-(Fab-scFv)-Fc (with CD28 scFv in VHVL or VLVH orientation), 1+1+1 stacked Fab2-scFv-Fc, and 1+1+1 mAb-scFv format PDL1xPDL2xCD28 in the presence of A) LnCAP (PDL1lowPDL2null), B) DU145 (PDL1medPDL2null), or C) LCLC103H (PDL1hiPDL2med) tumor cells. [Figure 156] Figure 1 shows the induction of IL-2 secretion by combining 1 μg / mL of an exemplary B7H3xCD3 bsAb with a 1+1+1 stacked Fab2-scFv-Fc PDL1xPDL2xCD28 with WT scFv (XENP43461) or stapled scFv (XENP43462) in the presence of DU145 (PDL1 med PDL2 null) tumor cells (1:1 effector:target ratio). [Figure 157]Figure 1 shows the induction of IL-2 secretion by combining 1 μg / mL of the exemplary B7H3xCD3 bsAb with PDL1xPDL2xCD28, which utilizes the 1A7 or 1A3 CD28-binding domain in various formats and orientations, in the presence of DU145 tumor cells. The data show that XENP43461, which contains the 1A7-binding domain, outperformed all formats of the triAb (with identical PDL1 and PDL2-binding domains) containing the 1A3-binding domain. [Figure 158] Figure 1 shows the induction of IL-2 secretion by combining 1 μg / mL of an exemplary B7H3xCD3 bsAb with PDL1xPDL2xCD28, which utilizes 1A7 with a germline VL and affinity-optimized VH, in the presence of DU145 tumor cells. The data demonstrate that several of the full CLC triAbs achieved potency comparable to that of XENP43465. [Figure 159] Figure 1 shows the monovalent KD of CD28 binding by the affinity-optimized 1A7 binding domain, as well as the EC50 for IL-2 secretion induction by combining 1 μg / mL of an exemplary B7H3xCD3 bsAb with PDL1xPDL2xCD28, which utilizes 1A7 with a germline VL and affinity-optimized VH, in the presence of DU145 tumor cells. The data demonstrate that several of the full CLC triAbs achieved potency comparable to that of XENP43465. [Figure 160] Figure 1 shows the induction of IL-2 secretion by combining 1 μg / mL of the exemplary B7H3xCD3 bsAb with PDL1xPDL2xCD28, which utilizes the CD28-binding domain from 1A7 or 1A3, in the presence of Caki-1 tumor cells (1:1 effector:target). The data demonstrate that the 1A7-based full CLC triAb is more potent and effective than the 1A3-based full CLC triAb. [Figure 161A]In addition to the novel PDL1-binding domain clone 2A3A4.248, an exemplary affinity-engineered VH 2A3A4.248_H1.9 (an additional variant shown as SEQ ID NOS: 3235-3266) is shown, which may pair with L1 (note that L1 is the IGKV1-39 germline sequence). Additionally, the consensus framework regions (FR) and complementarity-determining regions (CDRs) of the 1F12A4.249 VH are shown. As is true for all sequences herein described and containing CDRs, the precise identification of CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either scFv or Fab formats. [Figure 161B] In addition to the novel PDL1-binding domain clone 2A3A4.248, an exemplary affinity-engineered VH 2A3A4.248_H1.9 (an additional variant shown as SEQ ID NOS: 3235-3266) is shown, which may pair with L1 (note that L1 is the IGKV1-39 germline sequence). Additionally, the consensus framework regions (FR) and complementarity-determining regions (CDRs) of the 1F12A4.249 VH are shown. As is true for all sequences herein described and containing CDRs, the precise identification of CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either scFv or Fab formats. [Figure 161C]In addition to the novel PDL1-binding domain clone 2A3A4.248, an exemplary affinity-engineered VH 2A3A4.248_H1.9 (an additional variant shown as SEQ ID NOS: 3235-3266) is shown, which may pair with L1 (note that L1 is the IGKV1-39 germline sequence). Additionally, the consensus framework regions (FR) and complementarity-determining regions (CDRs) of the 1F12A4.249 VH are shown. As is true for all sequences herein described and containing CDRs, the precise identification of CDR positions may vary slightly depending on the numbering used, as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either scFv or Fab formats. [Figure 162A] In addition to the novel PDL2-binding domain clone 1F12A4.249, an exemplary affinity-engineered VH 1F12A4.249_H1.45 (an additional variant shown as SEQ ID NOS: 3267-3353) that may pair with L1 (note that L1 is the IGKV1-39 germline sequence) is shown. Additionally, the consensus framework regions (FR) and complementarity-determining regions (CDRs) of the 1F12A4.249 VH are shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either scFv or Fab formats. [Figure 162B]In addition to the novel PDL2-binding domain clone 1F12A4.249, an exemplary affinity-engineered VH 1F12A4.249_H1.45 (an additional variant shown as SEQ ID NOS: 3267-3353) that may pair with L1 (note that L1 is the IGKV1-39 germline sequence) is shown. Additionally, the consensus framework regions (FR) and complementarity-determining regions (CDRs) of the 1F12A4.249 VH are shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either scFv or Fab formats. [Figure 162C] In addition to the novel PDL2-binding domain clone 1F12A4.249, an exemplary affinity-engineered VH 1F12A4.249_H1.45 (an additional variant shown as SEQ ID NOS: 3267-3353) that may pair with L1 (note that L1 is the IGKV1-39 germline sequence) is shown. Additionally, the consensus framework regions (FR) and complementarity-determining regions (CDRs) of the 1F12A4.249 VH are shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either scFv or Fab formats. [Figure 162D]In addition to the novel PDL2-binding domain clone 1F12A4.249, an exemplary affinity-engineered VH 1F12A4.249_H1.45 (an additional variant shown as SEQ ID NOS: 3267-3353) that may pair with L1 (note that L1 is the IGKV1-39 germline sequence) is shown. Additionally, the consensus framework regions (FR) and complementarity-determining regions (CDRs) of the 1F12A4.249 VH are shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either scFv or Fab formats. [Figure 162E] In addition to the novel PDL2-binding domain clone 1F12A4.249, an exemplary affinity-engineered VH 1F12A4.249_H1.45 (an additional variant shown as SEQ ID NOS: 3267-3353) that may pair with L1 (note that L1 is the IGKV1-39 germline sequence) is shown. Additionally, the consensus framework regions (FR) and complementarity-determining regions (CDRs) of the 1F12A4.249 VH are shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either scFv or Fab formats. [Figure 162F]In addition to the novel PDL2-binding domain clone 1F12A4.249, an exemplary affinity-engineered VH 1F12A4.249_H1.45 (an additional variant shown as SEQ ID NOS: 3267-3353) that may pair with L1 (note that L1 is the IGKV1-39 germline sequence) is shown. Additionally, the consensus framework regions (FR) and complementarity-determining regions (CDRs) of the 1F12A4.249 VH are shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either scFv or Fab formats. [Figure 162G] In addition to the novel PDL2-binding domain clone 1F12A4.249, an exemplary affinity-engineered VH 1F12A4.249_H1.45 (an additional variant shown as SEQ ID NOS: 3267-3353) that may pair with L1 (note that L1 is the IGKV1-39 germline sequence) is shown. Additionally, the consensus framework regions (FR) and complementarity-determining regions (CDRs) of the 1F12A4.249 VH are shown. As is true for all sequences herein described and containing CDRs, the exact identification of the CDR positions may vary slightly depending on the numbering used as shown in Table 2; therefore, the underlined CDRs are included herein, as well as CDRs contained within VH and VL domains using other numbering systems. Furthermore, for all sequences in the figures, these VH and VL sequences may be used in either scFv or Fab formats. [Figure 163A] The consensus framework regions (FR) and complementarity determining regions (CDR) are shown, along with the novel 1A7 VH that was affinity engineered to match the IGKV1-39 germline sequence. [Figure 163B]The consensus framework regions (FR) and complementarity determining regions (CDR) are shown, along with the novel 1A7 VH that was affinity engineered to match the IGKV1-39 germline sequence. [Figure 164A] 1 shows an exemplary 1A7-based scFv in which cysteines were engineered for disulfide stabilization. [Figure 164B] 1 shows an exemplary 1A7-based scFv in which cysteines were engineered for disulfide stabilization. [Figure 164C] 1 shows an exemplary 1A7-based scFv in which cysteines were engineered for disulfide stabilization. [Figure 164D] 1 shows an exemplary 1A7-based scFv in which cysteines were engineered for disulfide stabilization. [Figure 165] Figure 1 shows the CD28 affinity of additional 1A7 scFv variants with and without disulfide stabilization (as indicated by [SS]) from different experiments in the context of different CD28 multispecific antibodies. [Figure 166] 1 shows the increased thermal stability of disulfide-stabilized 1A7 scFv. DETAILED DESCRIPTION OF THE INVENTION

[0066] I. Overview Provided herein are novel anti-PD-L1, anti-PD-L2, and anti-CD28 antibodies (including novel anti-CD28×anti-PD-L1 and anti-CD28×anti-PD-L2 bispecific antibodies, and novel anti-CD28×anti-PD-L1×anti-PD-L2 trispecific antibodies). Also provided herein are methods for making and using such antibodies for the treatment of cancer. The subject bispecific and trispecific antibodies are capable of agonistically binding to the CD28 costimulatory molecule on T cells and targeting PD-L1 and / or PD-L2 on tumor cells. Thus, such antibodies selectively enhance anti-tumor activity at the tumor site while minimizing peripheral toxicity. The subject antibodies provided herein are particularly useful for enhancing anti-tumor activity when used in combination with other anti-cancer therapies.

[0067] Thus, in one aspect, provided herein are heterodimeric antibodies that bind to two or three different antigens. In some embodiments, the antibodies are "bispecific" and bind to two different target antigens, generally CD28 and PD-L1 or CD28 and PD-L2, as described below. In some embodiments, the antibodies are "trispecific" and bind to three different target antigens, generally CD28, PD-L1, and PD-L2, as described below. These heterodimeric antibodies can bind each of the target antigens either monovalently (e.g., where there is a single antigen-binding domain, e.g., a variable heavy and variable light domain pair) or bivalently (where there are two antigen-binding domains that independently bind each antigen). In some embodiments, the heterodimeric antibodies provided herein include 1) one CD28-binding domain and 2) one PD-L1 or one PD-L2-binding domain (e.g., a "1+1 Fab-scFv-Fc" format heterodimeric antibody described herein). In other embodiments, the heterodimeric antibodies provided herein comprise 1) one CD28-binding domain, and 2) two PD-L1 or PD-L2-binding domains (e.g., heterodimeric antibodies in the "2+1 Fab2-scFv-Fc" format described herein). In some embodiments, the heterodimeric antibodies provided herein are trispecific antibodies, which comprise three different antigen-binding domains, each binding to a different target antigen, typically PD-L1, PD-L2, and CD28 (e.g., heterodimeric antibodies in the "1+1+1 stacked Fab2-scFv-Fc," "1+1+1 Fab-(Fab-scFv)-Fc," "1+1+1 mAb-scFv," and "1+1+1 stacked Fab2-Fab-Fc" formats; Figure 83). The heterodimeric antibodies provided herein are based on the use of different monomers that contain amino acid substitutions (i.e., "scubariant") that "skew" heterodimer formation relative to homodimers, as more fully outlined below. In some embodiments, heterodimeric antibodies are also combined with "pI variants" that allow for simple purification of heterodimers from homodimers, as also outlined below.The heterodimeric bispecific antibodies provided typically rely on the use of engineered or variant Fc domains that can self-assemble in the production cells to produce heterodimeric proteins, and methods for producing and purifying such heterodimeric proteins.

[0068] II. Nomenclature The antibodies provided herein are listed in several different formats. In some cases, each monomer of a particular antibody is assigned a unique "XENP" number, although longer sequences may contain shorter sequences, as will be understood in the art. For example, the "scFv-Fc" monomer of a 1+1 Fab-scFv-Fc format antibody may have a first XENP number, but the scFv domain itself will have a different XENP number. Because some molecules have three polypeptides, the XENP number containing the components is used as the name. Thus, the molecule XENP34961, which is a 2+1 Fab2-scFv-Fc format, contains three sequences (see Figure 85A): 1) a "Fab-Fc heavy chain" monomer ("chain 1"), 2) a "Fab-scFv-Fc heavy chain" monomer ("chain 2"), and 3) a "light chain" monomer or equivalent, which will be readily identifiable by one of skill in the art through sequence alignment. These XENP numbers are in the identifiers in the sequence listings and are used in the figures. Additionally, a molecule containing three components results in multiple sequence identifiers. For example, the listing for Fab includes the complete heavy chain sequence, the variable heavy chain domain sequence, and three CDRs of the variable heavy chain domain sequence, the complete light chain sequence, the variable light chain domain sequence, and three CDRs of the variable light chain domain sequence. A Fab-scFv-Fc monomer contains the full-length sequence, the variable heavy chain domain sequence, the three heavy chain CDR sequences, and the scFv sequence (including the scFv variable heavy chain domain sequence, the scFv variable light chain domain sequence, and the scFv linker). Note that while some molecules herein with scFv domains use a single charged scFv linker (+H), others may be used. Additionally, the nomenclature for certain antigen-binding domains (e.g., PD-L1-, PD-L2-, and CD28-binding domains) uses a format of "Hx.xx_Ly.yy" or "Hx.xxLy.yy," with numbers that are unique identifiers for the particular variable chain sequence. Thus, the Fab-side variable domain of PD-L1-binding domain 2G4[PDL1] (e.g., Figure 17) is "H1L1," indicating that the variable heavy chain domain H1 is combined with the light chain domain L1.When these sequences are used as scFvs, the designation "H1_L1" or "H1L1" indicates that the variable heavy chain domain H1 is combined with the light chain domain L1 in a VH-linker-VL orientation from N to C terminus. A molecule with the same sequence of heavy and light variable domains but in the reverse order (VL-linker-VH orientation from N to C terminus) would be designated "L1_H1". Similarly, different constructs can "mix and match" heavy and light chains, as will be apparent from the sequence listing and figures.

[0069] III. Definition In order that this application may be more fully understood, certain definitions are set forth below. Such definitions are meant to encompass all grammatical equivalents.

[0070] As used herein, "CD28," "cluster of differentiation 28," and "Tp44" (e.g., Genebank accession numbers NP_001230006 (human), NP_001230007 (human), NP_006130 (human), and NP_031668 (mouse)) refer to the B7 receptor expressed on T cells, which provides costimulatory signals required for T cell activation and survival. T cell stimulation via CD28 in addition to the T cell receptor (TCR) provides a potent signal for the production of various interleukins. CD28 is a receptor for the CD80 (B7.1) and CD86 (B7.2) proteins. CD28 contains an intracellular domain with a YMNM motif that is important for the recruitment of SH2 domain-containing proteins, particularly PI3K. CD28 also contains two proline-rich motifs that can bind SH3-containing proteins. An exemplary CD28 sequence is shown in Figure 1.

[0071] As used herein, "reducing" refers to reducing or eliminating activity. Thus, for example, "reducing FcγR binding" means that an Fc region amino acid variant has less than 50% of the starting binding compared to an Fc region not containing the particular variant, with a loss of activity of 70-80-90-95-98% or more being preferred, and typically activity below the level of binding detectable in Biacore, SPR, or BLI assays. Particularly useful in reducing FcγR binding are those shown in Figure 5, which are typically added to both monomers.

[0072] "ADCC," or antibody-dependent cell-mediated cytotoxicity, as used herein, refers to a cell-mediated reaction in which nonspecific cytotoxic cells expressing FcγR recognize bound antibody on target cells and subsequently cause lysis of the target cells. ADCC correlates with binding to FcγRIIIa; increased binding to FcγRIIIa results in increased ADCC activity.

[0073] "ADCP" or "antibody-dependent cell-mediated cytophagocytosis," as used herein, refers to a cell-mediated reaction in which nonspecific cytophagic cells expressing FcγR recognize bound antibody on a target cell, and subsequently cause phagocytosis of the target cell.

[0074] As used herein, the term "antibody" is used generically. The antibodies provided herein can take many forms as described herein, including conventional antibodies as well as antibody derivatives, fragments, and mimetics as described herein.

[0075] Conventional immunoglobulin (Ig) antibodies are "Y" shaped tetramers. Each tetramer is typically composed of two identical pairs of polypeptide chains, each pair having one "light chain" monomer (typically having a molecular weight of about 25 kDa) and one "heavy chain" monomer (typically having a molecular weight of about 50-70 kDa).

[0076] Useful bispecific antibody formats include, but are not limited to, the "1+1 Fab-scFv-Fc," "2+1 Fab2-scFv-Fc," "1+1 common light chain," and "2+1 common light chain" formats provided herein (see, e.g., Figure 82). Useful trispecific antibodies include, but are not limited to, the "1+1+1 stacked Fab2-scFv-Fc," "1+1+1 Fab-(Fab-scFv)-Fc," "1+1+1 mAb-scFv," and "1+1+1 stacked Fab2-Fab-Fc" formats (see, e.g., Figure 83). Additional useful antibody formats include, but are not limited to, "mAb-Fv," "mAb-scFv," "central-Fv," "one-arm scFv-mAb," "scFv-mAb," "double-scFv," and "trident" format antibodies, as disclosed in US20180127501A1 (which is incorporated herein by reference, particularly in the relevant portion regarding antibody formats (see, e.g., Figure 2)).

[0077] An antibody heavy chain typically comprises a variable heavy (VH) domain comprising vhCDR1-3, and an Fc domain comprising CH2-CH3 monomers. In some embodiments, an antibody heavy chain comprises a hinge and a CH1 domain. A conventional antibody heavy chain is a monomer organized from N to C terminus: VH-CH1-hinge-CH2-CH3. The CH1-hinge-CH2-CH3 are collectively referred to as the heavy chain "constant domain" or "constant region" of antibodies, which exist in five different categories or "isotypes": IgA, IgD, IgG, IgE, and IgM.

[0078] In some embodiments, the antibodies provided herein comprise an IgG isotype constant domain, which has several subclasses, including, but not limited to, IgG1, IgG2, IgG3, and IgG4. In the IgG subclass of immunoglobulins, several immunoglobulin domains are present in the heavy chain. An "immunoglobulin (Ig) domain," as used herein, refers to a region of an immunoglobulin that has a unique tertiary structure. Heavy chain domains comprising a constant heavy (CH) domain and a hinge domain are of interest in the present invention. In the context of IgG antibodies, each IgG isotype has three CH regions. Thus, in the context of IgG, the "CH" domains are as follows: "CH1" refers to positions 118-215 according to the EU index of Kabat; "hinge" refers to positions 216-230 according to the EU index of Kabat; "CH2" refers to positions 231-340 according to the EU index of Kabat; and "CH3" refers to positions 341-447 according to the EU index of Kabat. The exact numbering and arrangement of heavy chain domains may vary between different numbering systems, as shown in Figure 1. As shown herein and described below, pI variants may be in one or more of the CH regions, and in the hinge region as discussed below.

[0079] It should be noted that IgG1 has different allotypes, with polymorphisms at 356 (D or E) and 358 (L or M). The sequences shown herein use the 356E / 358M allotype, but other allotypes are encompassed herein. That is, any sequence encompassing an IgG1 Fc domain encompassed herein may have the 356D / 358L allotype replacing the 356E / 358M allotype. It should be understood that therapeutic antibodies may also comprise isotype and / or subclass hybrids. For example, as shown in U.S. Publication 2009 / 0163699 (incorporated by reference), the present antibodies in some embodiments comprise human IgG1 / G2 hybrids.

[0080] "Fc" or "Fc region" or "Fc domain," as used herein, refers to a polypeptide comprising the constant region of an antibody, in some instances excluding all or a portion of the first constant region immunoglobulin domain (e.g., CH1), and in some cases, optionally including all or a portion of the hinge. In the case of an IgG, the Fc domain comprises immunoglobulin domains CH2 and CH3 (Cγ2 and Cγ3), and optionally all or a portion of the hinge region between CH1 (Cγ1) and CH2 (Cγ2). Thus, in some cases, the Fc domain comprises, from N- to C-terminus, CH2-CH3 and hinge-CH2-CH3. In some embodiments, the Fc domain is from IgG1, IgG2, IgG3, or IgG4, with IgG1 hinge-CH2-CH3 and IgG4 hinge-CH2-CH3 being particularly utilized in many embodiments. Also, in the case of a human IgG1 Fc domain, the hinge may include a C220S amino acid substitution. Additionally, in the case of a human IgG4 Fc domain, the hinge may include a S228P amino acid substitution. Although the boundaries of the Fc region may vary, the human IgG heavy chain Fc region is usually defined as comprising residues E216, C226, or A231 at its carboxyl-terminus, numbering according to the EU index in Kabat. In some embodiments, amino acid modifications are made in the Fc region to alter binding to one or more FcγRs or FcRn, as described more fully below.

[0081] By "heavy chain constant region" herein is meant the CH1-hinge-CH2-CH3 portion of an antibody (or fragment thereof), excluding the variable heavy domain; in the EU numbering of human IgG1, this is amino acids 118 to 447. By "heavy chain constant region fragment" herein is meant a heavy chain constant region that contains fewer amino acids from either or both the N- and C-termini but still retains the ability to form a dimer with another heavy chain constant region.

[0082] Another type of domain in a heavy chain is the hinge region. By "hinge" or "hinge region" or "antibody hinge region" or "hinge domain" herein is meant a flexible polypeptide comprising the amino acids between the first and second constant domains of an antibody. Structurally, the IgG CH1 domain ends at EU position 215, and the IgG CH2 domain begins at residue EU position 231. Thus, in the case of IgG, the antibody hinge is defined herein to comprise positions 216 (E216 in IgG1) to 230 (P230 in IgG1), where the numbering is according to the EU index in Kabat. In some cases, "hinge fragments" containing fewer amino acids at either or both the N- and C-termini of the hinge domain are used. As described herein, pI variants can also be made in the hinge region. Many of the antibodies herein have at least one cysteine ​​at position 220 according to the EU numbering (hinge region) replaced with serine. Typically, this modification is on the "scFv monomer" side for most of the sequences shown herein, but it can also be on the "Fab monomer" side to reduce disulfide formation, or both. One or both of these replaced cysteines (C220S) are specifically included within the sequences herein.

[0083] As will be appreciated by those skilled in the art, the exact numbering and arrangement of heavy chain constant region domains (i.e., CH1, hinge, CH2, and CH3 domains) can vary between different numbering systems. A useful comparison of heavy constant region numbering according to EU and Kabat is as follows: see Edelman et al., 1969, Proc Natl Acad Sci USA 63:78-85 and Kabat et al., 1991, Sequences of Proteins of Immunological Interest, 5th Ed., United States Public Health Service, National Institutes of Health, Bethesda (incorporated by reference in their entirety).

[0084] [Table 1]

[0085] An antibody light chain typically comprises two domains: a variable light domain (VL) containing the light chain CDRs v1CDR1-3, and a constant light chain region (often referred to as CL or Cκ). Antibody light chains are typically organized from N- to C-terminus as follows: VL-CL.

[0086] As used herein, the term "antigen-binding domain" or "ABD," when present as part of a polypeptide sequence, refers to a set of six complementarity-determining regions (CDRs) that specifically bind a target antigen (e.g., PD-L1 or CD28) discussed herein. As is known in the art, these CDRs typically exist as a first set of variable heavy CDRs (vhCDRs or VH CDRs) and a second set of variable light CDRs (vlCDRs or VL CDRs), each comprising three CDRs: vhCDR1, vhCDR2, and vhCDR3, the variable heavy CDRs, and vlCDR1, vlCDR2, and vlCDR3, the variable light CDRs. The CDRs are present in the variable heavy domain (vhCDR1-3) and the variable light domain (vlCDR1-3). The variable heavy and variable light domains form the Fv region.

[0087] The present invention provides a number of different CDR sets. In this case, a "complete CDR set" comprises three variable light CDRs and three variable heavy CDRs, e.g., vlCDR1, vlCDR2, vlCDR3, vhCDR1, vhCDR2, and vhCDR3. These may be part of a larger variable light or variable heavy domain, respectively. Also, as more fully outlined herein, the variable heavy and variable light domains may be on separate polypeptide chains, when heavy and light chains are used (e.g., when a Fab is used), or on a single polypeptide chain, in the case of an scFv sequence.

[0088] As will be understood by those skilled in the art, the exact numbering and arrangement of CDRs may vary between different numbering systems. However, it should be understood that the disclosure of a variable heavy and / or variable light sequence includes the disclosure of the associated (internal) CDRs. Thus, the disclosure of each variable heavy region is the disclosure of the vhCDRs (e.g., vhCDR1, vhCDR2, and vhCDR3), and the disclosure of each variable light region is the disclosure of the vlCDRs (e.g., vlCDR1, vlCDR2, and vlCDR3). A useful comparison of CDR numbering is as follows: See Lafranc et al., Dev. Comp. Immunol. 27(1):55-77(2003):

[0089] [Table 2]

[0090] Throughout this specification, the Kabat numbering system is generally used when referring to residues in the variable domain (approximately residues 1-107 for light chain variable regions and residues 1-113 for heavy chain variable regions), and the EU numbering system is used for the Fc region (e.g., Kabat et al., supra (1991)).

[0091] CDRs contribute to the antigen-binding domain and the formation of the antigen-binding, or more specifically, epitope-binding, site of an antibody. "Epitope" refers to a determinant that interacts with a specific antigen-binding site in the variable region of an antibody molecule, known as a paratope. An epitope is a group of molecules, such as amino acids or sugar side chains, that usually have specific structural and charge characteristics. A single antigen can have multiple epitopes.

[0092] An epitope can include amino acid residues that are directly involved in binding (also called the immunodominant component of the epitope) and other amino acid residues that are not directly involved in binding, for example, amino acid residues that are effectively blocked by the specific antigen-binding peptide (i.e., the amino acid residues are within the footprint of the specific antigen-binding peptide).

[0093] Epitopes can be either conformational or linear. Conformational epitopes are formed by spatially juxtaposed amino acids from different segments of a linear polypeptide chain. Linear epitopes are formed by adjacent amino acid residues in a polypeptide chain. Conformational and nonconformational epitopes can be distinguished in that the binding to the former but not the latter is lost in the presence of denaturing solvents.

[0094] An epitope typically includes at least three, more usually at least five or eight to ten amino acids in a unique spatial conformation. Antibodies that recognize the same epitope can be identified in a simple immunoassay, e.g., "binning," demonstrating the ability of one antibody to block the binding of another antibody to a target antigen. As outlined below, the present invention includes not only the antigen-binding domains and antibodies listed herein, but also those that compete for binding to the epitope bound by the listed antigen-binding domains.

[0095] In some embodiments, the six CDRs of an antigen-binding domain are provided by a variable heavy domain and a variable light domain. In the "Fab" format, the set of six CDRs is provided by two different polypeptide sequences: a variable heavy domain (vh or VH; containing vhCDR1, vhCDR2, and vhCDR3) and a variable light domain (vl or VL; containing vlCDR1, vlCDR2, and vlCDR3), with the C-terminus of the vh domain attached to the N-terminus of the CH1 domain of the heavy chain and the C-terminus of the vl domain attached to the N-terminus of the constant light domain (thus forming the light chain). In the scFv format, the vh and vl domains are covalently linked into a single polypeptide sequence, usually via the use of a linker as outlined herein (an "scFv linker"), which can be either vh-linker-vl or vl-linker-vh (starting from the N-terminus), with the former usually being preferred (including optional domain linkers on each side, depending on the format used (e.g., in Figure 26). Usually, the C-terminus of the scFv domain is joined to the N-terminus of the hinge in the second monomer.

[0096] As used herein, a "variable region" or "variable domain" refers to a region of an immunoglobulin that includes one or more Ig domains substantially encoded by any of the Vκ, Vλ, and / or VH genes constituting the kappa, lambda, and heavy chain immunoglobulin loci, respectively, and contains the CDRs that confer antigen specificity. Thus, a "variable heavy domain" pairs with a "variable light domain" to form an antigen-binding domain ("ABD"). Each variable domain also includes three hypervariable regions ("complementarity-determining regions," "CDRs") (vhCDR1, vhCDR2, and vhCDR3 for the variable heavy domain and vlCDR1, vlCDR2, and vlCDR3 for the variable light domain) and four framework (FR) regions, arranged from amino-terminus to carboxy-terminus in the following order: FR1-CDR1-FR2-CDR2-FR3-CDR3-FR4.

[0097] "Fab" or "Fab region," as used herein, refers to an antibody region comprising the VH, CH1, VL, and CL immunoglobulin domains, usually on two different polypeptide chains (e.g., VH-CH1 on one chain and VL-CL on the other chain). Fab can refer to this region in isolation or this region in the context of a bispecific antibody of the invention. In the context of Fab, Fab comprises the Fv region in addition to the CH1 and CL domains.

[0098] "Fv" or "Fv fragment" or "Fv region," as used herein, refers to the antibody region comprising the VL and VH domains. The Fv region can be formatted as both a Fab (as discussed above, which are typically two separate polypeptides that also contain the constant regions outlined above) and a single-chain Fv (scFv) (in which the vl and vh domains are typically contained in a single peptide joined by a linker as discussed herein).

[0099] By "single-chain Fv" or "scFv" herein is meant a variable heavy domain covalently linked to a variable light domain, usually using an scFv linker as discussed herein, to form an scFv or scFv domain. The scFv domain can be in either orientation from N to C terminus (vh-linker-vl or vl-linker-vh). In sequences shown in the sequence listing and figures, the order of the vh and vl domains is included in the name. For example, H.X_L.Y means that from N to C terminus it is vh-linker-vl, and L.Y_H.X is vl-linker-vh.

[0100] Some embodiments of the subject antibodies provided herein are not naturally occurring but comprise at least one scFv domain that typically comprises a variable heavy domain and a variable light domain linked together by an scFv linker. As outlined herein, scFv domains are typically oriented N- to C-terminus as VH-scFv linker-VL, whereas this can be reversed for any of the scFv domains (or those constructed using vh and vl sequences from a Fab) to VL-scFv linker-VH, with optional linkers at either or both ends depending on the format.

[0101] As used herein, "modification" or "variant" refers to an amino acid substitution, insertion, and / or deletion in a polypeptide sequence or an alteration to a site chemically linked to a protein. For example, a modification can be an altered carbohydrate or PEG structure attached to a protein. As used herein, an "amino acid modification" refers to an amino acid substitution, insertion, and / or deletion in a polypeptide sequence. For clarity, unless otherwise stated, an amino acid modification always refers to an amino acid encoded by DNA, e.g., the 20 amino acids for which DNA and RNA have codons.

[0102] As used herein, an "amino acid substitution" or "substitution" refers to the replacement of an amino acid at a particular position in a parent polypeptide sequence with a different amino acid. In particular, in some embodiments, the substitution is for an amino acid that does not naturally occur at the particular position and does not naturally occur in or on any organism. For example, the substitution E272Y refers to a variant polypeptide, in this case an Fc variant, in which glutamic acid at position 272 is replaced with tyrosine. For clarity, a protein engineered to alter a nucleic acid coding sequence but not change the starting amino acid (e.g., swapping CGG (which encodes arginine) for CGA (which still encodes arginine) to increase host organism expression levels) is not an "amino acid substitution"; i.e., if a protein has the same amino acid at the particular position where it starts, despite the generation of a new gene encoding the same protein, it is not an amino acid substitution.

[0103] "Amino acid insertion" or "insertion," as used herein, refers to the addition of an amino acid sequence at a particular position in a parent polypeptide sequence. For example, -233E or 233E designates the insertion of glutamic acid after position 233 and before position 234. Also, -233ADE or A233ADE designates the insertion of AlaAspGlu after position 233 and before position 234.

[0104] "Amino acid deletion" or "deletion," as used herein, refers to the removal of an amino acid sequence at a particular position in a parent polypeptide sequence. For example, E233- or E233#, E233() or E233del designates the deletion of glutamic acid at position 233. Also, EDA233- or EDA233# designates the deletion of the sequence GluAspAla starting at position 233.

[0105] "Variant protein" or "protein variant" or "variant," as used herein, refers to a protein that differs from that of a parent protein by at least one amino acid modification. A protein variant has at least one amino acid modification compared to the parent protein, but not so many that the variant protein does not align with the parent protein using an alignment program such as those described below. Typically, a variant protein (e.g., a variant Fc domain as outlined herein, etc.) is typically at least 75, 80, 85, 90, 91, 92, 93, 94, 95, 96, 97, 98, or 99% identical to the parent protein using an alignment program such as BLAST, as described below.

[0106] "Variant," as used herein, also refers to particular amino acid modifications that confer particular functions (e.g., "heterodimerization variants," "pI variants," "reduced variants," etc.).

[0107] As described below, in some embodiments, the parent polypeptide, e.g., an Fc parent polypeptide, is a human wild-type sequence, e.g., a heavy constant domain or Fc region from IgG1, IgG2, IgG3, or IgG4, although human sequences having variants can also function as "parent polypeptides," including, for example, the IgG1 / 2 hybrids of U.S. Publication No. 2006 / 0134105. Protein variant sequences herein preferably retain at least about 80% identity, most preferably at least about 90% identity, and more preferably at least about 95-98-99% identity to the parent protein sequence. Thus, "antibody variant" or "variant antibody," as used herein, refers to an antibody that differs from a parent antibody by at least one amino acid modification; "IgG variant" or "variant IgG," as used herein, refers to an antibody that differs from a parent IgG (again, often from a human IgG sequence) by at least one amino acid modification; and "immunoglobulin variant" or "variant immunoglobulin," as used herein, refers to an immunoglobulin sequence that differs from that of the parent immunoglobulin sequence by at least one amino acid modification. "Fc variant" or "variant Fc," as used herein, refers to a protein that comprises an amino acid modification in the Fc domain when compared to the Fc domain of human IgG1, IgG2, or IgG4.

[0108] "Fc variant" or "variant Fc," as used herein, refers to a protein comprising an amino acid modification in the Fc domain. The modification can be an addition, deletion, or substitution. Fc variants are defined according to the amino acid modification that constitutes them. Thus, for example, N434S or 434S is an Fc variant having a substitution of serine at position 434 relative to the parent Fc polypeptide, where the numbering is according to the EU index. Similarly, M428L / N434S is an Fc variant having the substitutions M428L and N434S relative to the parent Fc polypeptide. The identity of the WT amino acids may not be specified, in which case the aforementioned variant is referred to as 428L / 434S. It is noted that the substitutions may be provided in any order (i.e., for example, 428L / 434S is the same Fc variant as 434S / 428L, etc.). For all positions discussed herein with respect to antibodies or derivatives and fragments thereof (e.g., Fc domains), unless otherwise stated, amino acid position numbering is according to the EU index. The "EU index" or "EU index as in Kabat" or "EU numbering" scheme refers to EU antibody numbering (Edelman et al., 1969, Proc Natl Acad Sci USA 63:78-85, hereby incorporated by reference in its entirety). Modifications can be additions, deletions, or substitutions.

[0109] Typically, the variant Fc domain will have at least about 80, 85, 90, 95, 97, 98, or 99 percent identity (using the identity algorithms discussed below (one embodiment utilizes the BLAST algorithm as known in the art using default parameters)) to the corresponding parent human IgG Fc domain. Alternatively, the variant Fc domain may have 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 11, 12, 13, 14, 15, 16, 17, 18, 19, or 20 or more amino acid modifications when compared to the parent Fc domain. Alternatively, the variant Fc domain may have up to 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 11, 12, 13, 14, 15, 16, 17, 18, 19, or 20 amino acid modifications compared to the parent Fc domain. Also, as discussed herein, the variant Fc domains described herein still retain the ability to form dimers with another Fc domain as determined using known techniques, e.g., non-denaturing gel electrophoresis, as described herein.

[0110] As used herein, "protein" refers to at least two covalently linked amino acids, and includes proteins, polypeptides, oligopeptides, and peptides. Polypeptides constituting the antibodies of the present invention may also include synthetic derivatization of one or more side chains or termini, glycosylation, PEGylation, circular permutation, cyclization, linkers to other molecules, fusion to proteins or protein domains, and addition of peptide tags or labels.

[0111] "Residue," as used herein, refers to a position in a protein and its associated amino acid identity. For example, asparagine 297 (also called Asn297 or N297) is the residue at position 297 in human antibody IgG1.

[0112] "IgG subclass modification" or "isotype modification," as used herein, refers to an amino acid modification that converts an amino acid in one IgG isotype to the corresponding amino acid in a different aligned IgG isotype. For example, IgG1 contains a tyrosine at EU position 296, and IgG2 contains a phenylalanine, so an F296Y substitution in IgG2 is considered an IgG subclass modification.

[0113] "Non-naturally occurring modification," as used herein, means an amino acid modification that is not isomorphic. For example, because none of the human IgGs contain a serine at position 434, the substitution 434S in IgG1, IgG2, IgG3, or IgG4 (or hybrids thereof) is considered a non-naturally occurring modification.

[0114] "Amino acid" and "amino acid identity" as used herein refer to one of the 20 naturally occurring amino acids encoded by DNA and RNA.

[0115] "Effector function," as used herein, refers to a biochemical event that results from the interaction of an antibody Fc region with an Fc receptor or ligand. Effector functions include, but are not limited to, ADCC, ADCP, and CDC.

[0116] As used herein, "IgG Fc ligand" refers to a molecule, preferably a polypeptide, from any organism that binds to the Fc region of an IgG antibody to form an Fc / Fc ligand complex. Fc ligands include, but are not limited to, FcγRI, FcγRII, FcγRIII, FcRn, C1q, C3, mannan-binding lectin, mannose receptor, staphylococcal protein A, streptococcal protein G, and viral FcγR. Fc ligands also include Fc receptor homologs (FcRHs), a family of Fc receptors that are homologous to FcγRs (Davis et al., 2002, Immunological Reviews 190:123-136, incorporated by reference in its entirety). Fc ligands may also include undiscovered molecules that bind Fc. Particular IgG Fc ligands are FcRn and Fc gamma receptors. "Fc ligand," as used herein, means a molecule, preferably a polypeptide, from any organism that binds to the Fc region of an antibody to form an Fc / Fc ligand complex.

[0117] "Fc gamma receptor," "FcγR," or "Fc gamma R," as used herein, means any member of a family of proteins that bind the IgG antibody Fc region and are encoded by the FcγR gene. In humans, this family includes, but is not limited to, FcγRI (CD64), which includes the isoforms FcγRIa, FcγRIb, and FcγRIc; FcγRII (CD32), which includes the isoforms FcγRIIa (including allotypes H131 and R131), FcγRIIb (including FcγRIIb-1 and FcγRIIb-2), and FcγRIIc; and FcγRIII (CD16), which includes the isoforms FcγRIIIa (including allotypes V158 and F158) and FcγRIIIb (including allotypes FcγRIIb-NA1 and FcγRIIb-NA2) (Jefferis et al., 2002, Immunol Lett 82:57-65 (incorporated by reference in its entirety)), as well as any undiscovered human FcγR or FcγR isoform or allotype. FcγRs can be from any organism, including, but not limited to, humans, mice, rats, rabbits, and monkeys. Mouse FcγRs include, but are not limited to, FcγRI (CD64), FcγRII (CD32), FcγRIII (CD16), and FcγRIII-2 (CD16-2), as well as any undiscovered mouse FcγR or FcγR isoform or allotype.

[0118] As used herein, "FcRn" or "fetal Fc receptor" refers to a protein that binds the Fc region of an IgG antibody and is at least partially encoded by the FcRn gene. FcRn can be from any organism, including, but not limited to, humans, mice, rats, rabbits, and monkeys. As known in the art, a functional FcRn protein often comprises two polypeptides, referred to as a heavy chain and a light chain. The light chain is beta-2-microglobulin, and the heavy chain is encoded by the FcRn gene. Unless otherwise specified herein, FcRn or FcRn protein refers to the complex of the FcRn heavy chain and beta-2-microglobulin. Various FcRn variants are used to increase binding to the FcRn receptor and, in some cases, to increase serum half-life. An "FcRn variant" is an amino acid modification that contributes to increased binding to the FcRn receptor; suitable FcRn variants are listed below.

[0119] As used herein, "parent polypeptide" refers to a starting polypeptide that is subsequently modified to generate a variant. A parent polypeptide can be a naturally occurring polypeptide or a variant or engineered version of a naturally occurring polypeptide. Thus, a "parent immunoglobulin," as used herein, refers to an unmodified immunoglobulin polypeptide that is modified to generate a variant, and a "parent antibody," as used herein, refers to an unmodified antibody that is modified to generate a variant antibody. It should be noted that "parent antibody" includes known commercially available recombinantly produced antibodies, as outlined below. In this context, the "parent Fc domain" refers to the described variant; thus, a "variant human IgG1 Fc domain" is compared to the parent Fc domain of human IgG1, a "variant human IgG4 Fc domain" is compared to the parent Fc domain of human IgG4, etc.

[0120] "Position," as used herein, refers to a position in the sequence of a protein. Positions may be numbered consecutively or according to established formats, such as the EU index for numbering antibody domains (e.g., CH1, CH2, CH3, or hinge domains).

[0121] "Target antigen," as used herein, means the molecule that is specifically bound by the antigen binding domain comprising the variable region of a given antibody.

[0122] "Strandedness," as used herein in the context of the monomers of the heterodimeric antibodies of the present invention, means that the heterodimerization variants are incorporated into each monomer in a manner that preserves their ability to "match" to form heterodimers, similar to the "matching" of two strands of DNA. For example, if several pI variants are engineered into monomer A (e.g., to increase the pI), then a similarly available "charge-paired" steric variant would not interfere with the pI variant; for example, a charge variant that increases the pI would be placed in the same "strand" or "monomer" to preserve the functionality of both. Similarly, for "skew" variants provided in pairs, as more fully outlined below, one skilled in the art would consider the pI when determining which strand or monomer one set of the pair should reside in, so that pI separation is also maximized using the skewed pI.

[0123] "Target cell," as used herein, means a cell that expresses a target antigen.

[0124] By "host cell" herein in the context of producing a bispecific antibody according to the invention is meant a cell that contains exogenous nucleic acid encoding the components of the bispecific antibody and that is capable of expressing the bispecific antibody under suitable conditions. Suitable host cells are discussed below.

[0125] As used herein, "wild-type" or "WT" refers to an amino acid or nucleotide sequence found in nature, including allelic variations. A WT protein has an amino acid or nucleotide sequence that has not been intentionally modified.

[0126] Provided herein are numerous antibody domains (e.g., Fc domains) that share sequence identity with human antibody domains. Sequence identity between two similar sequences (e.g., antibody variable domains) can be determined using methods such as Smith, TF & Waterman, MS (1981) "Comparison of Biosequences," Adv. Appl. Math. 2:482 [local homology algorithm]; Needleman, SB & Wunsch, CD. (1970) "A General Method Applicable to the Search for Similarities in the Amino Acid Sequence of Two Proteins," J. Mol. Biol. 48:443 [homology alignment algorithm]; Pearson, WR & Lipman, DJ (1988) "Improved Tools for Biological Sequence Comparison," Proc. Natl. Acad. Sci. (USA) 85:2444 [similarity search method]; or Altschul, SF et al. (1990) "Basic Local Alignment Search Tool," J. Mol. Biol. 215:403-10, i.e., the "BLAST" algorithm (see https: / / blast.ncbi.nlm.nih.gov / Blast.cgi). When using any of the aforementioned algorithms, default parameters (with respect to window length, gap penalties, etc.) are used. In one embodiment, sequence identity is performed using the BLAST algorithm using default parameters.

[0127] The antibodies of the present invention are typically isolated or recombinant. "Isolated," when used to describe the various polypeptides disclosed herein, refers to a polypeptide that has been identified, separated, and / or recovered from the cell or cell culture in which it is expressed. Typically, an isolated polypeptide will be prepared by at least one purification step. "Isolated antibody" refers to an antibody that is substantially free of other antibodies with different antigen specificities. "Recombinant" means that the antibodies are produced in an exogenous host cell using recombinant nucleic acid technology, and they may also be isolated.

[0128] "Specific binding" or "specifically binds to" or "specific for" a particular antigen or epitope refers to binding that is distinct from non-specific interactions. Specific binding can be measured, for example, by determining the binding of a molecule compared to the binding of a control molecule, which is usually a molecule of similar structure that does not have binding activity. For example, specific binding can be determined by competition with a control molecule that is similar to the target.

[0129] Specific binding to a particular antigen or epitope is at least about 10 -4 M, at least about 10 -5 M, at least about 10 -6 M, at least about 10 -7 M, at least about 10 -8 M, at least about 10 -9 M, alternatively at least about 10 -10 M, at least about 10 -11 M, at least about 10 -12 This can be demonstrated by an antibody having a KD for an antigen or epitope that is at or above M, where KD refers to the dissociation rate of a particular antibody-antigen interaction. Typically, an antibody that specifically binds an antigen will have a KD for the antigen or epitope that is 20, 50, 100, 500, 1000, 5,000, 10,000, or more times higher than the control molecule.

[0130] Specific binding to a particular antigen or epitope can also be exhibited by an antibody having a K or K for the antigen or epitope that is at least 20, 50, 100, 500, 1000, 5,000, 10,000-fold or greater than that of a control, where K or K refers to the binding constant of a particular antibody-antigen interaction. Binding affinity is typically measured using Biacore, SPR, or BLI assays.

[0131] IV. PD-L1, PD-L2, and CD28 Antigen-Binding Domains Provided herein are antigen binding domains (ABDs) and ABD compositions that bind to either PD-L1, PD-L2, or CD28. In some embodiments, one or more of the ABDs are included in an antibody format described herein, including any of the bispecific and trispecific formats of Figures 82 and 83.

[0132] A. PD-L1 antigen-binding domain In one aspect, provided herein are PD-L1 antigen binding domains (ABDs), including anti-PD-L1 antibodies (e.g., anti-PD-L1 x anti-CD28 bispecific antibodies, and anti-PD-L1 x anti-PD-L2 x anti-CD28 trispecific antibodies), and compositions comprising such PD-L1 antigen binding domains (ABDs). Such PD-L1 binding domains, and related antibodies (e.g., the bispecific and trispecific antibodies disclosed herein), are utilized, for example, in the treatment of PD-L1-associated cancers. In some embodiments, the PD-L1 ABD can bind to human and cynomolgus monkey PD-L1 (see Figure 1 and Example 1).

[0133] As will be appreciated by those skilled in the art, a suitable PD-L1 binding domain may comprise a set of six CDRs as shown in Figures 25 and 26 and in the Sequence Listing. Suitable PD-L1 ABDs may also include these sequences used as scFvs or as Fab domains, as well as the entire VH and VL sequences as shown in Figures 25 and 26 and in the Sequence Listing.

[0134] In one embodiment, the PD-L1 antigen-binding domain comprises the six CDRs (i.e., vhCDR1-3 and vlCDR1-3) of the PD-L1 ABD described herein, including Figures 25 and 26, and the Sequence Listing, either as the underlined CDRs, or as the CDRs identified using other alignments within the variable heavy (VH) domain and variable light domain (VL) sequences of those shown in Figures 25 and 26, and the Sequence Listing (see Table 2), when a different numbering scheme is used as described herein and as shown in Table 2. Suitable PD-L1 ABDs may also include these sequences and the entire VH and VL sequences shown in the Figures, used as an scFv or as a Fab domain.

[0135] In one embodiment, the PD-L1 antigen-binding domain comprises the six CDRs of the PD-L1 ABD described herein (i.e., vhCDR1-3 and vlCDR1-3), including those shown in Figures 25 and 26, and the Sequence Listing. In an exemplary embodiment, the PD-L1 antigen-binding domain comprises the six CDRs of PD-L1, including a variable heavy domain (VH) and a variable light domain (VL), where the VH and VL are selected from the following:

[0136] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 21; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 25; and

[0137] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 29; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 33;

[0138] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 37; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 33;

[0139] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 29; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0140] (i) a VH comprising VH CDR1, VH CDR2, and VH CDR3, each having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3 of VH having the amino acid sequence of SEQ ID NO: 37, and (ii) a VL comprising VL CDR1, VL CDR2, and VL CDR3, each having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3 of VL having the amino acid sequence of SEQ ID NO: 20 (Figures 24 to 26).

[0141] In addition to the parent CDR sets disclosed in the Figures and Sequence Listing that form the ABD to PD-L1, provided herein are variant PD-L1 ABDs having CDRs that comprise at least one modification of a PD-L1 ABD CDR disclosed herein. In one embodiment, the PD-L1 ABD comprises a set of six CDRs that have 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 amino acid modifications when compared to the six CDRs of a PD-L1 ABD described herein, including the Figures and Sequence Listing. In an exemplary embodiment, the PD-L1 ABD comprises a set of six CDRs that have 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 amino acid modifications when compared to the six CDRs of a PD-L1 ABD comprising a variable heavy domain (VH) and a variable light domain (VL) selected from the following:

[0142] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 21; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 25; and

[0143] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 29; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 33;

[0144] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 37; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 33;

[0145] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 29; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0146] (i) a VH comprising VH CDR1, VH CDR2, and VH CDR3, each having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3 of VH having the amino acid sequence of SEQ ID NO: 37, and (ii) a VL comprising VL CDR1, VL CDR2, and VL CDR3, each having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3 of VL having the amino acid sequence of SEQ ID NO: 20 (Figures 24 to 26).

[0147] In certain embodiments, the variant PD-L1 ABD is capable of binding the PD-L1 antigen as measured by at least one of Biacore, surface plasmon resonance (SPR) and / or BLI (biolayer interferometry, e.g., Octet assay) assays (the latter of which is particularly utilized in many embodiments). In certain embodiments, the PD-L1 ABD is capable of binding to human and cynomolgus PD-L1.

[0148] In one embodiment, the anti-PD-L1 ABD comprises six CDRs that are at least 90, 95, 97, 98, or 99% identical to the six CDRs of an anti-PD-L1 ABD as described herein, including in Figures 24-26, and the Sequence Listing. In an exemplary embodiment, the anti-PD-L1 ABD comprises six CDRs that are at least 90, 95, 97, 98, or 99% identical to the six CDRs of a PD-L1 ABD comprising a variable heavy domain (VH) and a variable light domain (VL) selected from the following:

[0149] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 21; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 25; and

[0150] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 29; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 33;

[0151] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 37; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 33;

[0152] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 29; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0153] (i) a VH comprising VH CDR1, VH CDR2, and VH CDR3, each having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3 of VH having the amino acid sequence of SEQ ID NO: 37, and (ii) a VL comprising VL CDR1, VL CDR2, and VL CDR3, each having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3 of VL having the amino acid sequence of SEQ ID NO: 20 (Figures 24 to 26).

[0154] In certain embodiments, the anti-PD-L1 ABD is capable of binding to the PD-L1 antigen as measured by at least one of Biacore, surface plasmon resonance (SPR) and / or BLI (biolayer interferometry, e.g., Octet assay) assays (the latter of which is particularly utilized in many embodiments). In certain embodiments, the PD-L1 ABD is capable of binding to human and cynomolgus PD-L1.

[0155] In another exemplary embodiment, the anti-PD-L1 ABD comprises the variable heavy chain (VH) domain and / or variable light chain (VL) domain of any one of the PD-L1 ABDs described herein, including Figures 24-26 and the Sequence Listing. In an exemplary embodiment, the VH and VL are selected from the following:

[0156] (i) a VH having the amino acid sequence of SEQ ID NO: 21, and (ii) a VL having the amino acid sequence of SEQ ID NO: 25, and

[0157] (i) a VH having the amino acid sequence of SEQ ID NO: 29, and (ii) a VL having the amino acid sequence of SEQ ID NO: 33;

[0158] (i) a VH having the amino acid sequence of SEQ ID NO: 37, and (ii) a VL having the amino acid sequence of SEQ ID NO: 33;

[0159] (i) a VH having the amino acid sequence of SEQ ID NO: 29, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20, and

[0160] (i) VH having the amino acid sequence of SEQ ID NO: 37, and (ii) VL having the amino acid sequence of SEQ ID NO: 20 (Figures 24 to 26).

[0161] In addition to the parent anti-PD-L1 binding domain variable heavy and variable light domains disclosed herein, provided herein are anti-PD-L1 ABDs that comprise variable heavy and / or variable light domains that are variants of the anti-PD-L1 ABD VH and VL domains disclosed herein. In one embodiment, the variant VH and / or VL domains have 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid changes from the VH and / or VL domains of an anti-PD-L1 ABD described herein, including Figures 28-31 and the Sequence Listing. In exemplary embodiments, the variant VH and / or VL domains have 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid changes from the VH and / or VL selected from the following:

[0162] (i) a VH having the amino acid sequence of SEQ ID NO: 21, and (ii) a VL having the amino acid sequence of SEQ ID NO: 25, and

[0163] (i) a VH having the amino acid sequence of SEQ ID NO: 29, and (ii) a VL having the amino acid sequence of SEQ ID NO: 33;

[0164] (i) a VH having the amino acid sequence of SEQ ID NO: 37, and (ii) a VL having the amino acid sequence of SEQ ID NO: 33;

[0165] (i) a VH having the amino acid sequence of SEQ ID NO: 29, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20, and

[0166] (i) VH having the amino acid sequence of SEQ ID NO: 37, and (ii) VL having the amino acid sequence of SEQ ID NO: 20 (Figures 24 to 26).

[0167] In certain embodiments, the anti-PD-L1 ABD is capable of binding to PD-L1 as measured by at least one of Biacore, surface plasmon resonance (SPR) and / or BLI (biolayer interferometry, e.g., Octet assay) assays (the latter of which is particularly utilized in many embodiments). In certain embodiments, the PD-L1 ABD is capable of binding to human and cynomolgus PD-L1.

[0168] In some embodiments, the PD-L1 ABD comprises a variable heavy chain domain with 1, 2, 3, 4, 5, 6, 7, 8, 9 or 10 amino acid changes in one of the following PD-L1 ABD variable heavy chain domains: SEQ ID NOs: 21, 29, and 37. In some embodiments, the PD-L1 ABD comprises any of the variable light chain domains or variants thereof provided herein.

[0169] In some embodiments, the PD-L1 ABD comprises a variable light chain domain with 1, 2, 3, 4, 5, 6, 7, 8, 9 or 10 amino acid changes in one of the following PD-L1 ABD variable light chain domains: SEQ ID NOs: 20, 25, and 33. In some embodiments, the PD-L1 ABD comprises any of the variable heavy chain domains or variants thereof provided herein.

[0170] In one embodiment, the variant VH and / or VL Domains are at least 90, 95, 97, 98, or 99% identical to the VH and / or VL of an anti-PD-L1 ABD as described herein, including in Figures 25 and 26, and the Sequence Listing. In an exemplary embodiment, the variant VH and / or VL Domains are at least 90, 95, 97, 98, or 99% identical to the VH and / or VL of one of the anti-PD-L1 ABDs, the VH and VL of which are selected from the following:

[0171] (i) a VH having the amino acid sequence of SEQ ID NO: 21, and (ii) a VL having the amino acid sequence of SEQ ID NO: 25, and

[0172] (i) a VH having the amino acid sequence of SEQ ID NO: 29, and (ii) a VL having the amino acid sequence of SEQ ID NO: 33;

[0173] (i) a VH having the amino acid sequence of SEQ ID NO: 37, and (ii) a VL having the amino acid sequence of SEQ ID NO: 33;

[0174] (i) a VH having the amino acid sequence of SEQ ID NO: 29, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20, and

[0175] (i) VH having the amino acid sequence of SEQ ID NO: 37, and (ii) VL having the amino acid sequence of SEQ ID NO: 20 (Figures 24 to 26).

[0176] In certain embodiments, the anti-PD-L1 ABD is capable of binding to PD-L1 as measured by at least one of Biacore, surface plasmon resonance (SPR) and / or BLI (biolayer interferometry, e.g., Octet assay) assays (the latter of which is particularly utilized in many embodiments). In certain embodiments, the PD-L1 ABD is capable of binding to human and cynomolgus PD-L1.

[0177] In some embodiments, the PD-L1 ABD comprises a variable heavy chain domain that is at least 90, 95, 97, 98, or 99% identical to one of the following PD-L1 ABD variable heavy chain domains: SEQ ID NOs: 21, 29, and 37. In some embodiments, the PD-L1 ABD comprises any of the variable light chain domains or variants thereof provided herein.

[0178] In some embodiments, the PD-L1 ABD comprises a variable light chain domain that is at least 90, 95, 97, 98, or 99% identical to one of the following PD-L1 ABD variable light chain domains: SEQ ID NOs: 20, 25, and 33. In some embodiments, the PD-L1 ABD comprises any of the variable heavy chain domains or variants thereof provided herein.

[0179] In some embodiments, the PD-L1 ABD has variable heavy and variable light domains that are 90, 95, 97, 98, or 99% identical to the VH and / or VL domains of a PD-L1 ABD as described herein, but with identical CDRs. In some embodiments, the VH is selected from the group consisting of SEQ ID NOs: 21, 29, and 37. In some embodiments, the VL is selected from the group consisting of SEQ ID NOs: 20, 25, and 33.

[0180] In some embodiments, the PD-L1 ABD has variable heavy and variable light domains that are 90, 95, 97, 98, or 99% identical to the VH and / or VL domains of a PD-L1 ABD as described herein, and there may be 0-6 amino acid modifications in the CDRs. In some embodiments, the VH is selected from the group consisting of SEQ ID NOs: 21, 29, and 37. In some embodiments, the VL is selected from the group consisting of SEQ ID NOs: 20, 25, and 33.

[0181] In some embodiments, the PD-L1 ABD has variable heavy and variable light domains that are 90, 95, 97, 98, or 99% identical to the VH and / or VL domains of a PD-L1 ABD as described herein, and may have 0-6 amino acid modifications in the CDRs, but no CDR has more than 1 amino acid modification. In some embodiments, the VH is selected from the group consisting of SEQ ID NOs: 21, 29, and 37. In some embodiments, the VL is selected from the group consisting of SEQ ID NOs: 20, 25, and 33.

[0182] In certain embodiments, the PD-L1 ABD is capable of binding to PD-L1 as measured by at least one of a Biacore, surface plasmon resonance (SPR) and / or BLI (biolayer interferometry, e.g., Octet assay) assay (the latter of which is particularly utilized in many embodiments). In certain embodiments, the CD28 ABD is capable of binding human PD-L1 (see Figure 2) at the limit of detection of the assay.

[0183] Such PD-L1 binding domains may be included in any of the antibodies provided herein, including, for example, the bispecific and trispecific antibody formats provided in Figures 82 and 83.

[0184] 1. Additional PD-L1 binding domains In another aspect, provided herein are additional PD-L1 binding domains that may be used in anti-PD-L1 antibodies, including any of the anti-PD-L1 antibodies described herein. In some embodiments, the anti-PD-L1 binding domain comprises a variable heavy chain domain selected from any of the sequences in Figure 161 (SEQ ID NOS: 3335, and 3243-3260), or a variant thereof, and a common light chain having the variable light chain domain designated "IGKV1-39_L1" (also referred to as "2A3A4.248[PDL1]_L1" (SEQ ID NO: 3239, see Figure 161A) and "1F12A4.249[PDL2]_L1" (SEQ ID NO: 3271, see Figure 162A), or a variant thereof, where the common light chain may also be used as the light chain for the CD28 and / or PD-L2 binding domain. In some embodiments, the anti-PD-L1 antibodies provided herein (e.g., anti-CD28×anti-PD-L1 and anti-CD28×anti-PD-L1×anti-PDL2 antibodies) comprise a PD-L1 binding domain that comprises a common light chain with the IGKV1-39_L1 variable light chain domain. In some embodiments, the anti-PD-L1 antibodies are anti-CD28×anti-PD-L1 antibodies that have a CD28 binding domain and a PD-L1 binding domain, each comprising a common light chain with the IGKV1-39_L1 variable light chain domain (SEQ ID NO: 3239). In some embodiments, the anti-PD-L1 antibodies are anti-CD28×anti-PD-L1×anti-PD-L2 antibodies that have a CD28 binding domain, a PD-L1 binding domain, and a PD-L2 binding domain, each comprising a common light chain with the IGKV1-39_L1 variable light chain domain (SEQ ID NO: 3239). Such PD-L1 binding domains that utilize a common light chain having the IGKV1-39_L1 Variable Light Chain Domain (SEQ ID NO: 3239) may be used, for example, in any of the antibody formats provided herein that utilize a common light chain (see, for example, the 1+1 CLC, 2+1 CLC, 1+1+1 stacked Fab2-scFv-Fc, 1+1+1 Fab-(Fab-scFv)-Fc, 1+1+1 mAb-scFv, and 1+1+1 stacked Fab2-Fab-Fc formats disclosed herein in Figures 82 and 83).

[0185] In one embodiment, the PD-L1 antigen-binding domain comprises vhCDRs 1-3 of any of the PD-L1 variable heavy chain domains shown in Figure 161 (SEQ ID NOs: 3335, and 3243-3260), and vlCDRs 1-3 of the IGK1-39_L1 variable light chain domain (SEQ ID NO: 3239).

[0186] In one embodiment, the PD-L1 ABD of a subject anti-PD-L1 antibody described herein comprises a) a vhCDR1, vhCDR2, and / or vhCDR3 that have 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid modifications compared to the vhCDR1, vhCDR2, and / or vhCDR3, respectively, of one of the PD-L1 variable heavy chain domains shown in Figure 161 (SEQ ID NOs: 3335, and 3243-3260), and / or b) a vlCDR1, vlCDR2, and / or vlCDR3 that have 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid modifications compared to the vlCDR1, vlCDR2, and / or vlCDR3, respectively, of the IGK1-39_L1 variable light chain domain (SEQ ID NO: 3239). In certain embodiments, the PD-L1 ABD of a subject anti-PD-L1 antibody is capable of binding the PD-L1 antigen as measured by at least one of Biacore, surface plasmon resonance (SPR), flow cytometry, and / or BLI (bio-layer interferometry, e.g., Octet assay) assays (the latter of which is particularly utilized in many embodiments). In certain embodiments, the PD-L1 ABD is capable of binding the human PD-L1 antigen (see Figure 1).

[0187] In one embodiment, the PD-L1 ABD of a subject anti-PD-L1 antibody comprises vhCDR1, vhCDR2, and / or vhCDR3 that are at least 90, 95, 97, 98, or 99% identical, respectively, to the vhCDR1, vhCDR2, and / or vhCDR3 of one of the PD-L1 variable heavy chain domains shown in Figure 161 (SEQ ID NOs:3335, and 3243-3260), and / or vlCDR1, vlCDR2, and / or vlCDR3 that are at least 90, 95, 97, 98, or 99% identical, respectively, to the vlCDR1, vlCDR2, and / or vlCDR3 of the IGK1-39_L1 variable light chain domain (SEQ ID NO:3239). In certain embodiments, the PD-L1 ABD is capable of binding to PD-L1 as measured by at least one of Biacore, surface plasmon resonance (SPR), flow cytometry, and / or BLI (bio-layer interferometry, e.g., Octet assay) assays (the latter of which is particularly utilized in many embodiments). In certain embodiments, the PD-L1 ABD is capable of binding the human PD-L1 antigen (see Figure 1).

[0188] In another exemplary embodiment, the PD-L1 ABD of a subject anti-PD-L1 antibody comprises a variable heavy chain (VH) domain of one of the PD-L1 variable heavy chain domains shown in Figure 161 (SEQ ID NOs: 3335, and 3243-3260), and the IGK1-39_L1 variable light chain domain (SEQ ID NO: 3239).

[0189] In some embodiments, the anti-PD-L1 antibody comprises a PD-L1 ABD comprising a variable heavy chain domain that is a variant of one of the PD-L1 variable heavy chain domains shown in Figure 161 (SEQ ID NOs: 3335, and 3243-3260), and / or a variable light chain domain that is a variant of the IGK1-39_L1 variable light chain domain (SEQ ID NO: 3239). In some embodiments, the variant VH domain has 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid changes compared to one of the PD-L1 variable heavy chain domains shown in Figure 161 (SEQ ID NOs: 3335, and 3243-3260), and / or the variable light chain domain has 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid changes from the IGK1-39_L1 variable light chain domain (SEQ ID NO: 3239). In some embodiments, the one or more amino acid changes are in the VH and / or VL framework regions (FR1, FR2, FR3, and / or FR4). In some embodiments, the one or more amino acid change(s) are in one or more CDRs. In certain embodiments, the PD-L1 ABD is capable of binding to PD-L1 as measured by at least one of Biacore, surface plasmon resonance (SPR), flow cytometry, and / or BLI (bio-layer interferometry, e.g., Octet assay) assays (the latter of which is particularly utilized in many embodiments). In certain embodiments, the PD-L1 ABD is capable of binding the human PD-L1 antigen (see Figure 1).

[0190] In one embodiment, the variant VH domain is at least 90, 95, 97, 98, or 99% identical to one of the PD-L1 variable heavy chain domains shown in Figure 161 (SEQ ID NOs: 3335, and 3243-3260), and / or the variable light chain domain is at least 90, 95, 97, 98, or 99% identical to the IGK1-39_L1 variable light chain domain (SEQ ID NO: 3239). In certain embodiments, the PD-L1 ABD is capable of binding to PD-L1 as measured by at least one of Biacore, surface plasmon resonance (SPR), flow cytometry, and / or BLI (bio-layer interferometry, e.g., Octet assay) assays (the latter of which is particularly utilized in many embodiments). In certain embodiments, the PD-L1 ABD is capable of binding the human PD-L1 antigen (see Figure 1).

[0191] In some embodiments, the PD-L1 binding domain comprises a VH comprising any one of the VH CDR1-3 and / or HFR1-4 sequences shown in Figure 161C, and an IGK1-39_L1 variable light chain domain (SEQ ID NO: 3239) or a variant thereof.

[0192] In another aspect, provided herein is a PD-L1 binding domain that competes with any of the PD-L1 binding domains disclosed herein for binding to human PD-L1.

[0193] B. PD-L2 antigen-binding domain In one aspect, provided herein are PD-L2 antigen-binding domains (ABDs), including anti-PD-L2 antibodies (e.g., anti-PD-L1 x anti-CD28 bispecific antibodies, and anti-PD-L1 x anti-PD-L2 x anti-CD28 trispecific antibodies), and compositions comprising such PD-L2 antigen-binding domains (ABDs). Such PD-L2 binding domains, and related antibodies (e.g., the bispecific and trispecific antibodies disclosed herein), are utilized, for example, in the treatment of PD-L2-associated cancers. In some embodiments, the PD-L2 ABD can bind to human and cynomolgus monkey PD-L2 (see Figure 2 and Example 1).

[0194] As will be appreciated by those skilled in the art, a suitable PD-L2 binding domain may comprise a set of six CDRs as shown in Figures 29-32 and the Sequence Listing. A suitable PD-L2 ABD may also comprise these sequences used as an scFv or as a Fab domain, as well as the entire VH and VL sequences as shown in Figures 29-32 and the Sequence Listing.

[0195] In one embodiment, the PD-L2 antigen-binding domain comprises the six CDRs (i.e., vhCDR1-3 and vlCDR1-3) of the PD-L2 ABD described herein, including Figures 29-32 and the Sequence Listing, either as the underlined CDRs, or as the CDRs identified using other alignments within the variable heavy (VH) domain and variable light domain (VL) sequences of those shown in Figures 29-32 and the Sequence Listing (see Table 2), when a different numbering scheme is used as described herein and as shown in Table 2. Suitable PD-L2 ABDs can also include these sequences and the entire VH and VL sequences shown in the figures, used as an scFv or as a Fab domain.

[0196] In one embodiment, the PD-L2 antigen-binding domain comprises the six CDRs of the PD-L2 ABD described herein, including in Figures 29-32 and the Sequence Listing (i.e., vhCDR1-3 and vlCDR1-3). In an exemplary embodiment, the PD-L2 antigen-binding domain comprises the six CDRs of PD-L2, including a variable heavy domain (VH) and a variable light domain (VL), where the VH and VL are selected from the following:

[0197] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 41; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 45.

[0198] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of the VH CDR1, the VH CDR2, and the VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 49; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of the VL CDR1, the VL CDR2, and the VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 45.

[0199] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 41; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0200] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 49; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0201] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 53; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 57;

[0202] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 61; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 57;

[0203] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 53; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0204] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 61; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0205] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of the VH CDR1, the VH CDR2, and the VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 65; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of the VL CDR1, the VL CDR2, and the VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 69;

[0206] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 73; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 69;

[0207] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 65; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0208] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 73; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0209] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 77; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 81;

[0210] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 85; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 81;

[0211] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 77; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20; and

[0212] (i) a VH having the amino acid sequence of SEQ ID NO: 85, and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of the VL CDR1, the VL CDR2, and the VL CDR3, respectively, of a VL having the amino acid sequence of SEQ ID NO: 20 (Figures 29 to 32) (Figures 24 and 29 to 32).

[0213] In addition to the parent CDR sets disclosed in the Figures and Sequence Listing that form the ABD for PD-L2, provided herein are variant PD-L2 ABDs having CDRs that comprise at least one modification of a PD-L2 ABD CDR disclosed herein. In one embodiment, a PD-L2 ABD comprises a set of six CDRs that have 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 amino acid modifications compared to the six CDRs of a PD-L2 ABD described herein, including the Figures and Sequence Listing. In an exemplary embodiment, a PD-L2 ABD comprises a set of six CDRs that have 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 amino acid modifications compared to the six CDRs of a PD-L2 ABD comprising a variable heavy domain (VH) and a variable light domain (VL) selected from the following:

[0214] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 41; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 45.

[0215] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of the VH CDR1, the VH CDR2, and the VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 49; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of the VL CDR1, the VL CDR2, and the VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 45.

[0216] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 41; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0217] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 49; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0218] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 53; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 57;

[0219] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 61; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 57;

[0220] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 53; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0221] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 61; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0222] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of the VH CDR1, the VH CDR2, and the VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 65; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of the VL CDR1, the VL CDR2, and the VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 69;

[0223] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 73; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 69;

[0224] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 65; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0225] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 73; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0226] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 77; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 81;

[0227] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 85; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 81;

[0228] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 77; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20; and

[0229] (i) a VH having the amino acid sequence of SEQ ID NO: 85, and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of the VL CDR1, the VL CDR2, and the VL CDR3, respectively, of a VL having the amino acid sequence of SEQ ID NO: 20 (Figures 29 to 32) (Figures 24 and 29 to 32).

[0230] In certain embodiments, the variant PD-L2 ABD is capable of binding the PD-L2 antigen as measured by at least one of Biacore, surface plasmon resonance (SPR) and / or BLI (biolayer interferometry, e.g., Octet assay) assays (the latter of which is particularly utilized in many embodiments). In certain embodiments, the PD-L2 ABD is capable of binding to human and cynomolgus PD-L2.

[0231] In one embodiment, the anti-PD-L2 ABD comprises six CDRs that are at least 90, 95, 97, 98, or 99% identical to the six CDRs of an anti-PD-L2 ABD as described herein, including in Figures 24, and 29-32, and the Sequence Listing. In an exemplary embodiment, the anti-PD-L2 ABD comprises six CDRs that are at least 90, 95, 97, 98, or 99% identical to the six CDRs of a PD-L2 ABD comprising a variable heavy domain (VH) and a variable light domain (VL) selected from the following:

[0232] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 41; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 45.

[0233] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of the VH CDR1, the VH CDR2, and the VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 49; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of the VL CDR1, the VL CDR2, and the VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 45.

[0234] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 41; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0235] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 49; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0236] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 53; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 57;

[0237] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 61; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 57;

[0238] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 53; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0239] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 61; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0240] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of the VH CDR1, the VH CDR2, and the VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 65; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of the VL CDR1, the VL CDR2, and the VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 69;

[0241] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 73; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 69;

[0242] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 65; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0243] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 73; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20;

[0244] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 77; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 81;

[0245] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 85; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 81;

[0246] (i) a VH comprising a VH CDR1, a VH CDR2, and a VH CDR3 having the amino acid sequences of VH CDR1, VH CDR2, and VH CDR3, respectively, of the VH having the amino acid sequence of SEQ ID NO: 77; and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of VL CDR1, VL CDR2, and VL CDR3, respectively, of the VL having the amino acid sequence of SEQ ID NO: 20; and

[0247] (i) a VH having the amino acid sequence of SEQ ID NO: 85, and (ii) a VL comprising a VL CDR1, a VL CDR2, and a VL CDR3 having the amino acid sequences of the VL CDR1, the VL CDR2, and the VL CDR3, respectively, of a VL having the amino acid sequence of SEQ ID NO: 20 (Figures 29 to 32) (Figures 24 and 29 to 32).

[0248] In certain embodiments, the anti-PD-L2 ABD is capable of binding to the PD-L2 antigen as measured by at least one of Biacore, surface plasmon resonance (SPR), and / or BLI (biolayer interferometry, e.g., Octet assay) assays (the latter of which is particularly utilized in many embodiments). In certain embodiments, the PD-L2 ABD is capable of binding to human and cynomolgus PD-L2.

[0249] In another exemplary embodiment, the anti-PD-L2 ABD comprises the variable heavy (VH) domain and / or variable light (VL) domain of any one of the PD-L2 ABDs described herein, including Figures 24, and 29-32, and the Sequence Listing. In an exemplary embodiment, the VH and VL are selected from the following:

[0250] (i) a VH having the amino acid sequence of SEQ ID NO: 41, and (ii) a VL having the amino acid sequence of SEQ ID NO: 45;

[0251] (i) a VH having the amino acid sequence of SEQ ID NO: 49, and (ii) a VL having the amino acid sequence of SEQ ID NO: 45;

[0252] (i) a VH having the amino acid sequence of SEQ ID NO: 41, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20, and

[0253] (i) a VH having the amino acid sequence of SEQ ID NO: 49, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20;

[0254] (i) a VH having the amino acid sequence of SEQ ID NO: 53, and (ii) a VL having the amino acid sequence of SEQ ID NO: 57;

[0255] (i) a VH having the amino acid sequence of SEQ ID NO: 61, and (ii) a VL having the amino acid sequence of SEQ ID NO: 57;

[0256] (i) a VH having the amino acid sequence of SEQ ID NO: 53, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20;

[0257] (i) a VH having the amino acid sequence of SEQ ID NO: 61, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20;

[0258] (i) a VH having the amino acid sequence of SEQ ID NO: 65, and (ii) a VL having the amino acid sequence of SEQ ID NO: 69;

[0259] (i) a VH having the amino acid sequence of SEQ ID NO: 73, and (ii) a VL having the amino acid sequence of SEQ ID NO: 69;

[0260] (i) a VH having the amino acid sequence of SEQ ID NO: 65, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20;

[0261] (i) a VH having the amino acid sequence of SEQ ID NO: 73, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20;

[0262] (i) a VH having the amino acid sequence of SEQ ID NO: 77, and (ii) a VL having the amino acid sequence of SEQ ID NO: 81;

[0263] (i) a VH having the amino acid sequence of SEQ ID NO: 85, and (ii) a VL having the amino acid sequence of SEQ ID NO: 81;

[0264] (i) a VH having the amino acid sequence of SEQ ID NO: 77, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20, and

[0265] (i) VH having the amino acid sequence of SEQ ID NO: 85, and (ii) VL having the amino acid sequence of SEQ ID NO: 20 (Figures 24, and 29 to 32).

[0266] In addition to the parent anti-PD-L2 binding domain variable heavy and variable light domains disclosed herein, provided herein are anti-PD-L2 ABDs that comprise variable heavy and / or variable light domains that are variants of the anti-PD-L2 ABD VH and VL domains disclosed herein. In one embodiment, the variant VH and / or VL domains have 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid changes from the VH and / or VL domains of an anti-PD-L2 ABD described herein, including Figures 28-31 and the Sequence Listing. In exemplary embodiments, the variant VH and / or VL domains have 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid changes from the VH and / or VL selected from the following:

[0267] (i) a VH having the amino acid sequence of SEQ ID NO: 41, and (ii) a VL having the amino acid sequence of SEQ ID NO: 45;

[0268] (i) a VH having the amino acid sequence of SEQ ID NO: 49, and (ii) a VL having the amino acid sequence of SEQ ID NO: 45;

[0269] (i) a VH having the amino acid sequence of SEQ ID NO: 41, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20, and

[0270] (i) a VH having the amino acid sequence of SEQ ID NO: 49, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20;

[0271] (i) a VH having the amino acid sequence of SEQ ID NO: 53, and (ii) a VL having the amino acid sequence of SEQ ID NO: 57;

[0272] (i) a VH having the amino acid sequence of SEQ ID NO: 61, and (ii) a VL having the amino acid sequence of SEQ ID NO: 57;

[0273] (i) a VH having the amino acid sequence of SEQ ID NO: 53, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20;

[0274] (i) a VH having the amino acid sequence of SEQ ID NO: 61, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20;

[0275] (i) a VH having the amino acid sequence of SEQ ID NO: 65, and (ii) a VL having the amino acid sequence of SEQ ID NO: 69;

[0276] (i) a VH having the amino acid sequence of SEQ ID NO: 73, and (ii) a VL having the amino acid sequence of SEQ ID NO: 69;

[0277] (i) a VH having the amino acid sequence of SEQ ID NO: 65, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20;

[0278] (i) a VH having the amino acid sequence of SEQ ID NO: 73, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20;

[0279] (i) a VH having the amino acid sequence of SEQ ID NO: 77, and (ii) a VL having the amino acid sequence of SEQ ID NO: 81;

[0280] (i) a VH having the amino acid sequence of SEQ ID NO: 85, and (ii) a VL having the amino acid sequence of SEQ ID NO: 81;

[0281] (i) a VH having the amino acid sequence of SEQ ID NO: 77, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20, and

[0282] (i) VH having the amino acid sequence of SEQ ID NO: 85, and (ii) VL having the amino acid sequence of SEQ ID NO: 20 (Figures 24, and 29 to 32).

[0283] In certain embodiments, the anti-PD-L2 ABD is capable of binding to PD-L2 as measured by at least one of Biacore, surface plasmon resonance (SPR) and / or BLI (biolayer interferometry, e.g., Octet assay) assays (the latter of which is particularly utilized in many embodiments). In certain embodiments, the PD-L2 ABD is capable of binding to human and cynomolgus PD-L2.

[0284] In some embodiments, the PD-L2 ABD comprises a variable heavy chain domain with 1, 2, 3, 4, 5, 6, 7, 8, 9 or 10 amino acid changes in one of the following PD-L2 ABD variable heavy chain domains: SEQ ID NOs: 41, 49, 53, 61, 65, 73, 77, and 85. In some embodiments, the PD-L2 ABD comprises any of the variable light chain domains or variants thereof provided herein.

[0285] In some embodiments, the PD-L2 ABD comprises a variable light chain domain with 1, 2, 3, 4, 5, 6, 7, 8, 9 or 10 amino acid changes in one of the following PD-L2 ABD variable light chain domains: SEQ ID NOs: 20, 45, 57, 69, and 81. In some embodiments, the PD-L2 ABD comprises any of the variable heavy chain domains or variants thereof provided herein.

[0286] In one embodiment, the variant VH and / or VL Domains are at least 90, 95, 97, 98, or 99% identical to the VH and / or VL of an anti-PD-L2 ABD as described herein, including in Figures 29-32, and the Sequence Listing. In exemplary embodiments, the variant VH and / or VL Domains are at least 90, 95, 97, 98, or 99% identical to the VH and / or VL of one of the anti-PD-L2 ABDs, the VH and VL of which are selected from the following:

[0287] (i) a VH having the amino acid sequence of SEQ ID NO: 41, and (ii) a VL having the amino acid sequence of SEQ ID NO: 45;

[0288] (i) a VH having the amino acid sequence of SEQ ID NO: 49, and (ii) a VL having the amino acid sequence of SEQ ID NO: 45;

[0289] (i) a VH having the amino acid sequence of SEQ ID NO: 41, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20, and

[0290] (i) a VH having the amino acid sequence of SEQ ID NO: 49, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20;

[0291] (i) a VH having the amino acid sequence of SEQ ID NO: 53, and (ii) a VL having the amino acid sequence of SEQ ID NO: 57;

[0292] (i) a VH having the amino acid sequence of SEQ ID NO: 61, and (ii) a VL having the amino acid sequence of SEQ ID NO: 57;

[0293] (i) a VH having the amino acid sequence of SEQ ID NO: 53, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20;

[0294] (i) a VH having the amino acid sequence of SEQ ID NO: 61, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20;

[0295] (i) a VH having the amino acid sequence of SEQ ID NO: 65, and (ii) a VL having the amino acid sequence of SEQ ID NO: 69;

[0296] (i) a VH having the amino acid sequence of SEQ ID NO: 73, and (ii) a VL having the amino acid sequence of SEQ ID NO: 69;

[0297] (i) a VH having the amino acid sequence of SEQ ID NO: 65, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20;

[0298] (i) a VH having the amino acid sequence of SEQ ID NO: 73, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20;

[0299] (i) a VH having the amino acid sequence of SEQ ID NO: 77, and (ii) a VL having the amino acid sequence of SEQ ID NO: 81;

[0300] (i) a VH having the amino acid sequence of SEQ ID NO: 85, and (ii) a VL having the amino acid sequence of SEQ ID NO: 81;

[0301] (i) a VH having the amino acid sequence of SEQ ID NO: 77, and (ii) a VL having the amino acid sequence of SEQ ID NO: 20, and

[0302] (i) VH having the amino acid sequence of SEQ ID NO: 85, and (ii) VL having the amino acid sequence of SEQ ID NO: 20 (Figures 24, and 29 to 32).

[0303] In certain embodiments, the anti-PD-L2 ABD is capable of binding to PD-L2 as measured by at least one of Biacore, surface plasmon resonance (SPR) and / or BLI (biolayer interferometry, e.g., Octet assay) assays (the latter of which is particularly utilized in many embodiments). In certain embodiments, the PD-L2 ABD is capable of binding to human and cynomolgus PD-L2.

[0304] In some embodiments, the PD-L2 ABD comprises a variable heavy chain domain that is at least 90, 95, 97, 98, or 99% identical to one of the following PD-L2 ABD variable heavy chain domains: SEQ ID NOs: 41, 49, 53, 61, 65, 73, 77, and 85. In some embodiments, the PD-L2 ABD comprises any of the variable light chain domains or variants thereof provided herein.

[0305] In some embodiments, the PD-L2 ABD comprises a variable light chain domain that is at least 90, 95, 97, 98, or 99% identical to one of the following PD-L2 ABD variable light chain domains: SEQ ID NOs: 20, 45, 57, 69, and 81. In some embodiments, the PD-L2 ABD comprises any of the variable heavy chain domains or variants thereof provided herein.

[0306] In some embodiments, the PD-L2 ABD has variable heavy and variable light domains that are 90, 95, 97, 98, or 99% identical to the VH and / or VL domains of a PD-L2 ABD as described herein, but with identical CDRs. In some embodiments, the VH is selected from the group consisting of SEQ ID NOs: 41, 49, 53, 61, 65, 73, 77, and 85. In some embodiments, the VL is selected from the group consisting of SEQ ID NOs: 20, 45, 57, 69, and 81.

[0307] In some embodiments, the PD-L2 ABD has variable heavy and variable light domains that are 90, 95, 97, 98, or 99% identical to the VH and / or VL domains of a PD-L2 ABD as described herein, and there may be 0-6 amino acid modifications in the CDRs. In some embodiments, the VH is selected from the group consisting of SEQ ID NOs: 41, 49, 53, 61, 65, 73, 77, and 85. In some embodiments, the VL is selected from the group consisting of SEQ ID NOs: 20, 45, 57, 69, and 81.

[0308] In some embodiments, the PD-L2 ABD has variable heavy and variable light domains that are 90, 95, 97, 98, or 99% identical to the VH and / or VL domains of a PD-L2 ABD as described herein, and may have 0-6 amino acid modifications in the CDRs, but no CDR has more than 1 amino acid modification. In some embodiments, the VH is selected from the group consisting of SEQ ID NOs: 41, 49, 53, 61, 65, 73, 77, and 85. In some embodiments, the VL is selected from the group consisting of SEQ ID NOs: 20, 45, 57, 69, and 81.

[0309] In certain embodiments, the PD-L2 ABD is capable of binding to PD-L2 as measured by at least one of a Biacore, surface plasmon resonance (SPR) and / or BLI (biolayer interferometry, e.g., Octet assay) assay (the latter of which is particularly utilized in many embodiments). In certain embodiments, the CD28 ABD is capable of binding human PD-L2 (see Figure 2) at the limit of detection of the assay.

[0310] Such PD-L2 binding domains may be included in any of the antibodies provided herein, including, for example, the bispecific and trispecific antibody formats provided in Figures 82 and 83.

[0311] 1. Additional PD-L2 binding domain In another aspect, provided herein are additional PD-L2 binding domains that may be used in anti-PD-L2 antibodies, including any of the anti-PD-L1 antibodies described herein. In some embodiments, the anti-PD-L1 binding domain comprises a variable heavy chain domain selected from any of the sequences in Figure 162 (SEQ ID NOS: 3267, and 3275-3347), or a variant thereof, and a common light chain having the variable light chain domain designated "IGKV1-39_L1" (also referred to as "2A3A4.248[PDL1]_L1" (SEQ ID NO: 3239, see Figure 161A) and "1F12A4.249[PDL2]_L1" (SEQ ID NO: 3271, see Figure 162A), or a variant thereof, where the common light chain may also be used as the light chain for the CD28 and / or PD-L1 binding domain. In some embodiments, the anti-PD-L1 antibodies provided herein (e.g., anti-CD28×anti-PD-L2 and anti-CD28×anti-PD-L1×anti-PD-L2 antibodies) comprise a PD-L2 binding domain that comprises a shared light chain with the IGKV1-39_L1 variable light chain domain. In some embodiments, the anti-PD-L2 antibody is an anti-CD28×anti-PD-L2 antibody that has a CD28 binding domain and a PD-L2 binding domain, each comprising a shared light chain with the IGKV1-39_L1 variable light chain domain (SEQ ID NO: 3239). In some embodiments, the anti-PD-L2 antibody is an anti-CD28×anti-PD-L1×anti-PD-L2 antibody that has a CD28 binding domain, a PD-L1 binding domain, and a PD-L2 binding domain, each comprising a shared light chain with the IGKV1-39_L1 variable light chain domain (SEQ ID NO: 3239). Such PD-L2 binding domains utilizing a common light chain having the IGKV1-39_L1 Variable Light Chain Domain (SEQ ID NO: 3239) may be used, for example, in any of the antibody formats provided herein that utilize a common light chain (see, for example, the 1+1 CLC, 2+1 CLC, 1+1+1 stacked Fab2-scFv-Fc, 1+1+1 Fab-(Fab-scFv)-Fc, 1+1+1 mAb-scFv, and 1+1+1 stacked Fab2-Fab-Fc formats disclosed herein in Figures 82 and 83).

[0312] In one embodiment, the PD-L2 antigen-binding domain comprises vhCDRs 1-3 of any of the PD-L2 variable heavy chain domains shown in Figure 161 (SEQ ID NOs: 3335, and 3243-3260), and vlCDRs 1-3 of the IGK1-39_L1 variable light chain domain (SEQ ID NO: 3239).

[0313] In one embodiment, the PD-L2 ABD of a subject anti-PD-L2 antibody described herein comprises a) a vhCDR1, vhCDR2, and / or vhCDR3 that have 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid modifications compared to the vhCDR1, vhCDR2, and / or vhCDR3, respectively, of one of the PD-L2 variable heavy chain domains shown in Figure 162 (SEQ ID NOs: 3267, and 3275-3347), and / or b) a vlCDR1, vlCDR2, and / or vlCDR3 that have 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 amino acid modifications compared to the vlCDR1, vlCDR2, and / or vlCDR3, respectively, of the IGK1-39_L1 variable light chain domain (SEQ ID NO: 3239). In certain embodiments, the PD-L2 ABD of a subject anti-PD-L2 antibody is capable of binding the PD-L2 antigen as measured by at least one of Biacore, surface plasmon resonance (SPR), flow cytometry, and / or BLI (bio-layer interferometry, e.g., Octet assay) assays (the latter of which is particularly utilized in many embodiments). In certain embodiments, the PD-L2 ABD is capable of binding the human PD-L2 antigen (see Figure 1).

[0314] In one embodiment, the PD-L2 ABD of a subject anti-PD-L2 antibody compr...

Claims

[Claim 1] A multispecific antibody, a) PD-L1 antigen-binding domain, b) PD-L2 antigen-binding domain, c) CD28 antigen-binding domain and The multispecific antibody comprising the above.

Images