Immunomodulatory agents

Abstract

To provide antibodies that specifically binds to a fusion molecule comprising PD-L1 and a PD-L1 binding protein constructed with an IL15 receptor binding region, to provide nucleic acid molecules encoding the same, and to provide therapeutic compositions thereof.SOLUTION: This agent inhibits PD-L1 mediated immunosuppression and enhances immunity for the treatment of cell- and cytokine-mediated neoplastic and infectious diseases.SELECTED DRAWING: None

Description

[Technical Field]

[0001] The present invention provides monoclonal antibodies that specifically bind to PD-L1 and bispecific fusion molecules comprising PD-L1 binding proteins constructed with IL15 and the IL15 receptor alpha sushi domain, nucleic acid molecules encoding same, and therapeutic compositions thereof, which enhance T cell and NK cell function to augment cell- and cytokine-mediated immunity for the treatment of a variety of diseases associated with immune dysfunction, including cancer and infectious diseases.

[0002] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority to U.S. Provisional Patent Application No. 61 / 927,907, filed January 15, 2014, which is incorporated herein by reference in its entirety. [Background technology]

[0003] Programmed death 1 (PD-1) is a member of the CD28 family of receptors, which includes CD28, CTLA-4, PD-1, ICOS, and BTLA (Freeman et al. (2000) J Exp Med 192:1027-34; Latchman et al. (2001) Nat Immunol 2:261-8). PD-1 is a triggering immunosuppressive receptor that is primarily upregulated on activated T cells and B cells during the progression of immunopathological conditions. Interaction of PD-1 with its ligand, PD-L1, results in inhibition of TCR- and BCR-mediated proliferation and cytokine production, as well as induction of apoptosis of antigen-specific T cells through unique PD-1-mediated negative signaling of immunoreceptor tyrosine inhibitory motifs (ITIMs) (Agata et al. (1996) Int. Immunol. 8:765; Unkeless and Jin. (1997) Curr. Opin. Immunol. 9:338-343; Okzaki et al. (2001) PNAS 98:13866-71; Dong et al. (2002) Nat. Med. 8:793-800). PD-L1 is a cell surface glycoprotein and the primary ligand for PD-1. PD-L1 can also be induced on lymphoid and peripheral non-lymphoid tissues after cell activation. PD-L1 is upregulated in various pathological cell types, including immune cells, cancer cells, and stromal cells, and plays an active role in immunosuppression during the progression of disease (Iwai et al. (2002) PNAS 99:12293-7, Ohigashi et al. (2005) Clin Cancer Res 11:2947-53). PD-L1 upregulation is associated with poor clinical outcomes in various cancers and viral infections (Hofmeyer et al. (2011) J. BioMed. Biotech. 2011:1-9, McDermott and Atkins. (2013) Cancer Med. 2:66 2-73). Blocking PD-1 or PD-L1 with antibodies promoted CD8 T cell infiltration, CTL activity, and increased the presence of the Th1 cytokine IFN-γ in preclinical and clinical settings (Zhou et al. (2010) J. Immunol. 185:5082-92, Nomi et al. (2007) Clin Cancer Res. 13:2152-7, Flies et al. (2011) Yale J. Bio. Med. 48:409-21, Zitvogel and Kroemer. (2012) Oncolmmunol. 1:1223-25). PD-L1 antibodies as immunosuppressive agents have been shown to be effective when used as monotherapy or in combination with other immunosuppressive molecules.

[0004] However, immunomodulating interventions targeting immunosuppressants only partially resolve the problems associated with immune disorders in cancer, infectious diseases, and other diseases. It remains highly desirable to utilize biotherapeutics to directly stimulate and expand effector immune cells to elevate weakened innate and adaptive immune responses to levels more effective in controlling tumors and infectious diseases. While immunotherapy using interleukins (i.e., IL-2, IL-12, IL-15, IL-21) and cytokines, including TNFα and GM-CSF, has shown some effectiveness in treating cancer and infectious diseases, clinical success is often limited by systemic toxicity associated with the high blood concentrations of cytokines required for efficacy and a lack of target specificity in diseased cells and tissues.

[0005] Among the cytokines evaluated, IL15 is recognized to exert antitumor immunity by exclusively stimulating effector and central memory CD8 cells, a subset of antigen-specific CD8 cells, without regulating other T cell populations. Furthermore, unlike IL-2, which activates Tregs, IL15 has been shown to activate natural killer (NK) cells and effector and memory CD8 T cells, as well as rescue T cells from apoptosis induced by Tregs and other immunosuppressive cells (Van Belle et al. (2012) PLoS One 7:e45299; Obar and Lefrancois. (2010) J. Immunol. 185:263-72; Pelletier and Girard. (2006) J. Immunol. 177:100-108; Elpek et al. (2010) PNAS 107:21647-21652).

[0006] IL15 was identified as a γc cytokine in 1994 based on its ability to stimulate proliferation of the murine T cell line CTLL-2 (Grabstein et al. (1994) Science 264:965-8; Bamford et al. (1996) PNAS 93:2897-902). Human IL15 shares approximately 97% and 96% amino acid sequence identity with monkey and cynomolgus monkey IL15, respectively. Human and mouse IL15 share 73% homology and are equally active on mouse cells. IL-15 is a 12.5-kD protein (114 amino acids) secreted as a 14-15 kDa glycoprotein by DCs, macrophages, and granulocytes. It is also a member of a four-α-helical bundle cytokine family (Anderson et al. (1995) J Biol Chem. 270:29862-9; Steel et al. (2012) Trends Pharmacol. Sci. 33:35-41). IL-15 normally forms a complex with the IL-15 receptor α expressed on APCs before binding to functional IL-15 receptor β and γ units on T cells and NK cells. IL-15 may be presented in trans to responder cells expressing CD122 and CD132 by cells expressing the cytokine itself, bound to the membrane form of the receptor α chain (Dubois et al. (2002) Immunity 17:537-47). The IL-15 receptor α sushi domain (29.5 kD in size) is involved in the complex with IL-15 before properly associating with receptors β and γ. IL15 is an important component of the immune system (Wei et al. (2001) J. Immunol. 167:277-82). IL15 and IL15Rα complexes, as well as IL15 / IL15Rα sushi domain fusion proteins, have been reported to have significantly higher in vitro and in vivo stimulation abilities for CD8 T cells and NK cells compared with IL15 alone (Mortier et al. (2005) J. Biol. Chem. 281:1612-19, Stoklasek et al. (2006) J. Immunol. 177:6072-80). IL15 also induces proliferation and differentiation of stimulated human B cells (Armitage et al. (1995) J. Immunol. 154:483-90). IL15 has been suggested to act primarily to prolong T lymphocyte survival, interfering with activation-induced cell death (AICD) (Marks-Konczalik et al. (2000) PNAS 97:11445-50). IL15 has exceptional ability to support the maintenance of NK cells, as well as memory phenotypes and antigen-specific memory CD8 T cells (Ma et al. (2006) Annu Rev Immunol. 24:657-79). Thus, among the most active cytokines in immune regulation, IL15 has the unique ability to mediate many important aspects of immunity against various tumor types and viral infections, including HIV, HBV, HCV, and LCMV (Steel et al. (2012) Trends Pharmacol. Sci. 33:35-41; Verbist and Klonowski, (2012) Cytokine. 59:467-478). Summary of the Invention [Problem to be solved by the invention]

[0007] The present invention provides antibodies and binding proteins that bind to PD-L1. In certain embodiments of the invention, the antibodies bind to PD-L1 and block its interaction with PD-1. By blocking the interaction of PD-L1 with PD-1, such antibodies are useful for reducing or inhibiting immune suppression. [Means for solving the problem]

[0008] In another aspect, the invention provides antibodies and binding proteins that specifically bind to PD-L1 and at least one other molecule. Examples of such embodiments include PD-L1 binding proteins, which may be membrane bound or soluble, that also bind to one or more other ligands and / or receptors.

[0009] In another aspect, the present invention provides molecules, e.g., fusion proteins that bind PD-L1, which, in addition to reducing or inhibiting immunosuppression by binding to PD-L1, also stimulate one or more immune responses by interacting with other ligands or receptors. In one embodiment of the present invention, the molecule binds to PD-L1 on target cells and also stimulates a cellular immune response, e.g., by promoting the proliferation of T cells and / or NK cells. In one embodiment of the present invention, the molecule stimulates cells that respond to interleukins or interferons, e.g., but not limited to, IL2, IL7, IL15, and IL21. In one embodiment of the present invention, the molecule comprises a sequence or domain of the IL15 receptor (IL15R) that promotes stimulation of IL15. In one embodiment of the present invention, the molecule that promotes stimulation of IL15R is a portion of the IL15R α chain that includes a sushi domain. In one embodiment of the present invention, the molecule provides the sushi domain of the IL15R α chain. In one embodiment of the present invention, the molecule provides a complex of IL15 and the sushi domain of the IL15R α chain, which may be covalently or non-covalently bound. The experiments disclosed herein demonstrate that a single molecule containing both a PD-L1-binding domain, which blocks the binding of PD-L1 to PD-1, and an IL15R-stimulating domain promotes a better immune response than the combined use of the separate molecules. More specifically, providing a molecule that provides an anti-PD-L1 antibody domain and a hybrid domain containing IL15 and the IL15 α-chain sushi domain promoted increased proliferation, increased Th1 cytokine release, and increased molecules associated with NK and T cell killing activity, compared to providing the domains in separate molecules.

[0010] In one embodiment, the invention provides an antibody or fragment that binds PD-L1, comprising a heavy chain CDR-1H having the sequence X1YX2MX3 (SEQ ID NO: 328), where Xi is A, G, M, Q, S, Y, or W, X2 is A, L, M, Q, R, S, V, W, or Y, and X3 is A, F, L, M, S, T, V, or Y; a heavy chain CDR-2H having the sequence of SEQ ID NO: 243; and a heavy chain CDR-3H having the sequence of SEQ ID NO: 245. In certain such embodiments, the heavy chain CDR-1H has a sequence selected from SEQ ID NO:241, SEQ ID NO:264, SEQ ID NO:266, SEQ ID NO:268, SEQ ID NO:270, SEQ ID NO:272, SEQ ID NO:274, SEQ ID NO:276, SEQ ID NO:278, SEQ ID NO:280, SEQ ID NO:282, SEQ ID NO:284, SEQ ID NO:286, SEQ ID NO:288, SEQ ID NO:290, SEQ ID NO:292, SEQ ID NO:294, SEQ ID NO:296, SEQ ID NO:298, SEQ ID NO:300, SEQ ID NO:302, SEQ ID NO:304, SEQ ID NO:306, SEQ ID NO:308, SEQ ID NO:310, and SEQ ID NO:312. In such embodiments, the heavy chain variable region is at least 80%, or at least 85%, or at least 90%, or at least 95% identical to SEQ ID NO:246. The antibody may further comprise a light chain variable region comprising CDR-1L having SEQ ID NO:247, CDR-2L having SEQ ID NO:248, and CDR-3L having SEQ ID NO:249. In some such embodiments, the light chain variable region is at least 80%, or at least 85%, or at least 90%, or at least 95% identical to SEQ ID NO:250. In another embodiment, the invention provides an antibody or fragment thereof that binds to PD-L1, wherein the light chain comprises CDR-1L having SEQ ID NO:247, CDR-2L having SEQ ID NO:248, and CDR-3L having SEQ ID NO:249.

[0011] The present invention also provides conjugates of antibodies to, for example, but not limited to, imaging agents, therapeutic agents, or cytotoxic agents.

[0012] The present invention further provides compositions comprising the antibodies and conjugates and a pharmaceutically acceptable carrier.

[0013] In another aspect, the invention provides fusion proteins capable of binding to PDL1 that also stimulate immune responses mediated by, for example, T cells or NK cells. In one embodiment of the invention, the fusion protein comprises a moiety that binds to the IL15 receptor. In another embodiment, the fusion protein comprises a moiety that binds to, for example, an interleukin receptor or an interferon receptor. In one embodiment of the invention, the moiety of the fusion protein that binds to PD-L1 is an antibody or PD-L1-binding fragment thereof. In one embodiment of the invention, the IL15 receptor binding moiety is IL15, and this binding may be improved by the presence in the fusion protein of the IL15Rα sushi domain.

[0014] The invention provides a method for inhibiting the interaction of PD-1 with PD-L1 in a subject, comprising administering an effective amount of an antibody or fragment of the invention. The invention further provides a method for inhibiting PD-L1-mediated immune suppression, comprising administering an effective amount of an antibody or fragment of the invention, or a fusion protein of the invention.

[0015] The invention further provides methods for stimulating an immune response against cells or tissues that express PD-L1, comprising administering to a subject an effective amount of an antibody or fragment of the invention, or a fusion protein of the invention. In certain embodiments, the PD-L1-expressing cells or tissues are neoplastic or infected cells. [Brief explanation of the drawings]

[0016] [Figure 1] Binding of antibodies tccR3λF8, tccR3κA11, tccR3λH4, tctR3κA8, sR3λD7, and R2κA6 to human hPDL1-Fc (upper left panel), blocking of hPDL1 to hPD1 (lower left panel), binding to mouse mPDL1-Fc (upper right panel), and blocking of mPDL1 to hPD1 (lower right panel) are shown. [Figure 2] Binding of antibodies sR3λD7, tccR3κB7, tccR3κA4, tccR3λF8, tccR3λD7, tccR3λH4, and tccR3κD9 to human hPDL1-Fc (upper left panel), blocking of hPDL1 to hPD1 (lower left panel), binding of mouse mPDL1-Fc (upper right panel), and blocking of mPDL1 to hPD1 (lower right panel) are shown. [Figure 3] Binding of antibodies tccR3κF8, tccR3κD9, tccR3λD7, tccR3λD7, sR3κF10, sR3λD7, and tccR3λF8 to human hPDL1-Fc (upper left panel), blocking of hPDL1 to hPD1 (lower left panel), binding of mouse mPDL1-Fc (upper right panel), and blocking of mPDL1 to hPD1 (lower right panel) are shown. [Figure 4] Binding of antibodies R2κA6, sR3λD7, tccR3λD7, tccR3κB7, and tccR3κH4 to human hPDL1-Fc (upper left panel), blocking of hPDL1 to hPD1 (lower left panel), binding of mouse mPDL1-Fc (upper right panel), and blocking of mPDL1 to hPD1 (lower right panel) are shown. [Figure 5] Binding of antibodies sR3λD7, tctR3κA8, tccR3κA11, tccR3λD7, tccR3κD9, tccR3λF8, tccR3κF8, tccR3κF10, tccR3λH4, tccR3κB7, and tccR3κA4 to PDL1-293 cells (top) and MDS-MB-231 (bottom) cells is shown. [Figure 6] Figure 1 shows the binding of anti-PD-L1 antibodies to (A) human monocyte-derived dendritic cells, (B) a human cancer cell line expressing PD-L1 MDA-MB-231 cells, and (C) a murine cell line expressing PD-L1 B16-F10. [Figure 7]Figure 1 shows the function-blocking activity of anti-PD-L1 antibodies as measured by (A) increased CD4 proliferation when activated with aCD3 and PD-L1 Fc-coated beads, (B) increased cytokine secretion in human PBMCs activated with SEB, and (C) increased CD4 proliferation in a mixed lymphocyte reaction with mo-DCs. [Figure 8] Figure 1 shows activation of CD4 and CD8 when both anti-PDL1 antibody and IL15 are present during (A) mixed lymphocyte reaction with mo-DCs and (B) stimulation of CD8 with αCD3 and PD-L1 Fc-coated beads. [Figure 9] Anti-PD-L1 sushi domain-IL15 (referred to as anti-PDL1-SD15) fusion protein maintains binding to PD-L1 as measured by (A) solid-phase ELISA and (B) binding to CD4 activated by aCD3-coated beads. [Figure 10] PBMCs cultured in vitro with anti-PD-L1-SD15 fusion protein resulted in increased numbers of NK cells (A), CD8 cells (B), and activation status (C) as measured by % Granzyme B. No effect was observed on CD4 cells (D). [Figure 11] Figure 1 shows that in the presence of aCD3-coated beads (A), anti-PD-L1-SD15 fusion protein functions similarly to IL15 to activate CD8 when stimulated in vitro. However, in the presence of aCD3 and PD-L1 Fc-coated beads, anti-PD-L1-SD15 fusion protein is able to increase CD8 proliferation by more than five-fold when compared to IL15 (B). cD7-SD15neg is anti-PD-L1 cD7 with non-functional IL15, which served as a negative control. [Figure 12]Figure 1 shows CD8 activation in the presence of PD-L1 Fc on antigen-presenting cells. The anti-PD-L1-SD15 fusion protein, cD7-SD15, stimulated CD8 at significantly lower concentrations, as measured by (A) an increase in the percentage of granzyme B-positive CD8 and (B) an increase in total cytokine secretion. Addition of cD7-SD15 also increased the maximal amount of CD8 activation in CD8 activated with aCD3 and PD-L1 Fc compared to IL15. (C) Data for CD8 proliferation in the presence of both anti-PD-L1 antibody and free IL15 (dotted line) overlap with data for CD8 proliferation in the presence of anti-PD-L1-SD15 fusion protein (straight line). [Figure 13A1] The amino acid sequences of the VH (Figures 13A1-3) and VL (Figures 13B1-3) chains of anti-PD-L1 antibodies are shown. For the VH sequences, the boxed regions represent the CDRs. For CDR-1H, the Chothia CDR is italicized, and the Kabat CDR is underlined. For CDR-2H, the Kabat CDR is coextensive with the Chothia CDR (italicized) in the boxed sequence. For the VL sequences, the boxed regions represent the Kabat / Chothia CDRs. [Figure 13A2] Same as above. [Figure 13A3] Same as above. [Figure 13B1] Same as above. [Figure 13B2] Same as above. [Figure 13B3] Same as above. [Figure 14] Shown are the amino acid sequences of SD15 (SEQ ID NO: 261), which contains the IL15Rα sushi domain and IL15, tccλD7HC-SD15 (SEQ ID NO: 262), and the LALA mutant of tccλD7HC-SD15 (SEQ ID NO: 263), which contains alanine substitutions of two adjacent leucines at positions (Leu234 and Leu235) within the heavy chain constant region important for FcγRI. [Figure 15]Figure 1 shows the toxicity of the anti-PD-L1-SD15 fusion protein (CD7SD15) compared to the PD-L1 binding portion of the molecule alone (cD7) and a fusion protein containing the KLH-specific binding domain and the IL-15 domain (KLHSD15). Human CD8 T cells and MAD-MB-231 tumor cells were co-cultured in IMDM supplemented with 10% FBS for 7 days. Tumor cell killing activity was assessed by measuring the number of dead tumor cells stained with Viability Dye eFluor 780 in a FACS. [Figure 16] This figure shows that anti-PD-L1-SD15 fusion protein extended the survival rate of mice bearing tumors expressing PD-L1. Balb / c mice were intravenously injected with 2x105 murine CT26 colon tumor cells. 24 hours later, the mice received intraperitoneal injections of the anti-PD-L1 antibody cD7 (purple line: 75µg per dose), the anti-PD-L1-SD15 fusion protein cD7-SD15 (green line: 75µg per dose, blue line: 25µg per dose), or sD7-SD15 (gray line: 75µg per dose, red line: 25µg per dose) twice weekly for the first week, then once weekly for the remainder of the treatment course. Control mice received an equal volume of saline or normal IgG solution. Survival rates were measured using Kaplan-Meier plots. [Figure 17] Binding of two affinity-matured anti-PDL1 antibodies to soluble human PDL1, soluble mouse PDL1, and soluble rat PDL1, and non-binding to human PDL2, is shown. [Figure 18] Blockade of human PD1 to human PDL1 (left panel) and mouse PD1 to mouse PDL1 (right panel) by two affinity matured anti-PDL1 antibodies compared to the parental tccλD7 antibody is shown. [Figure 19] Two affinity matured variants of the anti-PD-L1 antibody tccλD7 show higher binding activity to PD-L1-expressing human MDA-MB-231 tumor cells as measured by flow cytometry. [Figure 20]Affinity matured variants of the anti-PD-L1 antibody tccλD7 with increased ability to promote production of the Th1 cytokines IL2 (top panel) and IFNγ (bottom panel) are shown. [Figure 21] 1 shows the binding of fusion proteins of the invention to PD-L1-expressing MDA-MB-231 tumor cells. [Figure 22] 1 shows the stimulatory activity of the protein of the present invention on IL15-responsive human megakaryoblastic leukemia cells. [Figure 23] Figure 1 shows the hPD-L1 binding (left panel) and ligand blocking (right panel) activity of fusion proteins of the invention. [Figure 24] 1 shows the results of size exclusion chromatography for the fusion protein of the present invention. [Figure 25] 1 shows the serum stability of the fusion protein of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0017] Interaction of PD-1 with PD-L1 on immune cells inhibits proliferation and cytokine production by the immune cells. PD-L1 is also inducible and upregulated in a variety of tissues, including cancer. PD-1 and PD-L1 together play an immunosuppressive role. The present invention provides novel antibodies, and antigen-binding fragments of such antibodies, that bind to and block interaction with PD-L1. In an embodiment of the invention, the antibodies reduce or inhibit immunosuppression.

[0018] The novel antibodies of the present invention are described in Table 1 and the accompanying sequence listing, which set forth the heavy and light chain CDRs (identified according to the Kabat and Chothia identification system) and the amino acid sequences of the complete heavy and light chain variable regions. The first two heavy chain CDRs are identified according to the general Kabat and Chothia system, which provides distinct, yet overlapping, CDR locations. Comparison of multiple heavy and light chains shows significant similarities among many CDR sequences. Therefore, it is expected that many CDRs can be mixed and matched within a sequence.

[0019] The antibody can have one or more amino acid substitutions, deletions, insertions, and / or additions. In certain embodiments, the antibody comprises one of the heavy chain variable regions described above and one of the light chain variable regions described above. In certain embodiments, the PD-L1 antibody or binding fragment thereof comprises one or more CDRs or one or more variable regions that have amino acid sequences at least 85%, at least 90%, at least 95%, at least 97%, at least 98%, or at least 99% identical to the CDR and variable region sequences set forth in Table 1.

[0020] "Identity" refers to the number or percentage of identical positions shared by two amino acid or nucleic acid sequences, taking into account the number of gaps and the length of each gap, which must be introduced for optimal alignment of the two sequences. "Substantially identical" refers to amino acid sequences that differ only by conservative amino acid substitutions, e.g., substitution of an amino acid for another amino acid of the same class (e.g., valine for glycine, arginine for lysine, etc.), or by one or more non-conservative substitutions, deletions, or insertions at positions in the amino acid sequence that do not disrupt protein function. Amino acid substitutions can be made, as the case may be, by selecting substitutions that do not significantly differ in their effect on maintaining (a) the structure of the peptide backbone in the region of substitution, (b) the charge or hydrophobicity of the molecule at the target site, or (c) the bulk of the side chain. For example, naturally occurring residues can be divided into groups based on the properties of their side chains: (1) hydrophobic amino acids (norleucine, methionine, alanine, valine, leucine, and isoleucine); (2) neutral hydrophilic amino acids (cysteine, serine, and threonine); (3) acidic amino acids (aspartic acid and glutamic acid); (4) basic amino acids (asparagine, glutamine, histidine, lysine, and arginine); (5) amino acids that influence chain orientation (glycine and proline); and (6) aromatic amino acids (tryptophan, tyrosine, and phenylalanine). Substitutions made within these groups can be considered conservative substitutions. Examples of substitutions include, but are not limited to, valine for alanine, lysine for arginine, glutamine for asparagine, glutamic acid for aspartic acid, serine for cysteine, asparagine for glutamine, aspartic acid for glutamic acid, proline for glycine, and acetylcholine for histidine. These include substitutions of arginine, leucine for isoleucine, isoleucine for leucine, arginine for lysine, leucine for methionine, leucine for phenylalanine, glycine for proline, threonine for serine, serine for threonine, tyrosine for tryptophan, phenylalanine for tyrosine, and / or leucine for valine.

[0021] Preferably, the amino acid sequence is at least 80%, or at least 85%, or at least 90%, or at least 95% identical to the amino acid sequences disclosed herein. Methods and computer programs for measuring sequence similarity are publicly available and include, but are not limited to, the GCG program package (Devereux et al., Nucleic Acids Research 12:387, 1984), BLASTP, BLASTN, FASTA (Altschul et al., J. Mol. Biol. 215:403 (1990), and the ALIGN program (version 2.0). The well-known Smith-Waterman algorithm may also be used to measure similarity. BLAST programs are available from NCBI and other sources (BLAST Manual, Altschul, et al., NCBI NLM NIH, Bethesda, Md. 20894; BLAST 2.0, available at http: / / www.ncbi.nlm.nih.gov / blast / ). In comparing sequences, these methods examine various substitutions, deletions, and other mutations. Conservative substitutions typically include substitutions within the following groups: glycine, alanine; valine, isoleucine, leucine; aspartic acid, glutamic acid, asparagine, glutamine; serine, threonine, lysine, arginine; and phenylalanine, tyrosine.

[0022] Antibodies of the present invention also include antibodies whose binding properties have been improved by direct mutation, affinity maturation, phage display, or chain shuffling. Affinity and specificity may be modified or improved by mutating the CDRs and screening for antigen-binding sites with the desired properties. CDRs can be mutated in a variety of ways. One method is to randomize individual residues or combinations of residues so that all 20 amino acids are found at a particular position in a population of otherwise identical antigen-binding sites. Alternatively, error-prone PCR can be used to mutate a range of CDR residues (see, e.g., Hawkins et al., J. Mol. Biol., 226:889-896 (1992)). For example, phage display vectors containing heavy and light chain variable region genes can be propagated in a mutator strain of Escherichia coli (see, e.g., Low et al., J. Mol. Biol., 250:359-368 (1996)). These methods of mutagenesis are representative of the many known to those of skill in the art.

[0023] To minimize immunogenicity, antibodies containing human constant region sequences are preferred. The antibody may be a member of any immunoglobulin class (e.g., IgG, IgM, IgA, IgD, or IgE), and subclasses thereof, or a combination thereof. The antibody class may be selected to optimize natural antibody effector functions (e.g., complement-dependent cytotoxicity (CDC) and antibody-dependent cellular cytotoxicity (ADCC)).

[0024] Certain embodiments of the invention involve the use of PD-L1-binding antibody fragments. Fv is the smallest fragment encompassing the complete heavy and light chain variable regions, including all six hypervariable loops (CDRs). The constant regions are absent, and the variable regions are non-covalently linked. H and V LThe heavy and light chains may be joined into a single polypeptide chain ("single-chain Fv" or "scFv") using a linker that is capable of joining the regions and forming the antigen-binding site. In one embodiment of the present invention, the linker is (Gly-Gly-Gly-Gly-Ser)3. Because scFv fragments lack the constant regions of a whole antibody, they are significantly smaller than whole antibodies. scFv fragments also lack the normal heavy chain constant region interactions with other biological molecules, which may be undesirable in certain embodiments.

[0025] V H , V L , and optionally C L , C H Antibody fragments containing the Fab, Fab', or other constant regions can also be used. Monovalent antibody fragments produced by papain digestion are called Fab and lack the heavy chain hinge region. Fragments produced by pepsin digestion are called F(ab')2 and retain the heavy chain hinge and are bivalent. Such fragments may also be produced recombinantly. Many other useful antigen-binding antibody fragments are known in the art, including, but not limited to, diabodies, triabodies, single-domain antibodies, and other monovalent and multivalent forms.

[0026] The present invention also provides multivalent antigen-binding proteins, which can be in the form of, but are not limited to, antibodies, antigen-binding fragments thereof, and proteins consisting of all or part of the antigen-binding portion of an antibody. Multivalent antigen-binding proteins may be monospecific, bispecific, or multispecific. The term "specificity" refers to the number of different types of antigenic determinants to which a particular molecule can bind. If an immunoglobulin molecule binds to only one type of antigenic determinant, the immunoglobulin molecule is monospecific. If an immunoglobulin molecule binds to different types of antigenic determinants, the immunoglobulin molecule is multispecific.

[0027] In one embodiment of the invention, the PD-L1 binding protein has a PD-L1 binding activity of at least about 10 as measured by surface plasmon resonance. 2 M-1 s -1 ; at least about 10 3 M -1 s -1 ; at least about 10 4 M -1 s -1 ; at least about 10 5 M -1 s -1 ; or at least about 10 6 M -1 s -1 In one embodiment, the PD-L1 binding protein has an association rate constant (Kon) of 10 as measured by surface plasmon resonance. 2 M -1 s -1 ~10 3 M -1 s -1 ;10 3 M -1 s -1 ~10 4 M -1 s -1 ;10 4 M -1 s -1 ~10 5 M -1 s -1 ; or 10 5 M -1 s -1 ~10 6 M -1 s -1 The binding rate constant (Kon) is

[0028] In another embodiment, the PD-L1 binding protein has a binding affinity of up to about 10 as measured by surface plasmon resonance. -3 s -1 ;Up to about 10 -4 s -1 ;Up to about 10 -5 s -1 ; or up to about 10 -6 s -1 In one embodiment, the PD-L1 binding protein has a dissociation rate constant (Koff) of 10 as measured by surface plasmon resonance. -3 s -1 ~10 -4 s -1 ;10 -4s -1 ~10 -5 s -1 ; or 10 -5 s -1 ~10 -6 s -1 It has a dissociation rate constant (Koff) of

[0029] In another embodiment, the PD-L1 binding protein is up to about 10 -7 M; maximum about 10 -8 M; maximum about 10 -9 M; maximum about 10 -10 M; maximum about 10 -11 M; maximum about 10 -12 M; or up to about 10 -13 Dissociation constant of M (K d In one embodiment, the binding protein has a 10 -7 M~10 -8 M;10 -8 M~10 -9 M;10 -9 M~10 -10 M;10 -10 M~10 -11 M;10 -11 M~10 -12 M;10 -12 M~10 -13 Dissociation constant of M (K d )

[0030] The binding proteins described herein may be conjugated to further comprise an imaging agent, a therapeutic agent, or a cytotoxic agent. In one embodiment, the imaging agent is a radiolabel, an enzyme, a fluorescent label, a luminescent label, a bioluminescent label, a magnetic label, or biotin. In another embodiment, the radiolabel is 3 H, 14 C. 35 S, 90 Y, 99 Tc, 111 In, 125 I, 131 I, 177 Lu, or 153In yet other embodiments, the therapeutic or cytotoxic agent is an antimetabolite, an alkylating agent, an antibiotic, a growth factor, a cytokine, an antiangiogenic agent, an antimitotic agent, an anthracycline, a toxin, or an apoptotic agent. As discussed below, immune-stimulating cytokines are of particular interest.

[0031] The present invention also provides a method for inhibiting immunosuppression by binding to PD-L1 and inhibiting immunosuppression by binding to other ligands or receptors. Also provided are molecules that also stimulate immune responses by interacting with IL15R. As exemplified herein, such molecules combine the PD-L1-binding region of an antibody with a region that stimulates NK or T cell function. Such a stimulatory region can be a region that binds to and stimulates a receptor responsive to an interleukin or interferon, including, but not limited to, IL2, IL7, IL15, and IL21. An exemplary stimulatory region herein is a hybrid region (e.g., SEQ ID NO: 261) comprising the sushi domain of the IL15Rα chain linked to IL15 by a linker. An example of the complete molecule is represented by SEQ ID NO: 262. A nearly identical molecule modified with two amino acid substitutions between the antibody region and the IL15R stimulatory region, which inhibits proteolysis within the region, is represented by SEQ ID NO: 263. As demonstrated herein, a molecule comprising a PD-L1-binding region that inhibits immunosuppression and a second region that stimulates an immune response confers increased immune cell activity compared to two different molecules that confer these functions separately.

[0032] As exemplified herein, the PD-L1-binding portion of the molecule is the antigen-binding region of an antibody. Several novel antibody heavy and light chain variable regions, and antibodies comprising them, are provided. According to the present invention, the PD-L1-binding portion can be any agent that binds to PD-L1 and blocks immunosuppression. These include anti-PD-L1 antibodies and fragments, including, but not limited to, the novel antibodies disclosed herein, as well as PD1-derived peptides and proteins (PD-L1's natural ligands).

[0033] As disclosed herein, a PD-L1-binding domain is linked to a domain that stimulates NK and T cell activity. This domain comprises IL15 and is linked to the stimulatory domain by a flexible linker, the "sushi" domain of the α chain of the IL15 receptor. The sushi domain binds to IL15 with high affinity, and the complex of the sushi domain and IL15 is particularly active in stimulating NK and T cell proliferation. Of particular note, as shown in the Examples, treatment with an agent that combines a PD-L1-binding domain in the same molecule with an IL15-stimulatory domain is more effective than combination treatment using a PD-L1-binding domain and an IL15-stimulatory domain as separate molecules.

[0034] Thus, in certain embodiments, the present invention contemplates hybrid molecules comprising a region that binds to PD-L1 and blocks binding to PD1, and a region that stimulates IL15R, thereby stimulating immune cell proliferation. As illustrated, the IL15R-stimulating region comprises the sushi domain of the IL15R α chain linked to IL15 by a flexible linker, e.g., similar to those used for single-chain Fv molecules (i.e., containing 15-20 amino acids, primarily serine and glycine). Indeed, other methods may be employed, and these may be preferable, for example, for manufacturing procedures. Furthermore, some methods recognize the structure of the domains and, thus, the modular structure and other features of the disclosed hybrid proteins. For example, a linker linking the sushi domain to IL15 is useful for expressing the hybrid as a single polypeptide, but other agents, linkers, or cross-linking agents may be substituted as well. Alternatively, the high affinity of IL15 for the sushi-containing region of the IL15R α chain indicates that the sushi domain and IL15 may form a stable complex that does not require covalent binding. Similarly, while the exemplary proteins constitute the constant region of a whole antibody, other antigen-binding fragments of PD-L1 binding antibodies will suffice.

[0035] Thus, the present invention provides a PD-L1 binding region linked to an IL15R stimulatory region, the IL15R stimulatory region comprising the sushi domain of the IL15R α chain or a variant thereof, and IL15 or a variant thereof. In certain embodiments, the variant is 80%, 85%, 87%, 90%, 91%, 92%, 93%, 94%, or 95% identical to a sequence disclosed herein. In one embodiment, the sushi domain of the IL15R α chain and IL15 form a covalent complex. In another embodiment, the sushi domain of the IL15R α chain The shi domain and IL15 form a non-covalent complex. The PD-L1 binding region can comprise one, two, three, four, five, or six CDRs or heavy and / or light chain variable regions of an antibody, antigen-binding fragment thereof, or variant thereof disclosed herein (e.g., a variant that is 80%, 85%, 87%, 90%, 91%, 92%, 93%, 94%, or 95% identical), or a PD-L1 antibody or antigen-binding fragment thereof known in the art that blocks binding to PD-1.

[0036] It will be understood that when the anti-PD-L1 antibodies and hybrid proteins of the invention are used in mammals for prophylactic or therapeutic purposes, they will be administered in the form of a composition further comprising a pharmaceutically acceptable carrier. Suitable pharmaceutically acceptable carriers include, for example, one or more of water, saline, phosphate-buffered saline, dextrose, glycerol, sucrose, polysorbate, ethanol, etc., as well as combinations thereof. Pharmaceutically acceptable carriers may further contain minor amounts of auxiliary substances, such as wetting or emulsifying agents, preservatives, or buffers, which enhance the preservation or effectiveness of the antibody.

[0037] In the methods of the present invention, a therapeutically effective amount of an antibody of the hybrid protein of the present invention is administered to a mammal in need thereof. As used herein, the term "administering" refers to delivering the antibodies and fusion proteins of the present invention to a mammal by any method capable of achieving the desired result. The antibody may be administered, for example, intravenously or intramuscularly. While exemplary antibodies of the present invention are particularly useful for administration to humans, they may be administered to other mammals as well. As used herein, the term "mammal" is intended to include, but is not limited to, humans, laboratory animals, pets, and livestock. A "therapeutically effective amount" refers to an amount of an antibody of the present invention that, when administered to a mammal, is effective to produce a desired therapeutic effect, such as inhibition of kinase activity.

[0038] The antibodies and hybrid proteins of the present invention are useful for inhibiting tumors and other neoplastic diseases, as well as for treating other conditions associated with immunosuppression. Treatable tumors can include primary tumors, metastatic tumors, and refractory tumors. Refractory tumors include tumors that do not respond to or are resistant to treatment with chemotherapeutic agents alone, antibodies alone, radiation alone, or a combination thereof. Refractory tumors also include tumors that appear to be inhibited by such agents but recur up to five years, and in some cases ten years or more, after treatment has ceased. The antibodies are effective in treating vascularized tumors as well as tumors that are not vascularized or remain substantially vascularized.

[0039] Examples of solid tumors that can be treated with antibodies include breast cancer, lung cancer, colon cancer, pancreatic cancer, glioma, and lymphoma. Some examples of such tumors include epidermoid tumors, squamous tumors (e.g., head and neck tumors), colon tumors, prostate tumors, breast tumors, lung tumors including small cell and non-small cell lung tumors, pancreatic tumors, thyroid tumors, ovarian tumors, and liver tumors. Other examples include Kaposi's sarcoma, CNS tumors, neuroblastoma, capillary hemangioblastoma, meningioma and brain metastases, melanoma, gastrointestinal and renal cancer / sarcoma, rhabdomyosarcoma, glioblastoma, preferably glioblastoma multiforme, and leiomyosarcoma. Examples of vascularized skin cancers in which the antagonists of the present invention are effective include squamous cell carcinoma, basal cell carcinoma, and skin cancers that can be treated by inhibiting the proliferation of malignant keratinocytes (e.g., human malignant keratinocytes).

[0040] Examples of non-solid tumors include leukemia, multiple myeloma, and lymphoma that are unresponsive to cytokines such as IL15. Some examples of leukemia include acute myeloid leukemia (AML), chronic myeloid leukemia (CML), acute lymphocytic leukemia (ALL), chronic lymphocytic leukemia (CLL), erythrocytic leukemia, or monocytic leukemia. Some examples of lymphomas include Hodgkin's lymphoma and non-Hodgkin's lymphoma.

[0041] PD-L1 antibodies and immune cells stimulating hybrid proteins of the present invention may also be used to treat viral infections. PD-1 expression on T cells is associated with HIV viral load and HCV-infected patients, and PD-1 expression is associated with depleted virus-specific CD8 + It has been identified as a marker for T cells, e.g., PD-1 + CD8 + T cells exhibit impaired effector function and PD-1-mediated T cell exhaustion, which can be rescued by blocking PD-1 / PD-L1 interactions. This is consistent with the loss of virus-specific CD8 +These findings demonstrate the restoration of T cell-mediated immunity and the use of antagonistic antibodies to block PD-1 signaling restores T cell effector function. Immunotherapy based on PD-1 / PD-L1 blockade not only results in the breakdown of T cell tolerance to tumor antigens, but also provides a strategy to reactivate virus-specific effector T cells and eradicate pathogens in chronic viral infections. Thus, the antibodies and hybrid proteins of the present invention are useful for the treatment of chronic viral infections, including, but not limited to, HCV and HIV, and lymphocytic choriomeningitis virus (LCMV).

[0042] The antibodies and hybrid proteins of the present invention can be advantageously administered to a patient in need thereof in conjunction with a second agent. For example, in some embodiments, an antibody or hybrid protein of the present invention is administered to a subject in conjunction with an anti-neoplastic agent. In some embodiments, an antibody or hybrid protein of the present invention is administered to a subject in conjunction with a second anti-angiogenic agent. In some embodiments, an antibody or hybrid protein of the present invention is administered to a subject in conjunction with an anti-inflammatory or immunosuppressant agent.

[0043] Antineoplastic agents include cytotoxic chemotherapeutic agents, targeted small molecules and biomolecules, and radiation. Non-limiting examples of chemotherapeutic agents include cisplatin, dacarbazine (DTIC), dactinomycin, irinotecan, mechlorethamine (nitrogen mustard), streptozocin, cyclophosphamide, carmustine (BCNU), lomustine (CCNU), doxorubicin (adriamycin), daunorubicin, procarbazine, mitomycin, cytarabine, etoposide, methotrexate, 5-fluorouracil, vinblastine, vincristine, bleomycin, paclitaxel (Taxol), docetaxel (Taxotere), aldesroy, anticoagulants, such as rifabutin, asparaginase, busulfan, carboplatin, cladribine, dacarbazine, floxuridine, fludarabine, hydroxyurea, ifosfamide, interferon alpha, leuprolide, megestrol, melphalan, mercaptopurine, plicamycin, mitotane, pegaspargase, pentostatin, pipobroman, plicamycin, streptozocin, tamoxifen, teniposide, testolactone, thioguanine, thiotepa, uracil mustard, vinorelbine, chlorambucil, taxol, and combinations thereof.

[0044] Targeted small molecules and biomolecules include, but are not limited to, inhibitors of signal transduction pathway components, such as tyrosine kinase modulators and receptor tyrosine kinase inhibitors, and agents that bind to tumor-specific antigens. Non-limiting examples of growth factor receptors implicated in tumorigenesis are receptors for platelet-derived growth factor (PDGFR), insulin-like growth factor (IGFR), nerve growth factor (NGFR), and fibroblast growth factor (FGFR), as well as receptors of the epidermal growth factor receptor family, including EGFR (erbB1), HER2 (erbB2), erbB3, and erbB4.

[0045] EGFR antagonists include antibodies that bind to EGFR or EGFR ligands and inhibit ligand binding and / or receptor activation. For example, the agent can block the formation of receptor dimers or heterodimers with other EGFR family members. Ligands for EGFR include, for example, EGF, TGF-α amphiregulin, heparin-binding EGF (HB-EGF), and beta-regulin. EGFR antagonists include EGFR-specific antibodies (e.g., EGFR-specific tyrosine kinase inhibitors) and EGFR-specific tyrosine kinase inhibitors (e.g., EGFR-specific tyrosine kinase inhibitors). EGFR antagonists can bind externally to the extracellular portion of EGFR (which may or may not inhibit ligand binding) or can bind internally to the tyrosine kinase domain. EGFR antagonists further include agents that inhibit EGFR-dependent signal transduction, for example, by inhibiting the function of components of the EGFR signaling pathway. Examples of EGFR antagonists that bind to EGFR include, but are not limited to, biomolecules such as EGFR-specific antibodies (and their functional equivalents) and small molecules such as synthetic kinase inhibitors that act directly on the cytoplasmic domain of EGFR.

[0046] Small molecule and biological inhibitors include inhibitors of epidermal growth factor receptor (EGFR), including gefitinib, erlotinib, and cetuximab, inhibitors of HER2 (e.g., trastuzumab, trastuzumab emtansine (trastuzumab-DM1; T-DM1), and pertuzumab), anti-VEGF antibodies and fragments (e.g., bevacizumab), antibodies that inhibit CD20 (e.g., rituximab, ibritumomab), anti-VEGFR antibodies (e.g., ramucirumab (IMC-1121B), IMC-1C11, and CDP791), anti-PDGFR antibodies, and imatinib. Small molecule kinase inhibitors can be specific for a particular tyrosine kinase or can be inhibitors of more than one kinase. For example, the compound N-(3,4-dichloro-2-fluorophenyl)-7-({[(3aR,6aS)-2-methyloctahydrocyclopenta[c]pyrrol-5-yl]methyl}oxy)-6-(methyloxy)quinazolin-4-amine (also known as XL647, EXEL-7647, and KD-019) is an in vitro inhibitor of multiple receptor tyrosine kinases (RTKs), including EGFR, EphB4, KDR (VEGFR), Flt4 (VEGFR3), and ErbB2, and is also an inhibitor of SRC kinase, a pathway involved in tumor non-responsiveness to certain TKIs. In one embodiment of the present invention, treating a subject in need thereof includes administering a Rho kinase inhibitor compound of Formula I, and administering KD-019.

[0047] Dasatinib (BMS-354825; Bristol-Myers Squibb, New York) is another orally available, ATP-site competitive Src inhibitor. Dasatinib also targets Bcr-Abl (FDA-approved for use in patients with chronic myeloid leukemia (CML) or Philadelphia chromosome-positive (Ph+) acute lymphoblastic leukemia (ALL)), as well as c-kit, PDGFR, c-FMS, EphA2, and SFKs. Two other oral tyrosine kinase inhibitors of Src and Bcr-Abl are bosutinib (SKI-606) and saracatinib (AZD0530).

[0048] In one embodiment of the invention, the PD-L1 antibodies or conjugates of the invention are used in combination with antiviral agents to treat chronic viral infections. For example, with respect to HCV, the following agents can be used: HCV protease inhibitors include, but are not limited to, boceprevir, telaprevir (VX-950), ITMN-191, SCH-900518, TMC-435, BI-201335, MK-7009, VX-500, VX-813, BMS790052, BMS650032, and VBY376. HCV nonstructural protein 4B (NS4B) inhibitors include, but are not limited to, clemizole and other NS4B-RNA binding inhibitors, including, but not limited to, benzimidazole RBI (B-RBI) and indazole RBI (I-RBI). HCV nonstructural protein 5A (NS5A) inhibitors include, but are not limited to, BMS-790052, A-689, A-831, EDP239, GS5885, and PP1461. HCV polymerase (NS5B) inhibitors include nucleoside analogs (e.g., valopicitabine, R1479, R1626, R7128), nucleotide analogs (e.g., IDX184, PSI-7851, PSI-7977), and non-nucleoside analogs (e.g., filibuvir, HCV-7 Antivirals include, but are not limited to, ribavirin or ribavirin analogs such as taribavirin (viramidine; ICN 3142), mizoribine, merimepodib (VX-497), mycophenolate mofetil, and mycophenolate.

[0049] In certain embodiments, a single dose of an antibody or hybrid protein of the invention is administered to a subject daily, every other day, every few days, every three days, once a week, twice a week, three times a week, or once every two weeks. In other embodiments, two, three, or four doses of the compound or composition are administered to a subject daily, every few days, every three days, once a week, or once every two weeks. In some embodiments, the dose of the compound or composition is administered for 2, 3, 5, 7, 14, or 21 days. In certain embodiments, the dose of the compound or composition is administered for 1 month, 1.5 months, 2 months, 2.5 months, 3 months, 4 months, 5 months, 6 months, or more.

[0050] Methods of administration include, but are not limited to, parenteral, intradermal, intravitreal, intramuscular, intraperitoneal, intravenous, subcutaneous, nasal, epidural, oral, sublingual, intranasal, intracerebral, intravaginal, transdermal, transmucosal, rectal, inhalation, or topical (especially to the ear, nose, eye, or skin). The method of administration is left to the discretion of the practitioner. In most cases, administration releases the compound into the bloodstream. For the treatment of ophthalmic diseases, intravitreal administration of biological agents is preferred.

[0051] In certain embodiments, it may be desirable to administer the compound locally. This administration may be achieved, for example, but not limited to, by local infusion or application, by injection, by catheter, or by implant (which may be porous, non-porous, or a gel-like material, including silastic membranes or fibers). In such instances, administration may selectively target local tissues without substantially releasing the compound into the bloodstream.

[0052] Pulmonary administration can also be employed, e.g., by use of an inhaler or nebulizer and formulation with an aerosolizing agent, or by perfusion in a fluorocarbon or synthetic pulmonary surfactant. In certain embodiments, the compounds may be formulated as a suppository, with traditional binders and excipients such as triglycerides.

[0053] In another embodiment, the compound is delivered as a vesicle, particularly a liposome (Langer, 1990, Science 249:1527-1533; Treat et al., in Liposomes in the Therapy of Infectious Disease and Bacterial infection, Lopez-Berestein and Fidler (eds.), Liss, New York, pp. 353-365 (1989); Lopez Berestein, ibid., pp. 317-327; see generally ibid.).

[0054] In another embodiment, the compound is delivered in a controlled release system (see, e.g., Goodson, in Medical Applications of Controlled Release, supra, vol. 2, pp. 115-138 (1984)). Examples of controlled release systems discussed in the text of Langer, 1990, Science 249:1527-1533 may be used. In one embodiment, a pump may be used (Langer, supra; Sefton, 1987, CRC Crit. Ref. Biomed. Eng. 14:201; Buchwald et al., 1980, Surgery 88:507; Saudek et al., 1989, N. Engl. J. .Med. 321:574). In another embodiment, polymeric materials can be used (see, Medical Applications of Controlled Release, Langer and Wise (eds.), CRC Pres., Boca Raton, Florida (1974); Controlled Drug Bioavailability, Drug Product Design and Performance, Smolen and Ball (eds.), Wiley, New York (1984); Ranger and Peppas, 1983, J. Macromol. Sci. Rev. Macro mo 1. Chem. 23:61; see also, Levy et al., 1985, Science 228:190; During et al., 1989, Ann. Neurol. 25:351; Howard et al., 1989, J. Neurosurg. 71:105).

[0055] The above dosing schedules are provided for illustrative purposes only and should not be considered limiting, as one of ordinary skill in the art will readily appreciate that any dosing regimen is within the scope of the present invention.

[0056] It is to be understood and contemplated that modifications of the inventive principles disclosed herein may be made by those skilled in the art, and it is intended that such modifications be included within the scope of the present invention.

[0057] Throughout this application, various publications are referenced. These publications are incorporated herein by reference in their entireties in order to more fully describe the state of the art to which this invention pertains. The present invention is further described in the following examples, which should not be construed as in any way limiting the scope of the invention. [Example]

[0058] Mixed lymphocyte reaction: CD14+ mononuclear cells were isolated from whole blood by negative selection using the RosetteSep Human Mononuclear Cell Enrichment Kit (StemCell Technologies). Immature lymphocyte-derived dendritic cells (mo-DCs) were generated by culturing CD14+ cells in IMDM supplemented with 10% FBS containing 150 ng / mL GM-CSF and 50 ng / mL IL-4 for 6–7 days. CD4+ cells were negatively isolated from whole blood using the RosetteSep Human CD4 Enrichment Kit (StemCell Technologies). Next, mo-DCs and CD4+ cells from different donors were co-cultured at mo-DC to CD4 cell ratios of 1–10. To evaluate the blocking function of anti-PDL1 antibodies, increasing amounts of anti-PDL1 antibodies were added at the beginning of the co-culture. In some cases, increasing amounts of IL-15 were also added at the beginning of the co-culture. Supernatants were collected on days 6 or 7 for measurement of secreted IL-2 and IFNγ by ELISA. CD4 cell counts and expression of the proliferation marker Ki67 were assessed by flow cytometry.

[0059] Activation of PBMCs PBMCs were isolated from whole blood using Histopaque-1077 (Sigma), cultured in IMDM supplemented with 10% FBS, and activated with either SEB (0.1 μg / mL), PHA (1 μg / mL), or anti-CD3 clone HiT3a (1 μg / mL, eBioscience) for 3 to 7 days. Binding of either anti-PDL1 antibody or anti-PDL1-SD15 fusion protein was assessed in activated PBMCs by flow cytometry after 3 days. Functional assessment of anti-PDL1 antibody was performed by adding increasing amounts of anti-PDL1 antibody during PBMC activation with SEB. On days 2 or 3, Supernatants were collected for measurement of IL-2 and IFNγ. For anti-PDL1-SD15 fusion protein, PBMCs were cultured in the presence of either anti-PDL1-SD15 or anti-PDL1 antibody without further stimulation. Cells were harvested on day 6, and CD8 and granzyme B, CD8 and perforin, and CD4 cell counts were measured by flow cytometry.

[0060] Activation of CD4 and CD8 cells: CD4 and CD8 positive cells were negatively isolated from whole blood using a RosetteSep enrichment kit (StemCell Technologies). CD4 cells were activated with either anti-CD3 or anti-CD3 and PDL1 Fc-coated beads in IMDM, 10% FBS in the presence of anti-PDL1 antibody. On day 5, supernatants were collected for IFNγ measurement by ELISA, and cellular expression of the proliferation marker Ki67 was assessed using flow cytometry. CD8 cells were activated with anti-CD3-coated beads and either IL15 or anti-PDL1-SD15 fusion protein. In some cases, anti-CD3 and PDL1 Fc were used instead of anti-CD3-coated beads. Supernatants were collected on day 6 or 7 for measurement of IFNγ and TNFα secretion by ELISA. Cells were collected for measurement of CD8 activation by granzyme B and perforin markers using flow cytometry.

[0061] Antibody fusion protein nomenclature: For experiments using anti-PD-L1 IL15 fusion proteins, the shorter names for the fusion proteins are identified in the legends. Fusion protein tcclD7HC-SD15 is identified in the figure legends as cD7-SD15. Fusion protein tcclF8HC-SD15 is identified in the figure legends as F8-SD15.

[0062] Specific high-affinity antibodies against PD-L1 from a phage display library High-affinity anti-PD-L1 antibodies were obtained using a phage display library. In one procedure, phage Fabs amplified from the Dyax library were panned for three rounds on either recombinant human PDL1-Fc (PDL1 ECD and human Fc fusion protein, Q9NZQ7) or mouse PDL1-Fc (Q9EP73) immobilized in immunotubes. ELISA-positive clones from round 2 (R2) and round 3 (R3) were sequenced.

[0063] In the second step, phage Fabs amplified from the Dyax library were panned on recombinant human PDL1-Fc (PDL1 ECD and human Fc fusion protein, Q9NZQ7) in the first round, and then on activated T cells in the second round. For the third round, either activated T cells or recombinant human PDL1-Fc were used for panning. Clones capable of binding to both soluble PDL1-Fc and cells expressing PDL1-Fc were sequenced. The V of these antibodies was analyzed. H and V L The variable region sequences are shown in FIG. 13 and in columns 1-26 of Table 1.

[0064] Unique clones were converted to IgG for further characterization. The variable regions were inserted into the Dyax expression vector pBh1. Both the wild-type CH1-CH2-CH3 region and the mutant CH1-CH2-CH3 (L234A and L235A, also referred to herein as the LALA mutant) were prepared in IgG format. [Table 1-1] [Table 1-2]

[0065] These antibodies were confirmed to have specific binding to PD-L1 by solid-phase ELISA (Figures 1-4) and to HEK-293 cells (Figure 5). Blockade of the PD-1:PD-L1 interaction in the presence of these antibodies was measured by solid-phase ELISA and in HEK-293 cells expressing PD-L1. The affinity constants for each antibody were calculated using Biacore. [Table 2] [Table 3]

[0066] These antibodies also confirmed their binding to PD-L1-expressing cells in vivo, as demonstrated by binding to immature monocyte-derived dendritic cells (Fig. 6A), human PD-L1-expressing breast cancer MDA-MB-231 cells (Fig. 6B), mouse PD-L1-expressing tumor B16-F10 cells (Fig. 6C), and human activated CD4 and CD8 T cells.

[0067] Functionally active anti-PD-L1 antibodies block the PD-1 / PD-L1 interaction and increase T cell proliferation and activation. High-affinity binding anti-PD-L1 antibodies were assayed for their ability to block PD-1 / PD-L1 interactions and increase T cell proliferation. Negatively purified CD4 T cells were activated in vitro with either αCD3 or αCD3 and PD-L1 Fc-coated beads in the presence of anti-PD-L1 antibodies. CD4 cells stimulated with αCD3 and PD-L1 Fc-coated beads showed reduced proliferation and secretion of IFNγ and IL-2 compared to CD4 cells stimulated with αCD3-only coated beads. Addition of functionally active anti-PD-L1 antibodies to CD4 cultures stimulated with αCD3 and PD-L1 Fc-coated beads increased CD4 proliferation (measured by either total CD4 counts or the percentage of the proliferation marker Ki67) compared to cultures without antibody (Figure 7A). Addition of a functional activity-blocking anti-PD-L1 antibody to CD4 cultures with αCD3 and PD-L1 Fc-coated beads also increased cytokine secretion by CD4 (IFNγ and IL-2 accumulated in the supernatant as measured by ELISA).

[0068] When PBMCs isolated from whole blood were stimulated with the superantigen Staphylococcal enterotoxin B (SEB) in the presence of anti-PD-L1 blocking antibodies, increased cytokine secretion was observed. Supernatants (pre-frozen) of PBMCs cultured with SEB for 48 hours were collected, and IFNγ and IL-2 were measured by ELISA. Compared to the control without antibody, no increase in T cell numbers was observed in the cultures treated with multiple anti-PD-L1 antibodies, but significant increases in IFNγ and IL-2 levels were observed (Figure 7B).

[0069] Additionally, increased CD4 proliferation and activation was also observed in mixed lymphocyte reactions (MLRs) of CD4 T cells and mo-DCs cultured in the presence of anti-PD-L1 blocking antibodies. Multiple anti-PD-L1 antibodies increased CD4 proliferation in MLRs compared with cultures without antibody (Figure 7C). These antibodies also increased IFNγ and IL-2 secretion, as measured by ELISA.

[0070] IL15 enhances the effects of anti-PD-L1 antibodies on T cell proliferation and activation in vitro. Compared to cultures of CD4 T cells and mo-DCs containing anti-PD-L1 antibody alone, MLR between CD4 T cells and mo-DCs in the presence of both anti-PD-L1 blocking antibody and the cytokine IL15 resulted in a significant increase in CD4 proliferation (Figure 8A), IFNγ, and IL-2 secretion. In these assays, IL15 was added at equimolar concentrations with the anti-PD-L1 antibody. Some synergistic effects on CD4 proliferation were observed when anti-PD-L1 antibody and IL15 were added at low concentrations (0.5 nM, Figure 8A).

[0071] Negatively purified CD8 cells from whole blood stimulated in vitro with aCD3 and PD-L1 Fc-coated beads also responded to IL-15 in a dose-dependent manner. Addition of IL-15 to the culture medium of CD8 cells containing αCD3 and PD-L1 Fc-coated beads and anti-PD-L1 antibody significantly increased CD8 proliferation (Figure 8B).

[0072] The anti-PD-L1-IL15 fusion protein targets IL15 to PD-L1-expressing antigen-presenting cells, increasing the proliferation and activation of responding CD8 cells. The anti-PD-L1 antibody and IL15 fusion protein was constructed by combining the Fc region of the antibody with the sushi domain of IL15R and the IL15 molecule itself. The fusion of the IL15Rα sushi domain, IRD-11 exon 3, linker, and IL15 (designated "SD15") is provided as SEQ ID NO: 261. SD15 was attached to the c-terminus of the heavy chain of a conventional IgG. The fusion protein with the IL15Rα sushi domain, IRD-11 exon 3, linker, and IL15 was constructed by combining the tccλD7 variable region and IgG1 C H 1-C H 2-C HThe heavy chain c-terminus of the IgG1 variable region (SEQ ID NO: 262) was attached. This construct also contained a K to S substitution at the end of the IgG1 heavy chain to (1) reduce the possibility of "GK" cleavage and (2) add a cloning site (BamHI) to the vector.

[0073] The light chain is that of a conventional antibody. Both the light chain and the fused heavy chain with or without the LALA mutation were inserted into the Dyax pBh1 vector for expression.

[0074] This fusion molecule is designated anti-PD-L1-sushi domain-IL15 or anti-PD-L1-SD15. We also constructed a different version of the fusion protein in which IL15 was linked to Fc instead of the sushi domain; as this fusion protein did not have IL15 functional activity, we used this protein as a negative control in some assays (referred to as anti-PD-L1-SD15neg).

[0075] No significant changes were observed in the binding of the anti-PD-L1-SD15 fusion protein compared to anti-PD-L1 antibodies in a solid-phase PD-L1 Fc-binding ELISA assay (Figure 9A). Some changes in binding affinity for activated CD4 cells expressing PD-L1 were observed when the binding of the anti-PD-L1-SD15 protein was compared to each of the original anti-PD-L1 antibodies (Figure 9B). Compared to each of the anti-PD-L1 antibodies, the anti-PD-L1-SD15 protein has lower affinity for cells expressing PD-L1. However, there is a significant difference in the binding affinity between the anti-PD-L1-SD15 bound on the cell surface and the bound anti-PD-L1. There may be differences in the binding of the secondary antibody to the target antigen.

[0076] To measure the incorporation of IL15 by anti-PD-L1-SD15 fusion protein, PBMCs isolated from whole blood were cultured in the presence of either anti-PD-L1-SD15 fusion protein or IL15. No other stimuli were added to the cultures. Similar to IL15, anti-PD-L1-SD15 fusion protein increased NK cell numbers (Figure 10A), CD8 proliferation (Figure 10B), and activation (measured as the percentage of granzyme B-positive CD8, Figure 10C). No significant increase in CD4 counts was observed in any culture (Figure 10D).

[0077] To assess anti-PD-L1-SD15 activity on CD8 cells, these fusion proteins were added to CD8 cell cultures in the presence of either αCD3 or beads coated with αCD3 and PDL1 Fc. In this case, when PDL1 Fc was present on the antigen-presenting cells, αCD3, and PDL1 Fc-coated beads, anti-PD-L1-SD15 significantly increased CD8 proliferation (Figure 11A, without PDL1 Fc; Figure 11B, with PD-L1 Fc on beads). Furthermore, a significant increase in CD8 activation was also observed. cD7-SD15 reduces the effective dose required to activate CD8 cells, as measured by an approximately 10-fold increase in the percentage of granzyme B-positive CD8 cells (Figure 12A) and IFNγ secretion (Figure 12B). cD7-SD15 also increases the maximal amount of CD8 activation compared to IL15 (Figures 12A and B). Compared to the addition of anti-PD-L1 antibody and free IL15, the anti-PD-L1-SD15 fusion protein increased CD8 proliferation to a higher level than the combination of these antibodies added separately (Figure 12C). These properties of the anti-PD-L1-SD15 fusion protein may be beneficial in immunotherapy settings, as higher levels of CD8 activation and proliferation can be achieved using lower doses of the anti-PD-L1-SD15 fusion protein. When antigen-presenting cells express PD-L1, the amplified CD8 response to the anti-PD-L1-SD15 fusion protein may be advantageous for achieving selective CD8 activation.

[0078] Cytotoxicity of anti-PD-L1-IL15 fusion proteins To determine whether anti-PD-L1-SD15 fusion protein enhanced IL15-induced cytotoxicity of CD8 T cells against PD-L1-expressing tumor cells, CD8 T cells were cocultured with human PD-L1-expressing MDA-MB-231 tumor cells in the presence of anti-PD-L1-SD15 fusion protein or anti-KLH-SD15, which have no binding activity against PD-L1-expressing tumor cells, for 7 days before measuring tumor cell death. Human CD8 T cells and tumor cells were cocultured in IMDM supplemented with 10% FBS for 7 days. Tumor cell killing activity was assessed by measuring the number of killed tumor cells stained with Viability Dye eFluor 780 in a FACS assay. CD8 T cell-mediated cytotoxicity of MDA-MB-231 tumor cells was significantly enhanced by anti-PD-L1-SD15 fusion protein compared to anti-KLH-SD15 treatment during coculture (Figure 15). Furthermore, the PD-L1-SD15 fusion protein cD7-SD15 significantly increased the survival rate of mice bearing PD-L1-expressing tumor cells in a mouse tumor model in which mouse CT26 colon tumor cells were intravenously injected, compared with mice treated with vehicle or the PD-L1-SD15 fusion protein sD7-SD15, which has no binding activity to mouse PD-L1 (Figure 16). These results suggest that targeting IL15-stimulated immune effector cells to key sites overexpressing PD-L1 using a bifunctional anti-PD-L1-SD15 fusion protein has the advantage of improving anti-tumor immunity while minimizing side effects. This type of bifunctional antibody-cytokine fusion protein has potential as a novel immunomodulatory therapeutic agent, achieving greater anti-tumor efficacy in controlling tumor progression.

[0079] affinity maturation We created mutants of the tccλD7 heavy chain by introducing amino acid substitutions at three methionine positions in CDR-1H and screening for improved affinity. More specifically, we created mutants of tccλD7 (approximately 1×10) in which the first, second, and fourth methionine positions were simultaneously changed. 8A library containing 25 mutants (PDL1 ECD and human Fc fusion protein, Q9NZQ7) was generated. The library was panned (four rounds) on recombinant human PDL1-Fc (PDL1 ECD and human Fc fusion protein, Q9NZQ7) or mouse PDL1-Fc (Q9EP73) immobilized on immunotubes. ELISA-positive clones from rounds 3 and 4 were sequenced. Unique clones were compared by competitive ELISA. Table 4 shows the affinity-matured CDR-1H sequences and amino acid substitutions for the heavy chain variable region containing the CDRs observed in 25 mutants obtained from the screen, along with their SEQ ID numbers. The amino acid sequences of these mutants are also described in the sequences listed in Table 1. [Table 4]

[0080] tccD7_#114 and tccD7_#102 (referred to herein as tcc Two mutants of tccD7_#1 and tccD7_#2 were converted to IgG and IgG forms containing Leu-Ala substitutions in the hinge region for reduced ADCC, as described elsewhere herein. The antibodies were expressed and purified for further characterization. The improved binding to soluble PDL1 for the two affinity-matured mutants is shown in Figure 17. Figure 18 shows two mutants that blocked the binding of human PD1 to human PDL1 (left panel) and mouse PD1 to mouse PDL1 (right panel). Compared to the parent, the mutants exhibited higher binding activity to MDA-MB-231 cells (Figure 19).

[0081] The affinity matured variants were tested for their ability to promote production of the Th1 cytokines IL2 and IFNγ. PBMCs isolated from whole blood were stimulated with the superantigen Staphylococcal enterotoxin B (SEB, 0.1 μg / mL) in the presence of anti-PD-L1 antibodies. Supernatants from PBMCs cultured with SEB for 7 days were collected and IFNγ and IL-2 were measured by ELISA. Significantly increased amounts of IFNγ and IL-2 were observed in cultures containing variants of the anti-PD-L1 antibody cD7#1 and #2 when compared to cD7 (Figure 20).

[0082] Several fusion protein variants containing the PD-L1 binding domain, the IL15Rα sushi domain, and IL15 were constructed. Some constructs contained a linker between the IL15Rα sushi domain and the IL15 portion. In one construct, 11 amino acids in exon 3, located at the C-terminus of the IL15 receptor α sushi domain, were replaced with "GS" linkers of various lengths. GS linkers include SGGSGGGGSGGGSGGGGS (SEQ ID NO:324; 18 amino acids), SGGSGGGGGSGGGSGGGGSLQ (SEQ ID NO:314; 20 amino acids), SGGGGSGGGGSGGGGSGGGGSGGGG (SEQ ID NO:316; 25 amino acids), SGGGGSGGGGSGGGGSGGGGSGGGGSGGGG (SEQ ID NO:318; 30 amino acids), SGGGGSGGGGSGGGGSGGGGSGGGGSGGGGSGGGGSGGGG (SEQ ID NO:320; 40 amino acids), and SGGGGSGGGGSGGGGSGGGGSGGGGSGGGGSGGGGSGGGGSGGGGSGGGG (SEQ ID NO:322; 50 amino acids) in constructs having SEQ ID NOs:325, 315, 317, 319, 321, and 323, respectively.

[0083] Fusion proteins were transiently or stably expressed in HEK293 cells and purified by Protein A column chromatography according to the manufacturer's instructions. In certain experiments, the C-terminal serine of the lambda light chain was deleted (referred to herein as "ds") to stabilize the association between the Ig heavy and light chain constant regions of the anti-PD-L1 portion of the molecule.

[0084] Fusion proteins containing tccλD7 affinity matured variant #102 with the sushi domain and IL15 (SEQ ID NO: 325) were tested for binding to MDA-MB-231 by flow cytometry. All showed improved binding compared to fusion proteins containing tccλD7 (Figure 21). Fusion proteins containing tccλD7 affinity matured variant #102 were also confirmed to have stimulatory activity on IL15-responsive human megakaryoblastic leukemia cells. Cells were cultured with anti-PD-L1-SD15 fusion proteins in RPMI1640 supplemented with 10% FBS and 20% conditioned medium from human bladder carcinoma 5637 cells for 48 hours. Cell proliferation was measured as relative light units (RLU) using the CellTiter-Glo® luminescent cell viability assay (Figure 22).

[0085] Analysis by size exclusion chromatography showed less than 5% aggregation of the expressed fusion protein (Figure 24) and improved serum stability (Figure 25).

Claims

1. Array X 1 YX 2 MX 3 (SEQ ID NO: 328) (wherein, X 1 is A, G, M, Q, S, Y, or W, and X 2 is A, L, M, Q, R, S, V, W, or Y, and X 3 is A, F, L, M, S, T, V, or Y), a heavy chain CDR-1H having the sequence of SEQ ID NO: 243, and a heavy chain CDR-3H having the sequence of SEQ ID NO:

245.

2. 2. The antibody or fragment of claim 1, wherein the heavy chain CDR-1H has a sequence selected from SEQ ID NO:241, SEQ ID NO:264, SEQ ID NO:266, SEQ ID NO:268, SEQ ID NO:270, SEQ ID NO:272, SEQ ID NO:274, SEQ ID NO:276, SEQ ID NO:278, SEQ ID NO:280, SEQ ID NO:282, SEQ ID NO:284, SEQ ID NO:286, SEQ ID NO:288, SEQ ID NO:290, SEQ ID NO:292, SEQ ID NO:294, SEQ ID NO:296, SEQ ID NO:298, SEQ ID NO:300, SEQ ID NO:302, SEQ ID NO:304, SEQ ID NO:306, SEQ ID NO:308, SEQ ID NO:310, and SEQ ID NO:

312.

3. The antibody or fragment of claim 1, wherein the heavy chain CDR-1H has the sequence of SEQ ID NO:

266.

4. The antibody or fragment of claim 1, wherein the heavy chain CDR-1H has the sequence of SEQ ID NO:

290.

5. The antibody or fragment of any one of claims 1 to 4, wherein the heavy chain variable region is at least 85% identical to SEQ ID NO:

246.

6. 6. The antibody or fragment of any one of claims 1 to 5, wherein the light chain comprises a CDR-1L having SEQ ID NO: 247, a CDR-2L having SEQ ID NO: 248, and a CDR-3L having SEQ ID NO:

249.

7. The antibody or fragment of claim 6, wherein the light chain variable region is at least 85% identical to SEQ ID NO:

250.

8. An antibody or fragment thereof that binds to PD-L1, wherein the light chain comprises a CDR-1L having SEQ ID NO: 247, a CDR-2L having SEQ ID NO: 248, and a CDR-3L having SEQ ID NO:

249.

9. The antibody or fragment of claim 8, wherein the light chain variable region is at least 85% identical to SEQ ID NO:

250.

10. An antibody or fragment thereof that binds to PD-L1, comprising a heavy chain variable region having CDR-1H, CDR-2H, and CDR-3H as described in Table 1.

11. An antibody or fragment thereof that binds to PD-L1, comprising a light chain variable region having CDR-1L, CDR-2L, and CDR-3L as described in Table 1.

12. 11. The antibody or fragment of claim 10, comprising a light chain variable region having CDR-1L, CDR-2L, and CDR-3L as set forth in Table 1.

13. A heavy chain variable region sequence set forth in Table 1, or a heavy chain variable region sequence set forth in Table 1 with conservative substitutions such that the heavy chain variable region sequence set forth in Table 1 is at least 95% identical. An antibody or fragment thereof that binds to PD-L1, comprising:

14. An antibody or fragment thereof that binds to PD-L1, comprising a light chain variable region sequence described in Table 1, or a light chain variable region sequence described in Table 1 with conservative substitutions such that the antibody or fragment thereof is at least 95% identical to said light chain variable region sequence described in Table 1.

15. 14. The antibody or fragment of claim 13, comprising a light chain variable region sequence set forth in Table 1, or a light chain variable region sequence set forth in Table 1 with conservative substitutions such that the light chain variable region sequence set forth in Table 1 is at least 95% identical.

16. 16. The antibody or fragment conjugate of any one of claims 1 to 15, further comprising an imaging agent, a therapeutic agent, or a cytotoxic agent.

17. An isolated nucleic acid sequence encoding an antibody variable region or fragment according to any one of claims 1 to 15.

18. A nucleic acid vector comprising the nucleic acid of claim 17.

19. 18. A prokaryotic or eukaryotic host cell comprising the nucleic acid of claim 17.

20. A composition comprising the antibody or fragment of claim 1 and a pharmaceutically acceptable carrier.

21. A fusion protein comprising a first domain that binds to PD-L1 and a second domain that binds to the IL15 receptor.

22. 22. The fusion protein of claim 21, wherein the region that binds to PD-L1 is an antibody or a PD-L1-binding fragment thereof.

23. The fusion protein of claim 22, wherein the region that binds to PD-L1 is an antibody or fragment thereof according to any one of claims 1 to 15.

24. 22. The fusion protein of claim 21, wherein the second region binds to the IL15 receptor (IL15R).

25. 25. The fusion protein of claim 24, wherein the second region comprises IL15, or an amino acid sequence at least 95% identical to IL-15, or an IL15R-binding fragment thereof.

26. 26. The fusion protein of claim 25, further comprising an IL15 receptor alpha sushi domain.

27. 27. The fusion protein of claim 26, comprising SEQ ID NO: 325, SEQ ID NO: 315, SEQ ID NO: 317, SEQ ID NO: 319, SEQ ID NO: 321, or SEQ ID NO:

323.

28. A method for inhibiting the interaction of PD1 with PD-L1 in a subject, comprising administering an effective amount of the antibody or fragment of any one of claims 1 to 15.

29. A method for inhibiting immunosuppression mediated by PD-L1 in a subject, comprising administering an effective amount of the antibody or fragment of any one of claims 1 to 15.

30. 28. A method for treating a rheumatoid arthritis comprising administering an effective amount of a fusion protein according to any one of claims 21 to 27. A method for inhibiting PD-L1-mediated immunosuppression in a subject.

31. A method for stimulating an immune response against cells expressing PD-L1, comprising administering to a subject an effective amount of the antibody or fragment of any one of claims 1 to 15.

32. A method for stimulating an immune response against cells expressing PD-L1, comprising administering to a subject the fusion protein of any one of claims 21 to 27.

33. 33. The method of claim 31 or 32, wherein the cell expressing PD-L1 is a tumor cell.

34. The method of claim 31 or 32, wherein the cells expressing PD-L1 are infected with a virus.

Images