WO2024038095A1

WO2024038095A1 - NOVEL ANTI-RGMb ANTIBODIES

Info

Publication number: WO2024038095A1
Application number: PCT/EP2023/072581
Authority: WO
Inventors: Abdijapar SHAMSHIEV; Zaki SELLAM; Maria MEIRA BOKALOT
Original assignee: Iome Bio
Priority date: 2022-08-16
Filing date: 2023-08-16
Publication date: 2024-02-22

Abstract

The present invention relates to a human antibody or antigen-binding fragment thereof with specificity for RGMb, which inhibits binding of human RGMb to human PD-L2, and inhibits binding of human RGMb to (i) neogenin, or to (ii) BMP2, BMP4, or BMP2 and BMP4, to a multispecific antibody-based binding composition with specificity for at least RGMb and a second antigen, to a nucleic acid encoding the human antibody or antigen-binding fragment thereof of the present invention, or the multispecific antibody- based binding composition of the present invention, to a pharmaceutical composition comprising the novel human antibody or antigen-binding fragment thereof of the present invention or the multispecific antibody-based binding composition of the present invention, including for use in the prevention or treatment of cancer.

Description

NOVEL ANTI-RGMb ANTIBODIES

FIELD OF THE INVENTION

[0001 ] The present invention relates to a human antibody or antigen-binding fragment thereof with specificity for RGMb, which inhibits binding of human RGMb to human PD-L2, and inhibits binding of human RGMb to (i) BMP2, BMP4, or BMP2 and BMP4, or to (ii) neogenin, to a multispecific antibody-based binding composition with specificity for at least RGMb and a second antigen, to a nucleic acid encoding the human antibody or antigen-binding fragment thereof of the present invention, or the multispecific antibody-based binding composition of the present invention, to a pharmaceutical composition comprising the novel human antibody or antigen-binding fragment thereof of the present invention or the multispecific antibody-based binding composition of the present invention, including for use in the prevention or treatment of cancer.

BACKGROUND OF THE INVENTION

[0002] Repulsive guidance molecule b (RGMb/Dragon) is a member of the RGM family, which consists of RGMa, RGMb, and RGMc. RGMb is a co-receptor for bone morphogenetic proteins (BMPs) (Shia et al., 2021 ).

[0003] RGMb, also known as DRAGON, functions in neuronal differentiation, and binds to different ligands, bone morphogenetic proteins 2 and 4 (BMP), neogenin, and PD-L2 (Pauken et al., 2021 ).

[0004] BMPs belong to the TGF[3 superfamily and regulate cell differentiation, proliferation, patterning and migration. BMPs are secreted cytokines and RGMb can function as a co-receptor for BMP engagement with Type I and Type II BMP serine/threonine kinase receptors which initiates Smad activation followed by Smad translocation into the nucleus resulting in context-dependent transcription. Separately, a dimer of RGMbs can bring together two neogenin molecules. Following engagement with BMP, RGMb inhibits IL-6 secretion in macrophages. PD-L2 and BMP can simultaneously bind to RGMb at distinct binding sites to form a trimeric complex.

[0005] Moreover, PD-1 and RGMb have distinct binding sites on PD-L2. Since RGMb binds to neogenin in cis, this may result in a supercomplex of BMP-BMPR-RGMb- neogenin. PD-L2 may bind in trans with the RGMb supercomplex to regulate downstream pathways. Neogenin belongs to the immunoglobulin superfamily, and in addition to RGMs also binds with netrin. The neogenin-netrin axis regulates angiogenesis and cell adhesion, while the neogenin-RGM interaction regulates neural cell differentiation and neural cell proliferation.

[0006] RGMb is an attractive drug target, since it has been demonstrated to be involved in tolerance and immunosuppression in the context of respiratory and neoplastic disorders (Pauken et al., 2021 ).

[0007] Monoclonal antibody based biotherapeutics have become an important avenue for new drug development. Monoclonal antibody technology offers specific targeting, precise signaling and/or payload delivery to specific cell population, and provides long lasting biological effect through its Fc functions. Efforts in antibody engineering have allowed developing humanized antibodies with improved pharmacological properties (potency, selectivity, stability, immunogenicity, etc.), thus expanding the scope of antibody drug development.

[0008] A number of anti-RGMb antibodies have already been developed in the past, including antibody 9D1 (Xiao et al., J Exp Med (2014) 211 ). However, a need exists for therapeutics targeting RGMb with a different mode of action and/or improved properties, including but not limited to high affinity binding to RGMb, cross-reactivity between human and cynomolgus monkey RGMb polypeptides, binding to immune and cancer cells, and the ability to induce T cell-mediated anti-tumor activities. In addition, there is a great need in the art for treating disorders that would benefit from using specific combinations of anti RGMb with immune checkpoints and other therapeutic agents.

[0009] Thus, in summary, there remains a clear need for novel antibody-based compositions with specificity for RGMb, in particular novel antibodies which block interaction of RGMb with at least two of its ligands.

SUMMARY OF THE INVENTION

[0010] It is an object of the present invention to provide a novel antibody-based binding composition with specificity for RGMb. More specifically, it was an object of the present invention to provide novel antibodies with specificity for RGMb which additionally block at least two of the ligands of RGMb.

[0011 ] The inventors have now surprisingly found that such novel anti-RGMb antibodies with properties superior to those of similar antibodies of the prior art could be obtained by extensive selection based on a phage-display antibody library.

[0012] Accordingly, in a first aspect, the present invention relates to a human antibody or antigen-binding fragment thereof that is specific for human RGMb, inhibits binding of human RGMb to human PD-L2, and inhibits binding of human RGMb to (i) BMP2, BMP4, or BMP2 and BMP4, or to (ii) neogenin.

[0013] In a second aspect, the present invention relates to a multispecific antibodybased binding composition with specificity for at least RGMb and a second antigen, wherein said multispecific antibody-based binding composition comprises a first human antibody or antigen-binding fragment thereof, which is a human antibody or antigen-binding fragment thereof of the present invention, and at least a second human antibody or antigen-binding fragment thereof with specificity for a different antigen.

[0014] In a third aspect, the present invention relates to a nucleic acid encoding the human antibody or antigen-binding fragment thereof of the present invention, or the multispecific antibody-based binding composition of the present invention.

[0015] In a fourth aspect, the present invention relates to a vector comprising the nucleic acid of the present invention.

[0016] In a fifth aspect, the present invention relates to a host cell comprising the nucleic acid of the present invention or the vector of the present invention.

[0017] In a sixth aspect, the present invention relates to a method for producing the human antibody or antigen-binding fragment thereof of the present invention, or the multispecific antibody-based binding composition of the present invention, comprising a step selected from (a) expressing the nucleic acid of the present invention in an expression system; (b) expressing a nucleic acid from the vector of the present invention; and (c) culturing the host cell of the present invention.

[0018] In a seventh aspect, the present invention relates to a multifunctional antibodybased binding composition, wherein said multifunctional antibody-based binding composition comprises a human antibody or antigen-binding fragment thereof, which is a human antibody or antigen-binding fragment thereof of the present invention, and at least a second non-antibody-based agent that selectively inhibits or blocks the expression or activity of PD-1 or PD-L1 .

[0019] In an eighth aspect, the present invention relates to a pharmaceutical composition comprising the human antibody or antigen-binding fragment thereof of the present invention, the multispecific antibody-based binding composition of the present invention, or the multifunctional antibody-based binding composition of the present invention and a pharmaceutically acceptable carrier.

[0020] In a nineth aspect, the present invention relates to the human antibody or antigen-binding fragment thereof of the present invention, the multispecific antibodybased binding composition of the present invention, the multifunctional antibodybased binding composition of the present invention, or the pharmaceutical composition of the present invention, for use in the prevention or treatment of cancer.

[0021 ] In a tenth aspect, the present invention relates to a method of treating an individual having a condition, where RGMb is expressed and that would benefit from upregulation of an immune response, in particular cancer, comprising the step of administering to said individual a therapeutically effective amount of the human antibody or antigen-binding fragment thereof of the present invention, the multispecific antibody-based binding composition of the present invention, the multifunctional antibody-based binding composition of the present invention, or the pharmaceutical composition of the present invention.

BRIEF DESCRIPTION OF THE DRAWINGS

[0022] FIGURE 1 shows the workflow of antibody discovery and selection: RGMb- positive clones were identified in panning rounds on huRGMb and muRGMb or on huRGMb alone; in total, 460 clones were found which were screened against huRGMb as soluble scFvs; out of them only 198 clones bound to huRGMb, and the VH and VL domains were sequenced identifying 80 mAbs with diverse CDRs; those hits were expressed as soluble scFvs and screened on huRGMb and muRGMb resulting in 58 hits, which were additionally screened for binding to cyno-RGMb resulting in 53 scFvs binding to all three targets. Those hits in scFv format were tested by flow cytometry on HEK293-RGMb cells and expressed as human lgG1 antibodies thereby reducing the number of hits to 51 ; for those hits, specificity and cross-reactivity were verified by ELISA and flow cytometry on HEK293-RGMb cells and screened in RGMb/PD-L2 inhibition assays; the VH and VL sequences of the 21 clones that performed best in binding (KD and ECso) and in inhibiting RGMb was analyzed resulting in the identification of the seven top-performing antibodies with diverse CDRs.

[0023] FIGURE 2 shows that anti-RGMb antibodies bind monomeric and dimeric forms of RGMb. Exemplary anti-RGMb scFvs (FIGURE 2A, FIGURE 2B) and full chain human lgG1 antibodies (FIGURE 2C) were added to microplate-immobilized huRGMb-His or huRGMb-rabbit-Fc. The bound scFvs and hulgG1 antibodies were detected by using anti-c-Myc-HRP antibody and anti-hulgG1 -HRP, respectively.

[0024] FIGURE 3 shows that the anti-RGMb antibodies recognize huRGMb expressed on the cell surface. HEK-RGMb.1 cells were stained with 3C09, 3C08, 5B05, 3E07, 5C10, 2C11 and 2G03 antibodies (bold solid lines), followed by staining with anti-huIgG-FITC. As positive and negative controls, a 9D1 antibody (upper left panel, bold solid line) and an isotype control antibody (shaded histograms) were used, respectively.

[0025] FIGURE 4 shows that the anti-RGMb antibodies recognize the natural form of huRGMb expressed on MDA-MB-231 cells. MDA-MB-231 cells were stained with 3C09, 5B05, 5C10, 2C11 and 2G03 antibodies (bold solid lines), followed by staining with anti-huIgG-FITC. As positive and negative controls, a 9D1 antibody (indicated panel, bold solid line) and an isotype control antibody (shaded histograms) were used, respectively.

[0026] FIGURE 5 shows a schematic outline of the three types of inhibition assays for assessing human RGMb-blocking antibodies.

[0027] FIGURE 6 shows that the anti-RGMb antibodies block the huRGMb/huPD-L2 interaction. Anti-RGMb (3C09, 3C08, 5B05, 3E07, 5C10, 2C11 and 2G03) IgG antibodies (30 ng/ml to 10 pg/ml) were added into wells with immobilized huRGMb- Fc. After incubation and washing, 600 ng/ml huPD-L2-Fc-biotin was added. The clone 9D1 and MAB3630 were used as positive controls, while isotype control Ig antibodies were used as negative control antibodies. The bound PD-L2-Fc-biotin was detected by streptavidin-poly-HRP. The interaction of huRGMb with huPD-L2 is weaker than the interaction of muRGMb with muPD-L2, as previously shown (Xiao et al., J Exp Med (2014) 211 ). Therefore, it was not possible to determine the ICso values for RGMb inhibitors in this experimental setup, although the inhibition is clearly demonstrated. [0028] FIGURE 7 shows blocking of RGMb by anti-RGMb antibodies in a huRGMb/muPD-L2 assay (hulgG1 ). Increasing amounts of anti-RGMb (3C09, 3C08, 5B05, 3E07, 5C10, 2C11 and 2G03) IgG antibodies were pre-incubated with 200 ng/ml huRGMb-Fc-biotin and added to immobilized muPD-L2. The clone 9D1 and MAB3630 were used as positive controls, while isotype control Ig antibodies were used as negative control antibodies. The bound huRGMb-Fc-biotin was detected by streptavidin-HRP. The lower graph shows that the blocking potency of anti-RGMb 2C11 is 10.5-fold superior to the reference antibody 9D1 (0.33 nM vs 3.49 nM respectively). ICso values are depicted in the table.

[0029] FIGURE 8 shows blocking of RGMb by anti-RGMb antibodies in a muRGMb/muPD-L2 assay. Anti-RGMb (3C09, 3C08, 5B05, 3E07, 5C10, 2C11 and 2G03) IgG antibodies at 10 pg/ml were added to immobilized muRGMb. The clone 9D1 and MAB3630 were used as positive controls, while isotype control Ig antibodies were used as negative control antibodies. The bound muPD-L2-biotin was detected by streptavidin-HRP.

[0030] FIGURE 9 shows the inhibition of the RGMb/Neogenin 1 interaction by anti- RGMb antibodies. Increasing amounts of anti-RGMb (3C09, 3C08, 5B05, 3E07, 5C10, 2C11 and 2G03) IgG antibodies were pre-incubated with 75 ng/ml huRGMb-rabbit-Fc and added to immobilized human Neogenin-His (A) or human Neogenin-Fc fusion (B) proteins. The bound huRGMb-rFc was detected by using anti-rabbit-Fc-HRP. 9D1 and MAB3630 were used as positive controls, while an isotype control IgG was used as a negative control. The results demonstrated that the anti-RGMb antibodies 3C08, 2G03 and 5C10 completely block huRGMb/Neogenin interaction.

[0031 ] FIGURE 10 shows the inhibition of RGMb/BMP4 and RGMb/BMP2 interactions by anti-RGMb antibodies. FIGURE 10A: RGMb-His was immobilized, and increasing amounts of indicated anti-RGMb (3C09, 3C08, 5B05, 3E07, 5C10, 2C11 and 2G03) IgG antibodies were added and followed by the addition of BPM4. The bound BMP4 were measured using a biotinylated anti-BMP4 antibody and detected by SAV-HRP. The lower graph shows that the blocking potency of RGMb/BMP4 interaction by anti-RGMB 2C11 is 3.2-fold superior to the reference antibody 9D1 (6.5 nM vs 21.0 nM, respectively). ICso values are depicted in the table. FIGURE 10B: HuRGMb-His was immobilized and incubated in the presence of 10 pg/ml anti-RGMb (3C09, 3C08, 5B05, 3E07, 5C10, 2C11 and 2G03) IgG antibodies or control mAbs, followed by addition of huBPM2. The bound huBMP2 was detected using a biotinylated anti-BMP2 antibody and SAV-HRP. 9D1 and MABAF3630 were used as positive controls. The results show that the anti-RGMb antibodies 5B05, 3C09 and 2C11 potently block the RGMb/BMP4 and RGMb/BMP2 interactions.

[0032] FIGURE 11 shows the inhibition of 9D1 antibody binding to RGMb by anti- RGMb antibodies. Increasing amounts of anti-RGMb (3C09, 3C08, 5B05, 3E07, 5C10, 2C11 and 2G03) IgG antibodies were pre-incubated with immobilized huRGMb-His, followed by the addition of biotinylated 9D1 . The bound 9D1 -biotin was detected using streptavidin-HRP. Unlabeled 9D1 was used as a positive control, while an isotype control hulgG1 was used as a negative control. The results demonstrated that the anti- RGMb antibodies 2C11 and 5B05 compete with 9D1 , whereas the 3E07 antibody partially competes with 9D1 .

[0033] FIGURE 12 shows a schematic illustration of anticipated epitopes of anti-RGMb antibodies (2C11 , 5B05, 5C10, 3C08, and 2G03) and 9D1 antibodies in relation to BPM2-, BMP4- and neogenin- and PD-L2-binding regions of huRGMb. 2C11 and 5B05 antibodies share a common epitope 1 (upper dotted oval), which overlaps with the BPM2- and BMP4-binding regions. 5C10, 3C08 and 2G03 antibodies share epitope 2 (lower dotted oval), which coincides with the RGMb/Neogenin binding region. 3C09 and 3E07 antibodies not indicated here partially inhibit RGMb/BMP2/4 and RGMb/Neogenin interactions. The latter antibodies likely bind an epitope shared with RGMb/PD-L2-binding region (dotted rectangle); however, it does not overlap with RGMb/neogenin- and RGMb/BMP2/4-binding regions. PD-L2-binding site is shown as a dotted rectangle, which covers both epitopes 1 and 2.

[0034] FIGURE 13 shows that the 2C11 anti-RGMb antibody is stable in human plasma and PBS over a period of 24 h. 2C11 antibody was incubated in human plasma or PBS at 37°C at a final concentration of 5 ug/ml for up to 24 h. Samples were collected at indicated time points (0, 2, 6, 8 and 24 h) and incubated with coated human RGMb-rFc protein. The bound 2C11 antibodies were detected by an anti-kappa-HRP antibody. The results demonstrated that the binding capacity of 2C11 anti-RGMb antibody was preserved in human plasma (left panel) and PBS (right panel).

[0035] FIGURE 14 shows the cross-reactivity of anti-RGMb antibodies to human, mouse and cynomolgus RGMb. Increasing amounts of anti-RGMb IgG antibodies (3C09, 3C08, 5B05, 3E07, 5C10, 2C11 , 2G03) and an isotype control antibody were incubated with coated human- mouse- and cyRGMb proteins. The bound anti-RGMb antibodies were detected by an anti-hulgG1 -HRP, while MAB3630 was detected by anti-mouse-HRP antibodies.

[0036] FIGURE 15 shows that anti-RGMb antibodies recognize huRGMb but not huRGMa and huRGMc. Antibodies 3C09, 3C08, 5B05, 3E07, 5C10, 2C11 and 2G03 at 10 pg/ml were incubated with immobilized RGMa, RGMb and RGMc proteins. An anti-hulgG1 -HRP antibody was used to detect the bound antibodies.

[0037] FIGURE 16 shows (A) the in vivo determination of maximum tolerated dose (MTD) and (B) pharmacokinetic properties of the 2C11 antibody. (A) For MTD testing, female C57BL/6 mice were administered intraperitoneally either 100 pg (5 mg/kg), 200 pg (10 mg/kg) or 400 pg (20 mg/kg) of 2C11 antibody at days 1 , 4, 7 and 10. Body weights and general signs were measured and recorded daily over the 14-day study period. The results demonstrated that no major toxicity was observed and that the maximum tolerated dose (MTD) was not reached at 400 pg. (B) For PK parameters determination, intravenous lateral tail vein single dosing of male CD-1 mice with 10 mg/kg of 2C11 or the control antibody 9D1 was performed. Animals divided into three subgroups (n = 3) were randomly assigned to the following time points: 5 min, 8 h, and 5 d for the first group of mice; 1 h, 24 h, and 7 d for the second group; 4 h and 72 h for the third group, for a maximum of three blood collections per mouse. Blood samples were processed and the concentrations of test substances in each plasma sample were assessed using a human lgG1 ELISA kit. PK parameters were calculated (right panel). The results demonstrate a much higher exposure of the 2C11 antibody compared to the control 9D1 antibody.

[0038] FIGURE 17 shows (A) the in vivo efficacy and (B) survival in the subcutaneous B16F10-OVA mouse melanoma syngeneic model in female C57BL/6 mice treated with an anti-PD-L1 antibody or a combination of anti-PD-L1 and 2C11 antibody. An isotype control was used as a negative control. (A) The mice were inoculated subcutaneously with B16F10-OVA tumor cells. The animals were randomized on tumor size and treated 4 times with either isotype control, anti-PD-L1 or a combination of anti-PD-L1 and 2C11 antibodies with a 3 days-interval. Body weights and tumor sizes were measured 3 times weekly. Tumor growth inhibition at day 10 is depicted. For survival analysis (B), animals were monitored within the window of individual tumor burden not exceeding the size of 2,000 mm³ or a diameter exceeding 2 cm in size in any direction. The results demonstrate that 2C11 in combination with aPD-L1 strongly increases tumor growth inhibition in comparison to control groups. Also, survival is significantly increased in this group.

[0039] FIGURE 18 shows the characterization of bispecific antibodies. (A) Representative scheme of the bivalent (left panel) and tetravalent (right panel) formats of the bispecific constructs. (B) Representative blockade assays of the RGMb/PD-L2 (upper panel) and PD-1/PD-L1 (lower panel) interactions by the 2C11/EH12 bivalent and tetravalent bispecific antibodies. For the RGMb/PD-L2 blocking assay, increasing amounts of bispecific 2C11/EH12 bivalent and tetravalent antibodies were preincubated with huRGMb-rabbit-Fc-biotin and added to the immobilized muPD-L2-huFc protein. The bound huRGMb-rFc-biotin was detected by using streptavidin-poly-HRP. For the PD-1/PD-L1 blocking assay, increasing amounts of bispecific 2C11/EH12 bivalent and tetravalent antibodies were added to the immobilized huPD-1 -huFc protein. HuPD-L1 -huFc-Biotin was added, and the bound protein was detected by using streptavidin-HRP. The results demonstrated that both bivalent and tetravalent formats of the 2C11/EH12 antibody completely block both RGMb/PD-L2 and PD-1/PD- L1 interactions.

DETAILED DESCRIPTION OF THE INVENTION

[0040] While first antibodies with specificity for RGMb have been described in the prior art, the data published for these antibodies make clear that there is still substantial room for additional improvement.

[0041 ] The present invention provides novel antibody-based binding composition with specificity for RGMb that exhibit superior properties, in particular related to their ability to compete with other ligands of RGMb. Those compositions of the present invention and pharmaceutical compositions comprising them thus provide distinct advantages over anti-RGMb antibodies that have been established so far.

[0042] Accordingly, in a first aspect, the present invention relates to a human antibody or antigen-binding fragment thereof that is specific for human RGMb, inhibits binding of human RGMb to human PD-L2, and inhibits binding of human RGMb to (i) BMP2, BMP4, or BMP2 and BMP4, or to (ii) neogenin. [0043] In a particular embodiment of the present invention, said human antibody or antigen-binding fragment thereof is characterized by one or more of the following properties selected from the following list: i) binds to human RGMb with a KD of between 10 nM and 100 pM, when expressed as full chain human lgG1 and tested in Bio-Layer Interferometry captured on an anti-human Fab biosensor and incubated with different concentrations of analyte huRGMb-rabbit Fc; ii) binds specifically to murine RGMb; iii) binds specifically to cynomolgus RGMb; iv) inhibits binding of human RGMb to murine PD-L2, particularly with an ICso value of between 2 nM and 50 pM, when expressed as full chain human IgG 1 and tested against dimeric human RGMb protein expressed as a fusion of the extracellular domain comprising amino acids 46-413 with rabbit Fc that has been immobilized onto Maxisorp 96-well microplates at 2 pg/ml by using increasing amounts of said human antibodies from 10 ng/ml to 10 pg/ml, followed by a constant amount of 600 ng/ml huPD-L2-Fc-biotin and detection of bound huPD-L2-Fc-biotin by Streptavidin Poly-HRP; v) does not bind to RGMa or RGMc; and vi) exhibits a monomeric content of 95 % of higher, when expressed as fullchain mouse lgG2a antibodies with Leu234Ala/Leu235Ala (LALA) mutations and formulated in PBS pH 7.4 at 2 mg/ml and exposed to three freeze/thaw cycles in liquid nitrogen and thawing at RT.

[0044] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this invention pertains.

[0045] The terms “comprising” and “including” are used herein in their open-ended and non-limiting sense unless otherwise noted. With respect to such latter embodiments, the term "comprising" thus includes the narrower term “consisting of”.

[0046] The terms “a” and “an” and “the” and similar references in the context of describing the invention (especially in the context of the following claims) are to be construed to cover both the singular and the plural, unless otherwise indicated herein or clearly contradicted by context. For example, the term “a cell” includes a plurality of cells, including mixtures thereof. Where the plural form is used for compounds, salts, and the like, this is taken to also mean a single compound, salt, or the like.

[0047] The term “antibody”, as used herein, refers naturally occurring glycoproteins that are central elements of the immune response to foreign substances. An antibody comprises at least two heavy (H) chains and two light (L) chains inter-connected by disulfide bonds. Each heavy chain is comprised of a heavy chain variable domain (abbreviated herein as VH) and a heavy chain constant region. The heavy chain constant region is comprised of three domains, CH1 , CH2 and CH3. Each light chain is comprised of a light chain variable domain (abbreviated herein as VL) and a light chain constant region. The light chain constant region is comprised of one domain (abbreviated herein as CL). The VH and VL regions can be further subdivided into regions of hypervariability, termed complementarity determining regions (CDRs), interspersed with regions that are more conserved, termed framework regions (FRs). Each VH and VL is composed of three CDRs and four FRs arranged from aminoterminus to carboxy-terminus in the following order: FR1 , CDR1 , FR2, CDR2, FR3, CDR3, FR4. The variable domains of the heavy and light chains form a binding domain that specifically interacts with an antigen. The constant regions of the antibodies may mediate the binding of the immunoglobulin to host tissues or factors, including various cells of the immune system (e. g., effector cells) and the first component (Clq) of the classical complement system.

[0048] The term “antibody or antigen-binding fragment thereof” as used herein refers to any proteinaceous composition comprising at least a part of an antibody that is able to specifically bind to the antigen of that antibody, in the present case to RGMb. Such part of an antibody may be referred to herein as “binding domain”, “antigen-binding fragment”, or “antigen-binding portion” of an antibody, and the like, as used herein. Thus, the term “antibody or antigen-binding fragment thereof” as used herein refers to compositions consisting of one antibody or compositions consisting of one or more fragments of an intact antibody that retain the ability to specifically bind to the antigen of that antibody, in the present case to RGMb.

[0049] The term “binding specificity” as used herein refers to the ability of an individual antibody to react with one antigenic determinant and not with a different antigenic determinant. As used herein, the term “specifically binds to” or is “specific for” refers to measurable and reproducible interactions such as binding between a target and an antibody, which is determinative of the presence of the target in the presence of a heterogeneous population of molecules including biological molecules. For example, an antibody that specifically binds to a target (which can be an epitope) is an antibody that binds this target with greater affinity, avidity, more readily, and/or with greater duration than it binds to other targets. In its most general form (and when no defined reference is mentioned), “specific binding” refers to the ability of the antibody to discriminate between the target of interest and an unrelated molecule, as determined, for example, in accordance with specificity assay methods known in the art. Such methods comprise, but are not limited to Western blots, ELISA, RIA, ECL, IRMA, SPR (Surface plasmon resonance) tests and peptide scans. For example, a standard ELISA assay can be carried out. The scoring may be carried out by standard color development (e. g. secondary antibody with horseradish peroxide and tetramethyl benzidine with hydrogen peroxide). The reaction in certain wells is scored by the optical density, for example, at 450 nm. Typical background (= negative reaction) may be about 0.1 OD; the typical positive reaction may be about 1 OD. This means the ratio between a positive and a negative score can be 10-fold or higher. In a further example, an SPR assay can be carried out, wherein at least 10-fold, particularly at least 100-fold difference between a background and signal indicates on specific binding. Typically, determination of binding specificity is performed by using not a single reference molecule, but a set of about three to five unrelated molecules, such as milk powder, transferrin or the like. In particular embodiments, the term “specificity” refers to the ability of an antibody to differentiate between the target of interest, in the present case RGMb, and closely related antigens, such as in the present case other proteins of the RGM family, such as RGMa or RGMc.

[0050] The term “Complementarity Determining Regions” (“CDRs”) refers to amino acid sequences with boundaries determined using any of a number of well-known schemes, including those described by Kabat et al. (1991 ), “Sequences of Proteins of Immunological Interest,” 5^th Ed. Public Health Service, National Institutes of Health, Bethesda, MD (“Kabat” numbering scheme); Al-Lazikani et al., (1997) JMB 273, 927- 948 (“Chothia” numbering scheme); ImMunoGenTics (IMGT) numbering (Lefranc, M.- P., The Immunologist, 7, 132-136 (1999); Lefranc, M.-P. et al., Dev. Comp. Immunol., 27, 55-77 (2003)) (“IMGT” numbering scheme); and the numbering scheme described in Honegger & Pluckthun, J. Mol. Biol. 309 (2001 ) 657-670 (“AHo” numbering). For example, for classic formats, under Kabat, the CDR amino acid residues in the heavy chain variable domain (VH) are numbered 31 -35 (HCDR1 ), 50-65 (HCDR2), and 95- 102 (HCDR3); and the CDR amino acid residues in the light chain variable domain (VL) are numbered 24-34 (LCDR1 ), 50-56 (LCDR2), and 89-97 (LCDR3). Under Chothia, the CDR amino acids in the VH are numbered 26-32 (HCDR1 ), 52-56 (HCDR2), and 95-102 (HCDR3); and the amino acid residues in VL are numbered 24- 34 (LCDR1 ), 50-56 (LCDR2), and 89-97 (LCDR3). By combining the CDR definitions of both Kabat and Chothia, the CDRs consist of amino acid residues 26-35 (HCDR1 ), 50-65 (HCDR2), and 95-102 (HCDR3) in human VH and amino acid residues 24-34 (LCDR1 ), 50-56 (LCDR2), and 89-97 (LCDR3) in human VL. Under IMGT the CDR amino acid residues in the VH are numbered approximately 26-35 (HCDR1 ), 51 -57 (HCDR2) and 93-102 (HCDR3), and the CDR amino acid residues in the VL are numbered approximately 27-32 (LCDR1 ), 50-52 (LCDR2), and 89-97 (LCDR3) (numbering according to “Kabat”). Under IMGT, the CDRs of an antibody can be determined using the program IMGT/DomainGap Align.

[0051 ] In the context of the present invention, and in contrast to the numbering systems referred to above that had been used by Kabat, Chothia etc., the numbering system suggested by Honegger & Pluckthun (“AHo”) is used (Honegger & Pluckthun, J. Mol. Biol. 309 (2001 ) 657-670), unless specifically mentioned otherwise. In particular, in the context of the present invention, the following residues are defined as CDRs according to AHo numbering scheme: LCDR1 (also referred to as CDR-L1 ): L24-L42; LCDR2 (also referred to as CDR-L2): L58-L72; LCDR3 (also referred to as CDR-L3): L107-L138; HCDR1 (also referred to as CDR-H1 ): H33-H42; HCDR2 (also referred to as CDR-H2): H57-H69; HCDR3 (also referred to as CDR-H3): H109-H138. Table 1 shows the sequences of the VH and VL domains of the antibodies of the present invention, where the CDR regions are indicated by bold and underlined letters. For the sake of clarity, the numbering system according to Honegger & Pluckthun takes the length diversity into account that is found in naturally occurring antibodies, both in the different VH and VL subfamilies and, in particular, in the CDRs, and provides for gaps in the sequences. Thus, in a given antibody variable domain usually not all positions 1 to 149 will be occupied by an amino acid residue.

[0052] As stated above, each of the variable domains VH and VL of antibodies comprises CDR regions CDR1 to CDR3, and framework regions FR1 to FR4. The CDRs are loop regions that together form the antigen-binding site, while the framework regions provide the globular structural backbone and additionally support the CDR loop regions, including certain residues that are critical for antigen-binding.

[0053] Based on the sequence homology of the framework regions, it is possible to group the variable domains into distinct variable domain families.

[0054] In the context of the present invention, the term “belonging to the VHx family” (or the VLx family) means that the framework sequences FR1 to FR3 show the highest degree of homology to said VHx family (or VLx, respectively). Examples of VH and VL families are given in WO 97/08320, Knappik et al., J. Mol. Biol. 296 (2000) 57-86, or in WO 2019/057787. In the context of the present invention, the set of VH and VL families as listed in WO 97/08320 are used as reference sequences. The VH domains of each of the seven antibodies of the present invention that are specifically listed herein belong to the VH3 family, and the VL domains of each of the seven antibodies of the present invention that are specifically listed herein belong to the Vlambda2 family. The consensus sequences for VH3 and Vlambda2 according to WO 97/08320 are shown in Table 1 (SEQ ID NOs:16 and 17, respectively).

[0055] While the grouping into families is determined based on the homology of, and while the residues in the framework regions that are critical for binding are comprised in, framework regions 1 to 3, the sequence of FR4 is less relevant, and FR4 sequences can be selected or exchanged more freely. This is not particularly surprising since antibody genes are generated in vivo by recombination of germline V, D, and J segments (VH domains) and V and J segments (VL domains), wherein the V segments encode frameworks 1 to 3, CDR1 , CDR2 and part of CDR3, whereas FR4 is encoded by the J segments. Thus, while the variable domains of the antibody or antigen-binding fragment thereof of the present invention are defined by reference to positions 3 to 138 only, /. e. a sequence stretch not including FR4, it is submitted that the antibody or antigen-binding fragment thereof of the present invention additionally comprise a FR4 region in each of VH and VL. In particular embodiments, that FR4 region is selected from human germ line J segments, which have all been sequenced and are known to anyone of ordinary skill in the art. In particular embodiments, that additional FR4 region is the FR4 region comprised in the corresponding sequence shown in Table 1. In that context it should be mentioned that it has even been shown that is possible to create hybrid variable domains comprising Vkappa-based frameworks FR1 to FR3 with a Vlambda-based FR4 sequences (see WO 2019/057787). Alternatively, the VH orVL domains of the present invention may comprise an alternative FR4 region that is found in a rearranged antibody sequence belonging to the VH3 or Vlambda2 family, respectively, or a variant of the FR4 as present in Table 1 , where such variant comprises one or two, particularly just one amino acid residue substitution.

[0056] Similarly, amino acid residues at the N-terminus of the variable domains are less critical for the structural integrity of the variable domains and/or the binding properties. In that context, it should be mentioned that it has been shown by Knappik et al. (Joe. cit.) that it is possible to generate and use a Vkappa3 consensus gene that starts with “DI...” as in the other Vkappa families instead of “El...” as present in the repertoire of germ line V segments belonging to the Vkappa3 family. Thus, in particular embodiments, the VH or VL domains of the present invention may comprise in positions 1 to 4 (AHo numbering) an alternative N-terminal end that is found in a rearranged antibody sequence belonging to the same VH3 or Vlambda2 family, respectively, or a variant of the N-terminal end as present in Table 1 , where such variant comprises one or two, particularly just one amino acid residue substitution.

[0057] Thus, while the variable domains of the antibody or antigen-binding fragment thereof of the present invention are listed in Table 1 by their full VH and VL sequences comprising the positions 1 to 149 according to the AHo numbering system, the core binding domains comprise at least positions 5 to 138 (AHo numbering), particularly at least positions 3 to 145 (AHo numbering).

[0058] In particular embodiments of the present invention, said human antibody or antigen-binding fragment thereof comprises a combination of a VH domain belonging to the VH3 family and a VL domain belonging to the Vlambda2 family, wherein

(i) the VH domain comprises a set of CDR1 , CDR2 and CDR3 sequences as comprised in the VH domain sequence according to SEQ ID NO:7 shown in Table 1 , and the VL domain comprises a set of CDR1 , CDR2 and CDR3 sequences as comprised in the VL domain sequence according to SEQ ID NO:8 shown in Table 1 ;

(ii) the VH domain comprises a set of CDR1 , CDR2 and CDR3 sequences as comprised in the VH domain sequence according to SEQ ID NO:9 shown in Table 1 , and the VL domain comprises a set of CDR1 , CDR2 and CDR3 sequences as comprised in the VL domain sequence according to SEQ ID NO: 10 shown in Table 1 ;

(iii) the VH domain comprises a set of CDR1 , CDR2 and CDR3 sequences as comprised in the VH domain sequence according to SEQ ID NO: 13 shown in Table 1 , and the VL domain comprises a set of CDR1 , CDR2 and CDR3 sequences as comprised in the VL domain sequence according to SEQ ID NO: 14 shown in Table 1 ;

(iv) the VH domain comprises a set of CDR1 , CDR2 and CDR3 sequences as comprised in the VH domain sequence according to SEQ ID NO:1 shown in Table 1 , and the VL domain comprises a set of CDR1 , CDR2 and CDR3 sequences as comprised in the VL domain sequence according to SEQ ID NO:2 shown in Table 1 ;

(v) the VH domain comprises a set of CDR1 , CDR2 and CDR3 sequences as comprised in the VH domain sequence according to SEQ ID NO:3 shown in Table 1 , and the VL domain comprises a set of CDR1 , CDR2 and CDR3 sequences as comprised in the VL domain sequence according to SEQ ID NO:4 shown in Table 1 ;

(vi) the VH domain comprises a set of CDR1 , CDR2 and CDR3 sequences as comprised in the VH domain sequence according to SEQ ID NO:5 shown in Table 1 , and the VL domain comprises a set of CDR1 , CDR2 and CDR3 sequences as comprised in the VL domain sequence according to SEQ ID NO:6 shown in Table 1 ;

(vii) the VH domain comprises a set of CDR1 , CDR2 and CDR3 sequences as comprised in the VH domain sequence according to SEQ ID NO:11 shown in Table 1 , and the VL domain comprises a set of CDR1 , CDR2 and CDR3 sequences as comprised in the VL domain sequence according to SEQ ID NO: 12 shown in Table 1 .

[0059] In particular embodiments of the present invention, said human antibody or antigen-binding fragment thereof comprises a combination of a VH domain belonging to the VH3 family and a VL domain belonging to the Vlambda2 family, wherein

(i) the VH domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO:7, and the VL domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO:8;

(ii) the VH domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO:9, and the VL domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NQ:10;

(iii) the VH domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO: 13, and the VL domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO:14;

(iv) the VH domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO:1 , and the VL domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO:2;

(v) the VH domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO:3, and the VL domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO:4;

(vi) the VH domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO:5, and the VL domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO:6;

(vii) the VH domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO: 11 , and the VL domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO:12. [0060] In particular embodiments of the present invention, said human antibody or antigen-binding fragment thereof comprises a combination of

(i) the VH domain having a sequence according to SEQ ID NO:7, and the VL domain having a sequence according to SEQ ID NO:8;

(ii) the VH domain having a sequence according to SEQ ID NO:9, and the VL domain having a sequence according to SEQ ID NO: 10;

(iii) the VH domain having a sequence according to SEQ ID NO: 13, and the VL domain having a sequence according to SEQ ID NO: 14;

(iv) the VH domain having a sequence according to SEQ ID NO:1 , and the VL domain having a sequence according to SEQ ID NO:2;

(v) the VH domain having a sequence according to SEQ ID NO:3, and the VL domain having a sequence according to SEQ ID NO:4;

(vi) the VH domain having a sequence according to SEQ ID NO:5, and the VL domain having a sequence according to SEQ ID NO:6;

(vii) the VH domain having a sequence according to SEQ ID NO:11 , and the VL domain having a sequence according to SEQ ID NO: 12.

[0061 ] In a particular aspect, the present invention relates to a variant of an antibody or antigen-binding fragment thereof as defined hereinabove, wherein such variant is still specific for human RGMb, inhibits binding of human RGMb to human PD-L2, and inhibits binding of human RGMb to (i) neogenin, or to (ii) BMP2, BMP4, or BMP2 and BMP4, but comprises at most eight amino acid residue substitutions compared to the corresponding sequences of SEQ ID NOs: 7 and 8, 9 and 10, 13 and 14, 1 and 2, 3 and 4, 5 and 6 or 11 and 12, particularly at most seven amino acid residue substitutions, particularly at most six amino acid residue substitutions, more particularly at most five, four, three, or two amino acid residue substitutions. In particular embodiments, the variant comprises one amino acid residue substitution.

[0062] Thus, in particular embodiments of the present invention, said human antibody or antigen-binding fragment thereof comprises a combination of a VH domain belonging to the VH3 family and a VL domain belonging to the Vlambda2 family, wherein

(i) the VH domain has an amino acid sequence that has, in the framework regions, a sequence identity of at least 90 %, particularly at least 91 %, at least 93 %, at least 94 %, at least 95 %, at least 96 %, at least 97 %, or at least 98 %, of SEQ ID NO:7, and the VL domain has an amino acid sequence that has, in the framework regions, a sequence identity of at least 90 %, particularly at least 91 %, at least 92 %, at least 93 %, at least 95 %, at least 96 %, at least 97 %, or at least 98 %, of SEQ ID NO:8;

(ii) the VH domain has an amino acid sequence that has, in the framework regions, a sequence identity of at least 90 %, particularly at least 91 %, at least 93 %, at least 94 %, at least 95 %, at least 96 %, at least 97 %, or at least 98 %, of SEQ ID NO:9, and the VL domain has an amino acid sequence that has, in the framework regions, a sequence identity of at least 90 %, particularly at least 91 %, at least 92 %, at least 93 %, at least 95 %, at least 96 %, at least 97 %, or at least 98 %, of SEQ ID NO: 10;

(iii) the VH domain has an amino acid sequence that has, in the framework regions, a sequence identity of at least 90 %, particularly at least 91 %, at least 93 %, at least 94 %, at least 95 %, at least 96 %, at least 97 %, or at least 98 %, of SEQ ID NO: 13, and the VL domain has an amino acid sequence that has, in the framework regions, a sequence identity of at least 90 %, particularly at least 91 %, at least 92 %, at least 93 %, at least 95 %, at least 96 %, at least 97 %, or at least 98 %, of SEQ ID NO: 14;

(iv) the VH domain has an amino acid sequence that has, in the framework regions, a sequence identity of at least 90 %, particularly at least 91 %, at least 93 %, at least 94 %, at least 95 %, at least 96 %, at least 97 %, or at least 98 %, of SEQ ID NO:1 , and the VL domain has an amino acid sequence that has, in the framework regions, a sequence identity of at least 90 %, particularly at least 91 %, at least 92 %, at least 93 %, at least 95 %, at least 96 %, at least 97 %, or at least 98 %, of SEQ ID NO:2;

(v) the VH domain has an amino acid sequence that has, in the framework regions, a sequence identity of at least 90 %, particularly at least 91 %, at least 93 %, at least 94 %, at least 95 %, at least 96 %, at least 97 %, or at least 98 %, of SEQ ID NO:3, and the VL domain has an amino acid sequence that has, in the framework regions, a sequence identity of at least 90 %, particularly at least 91 %, at least 92 %, at least 93 %, at least 95 %, at least 96 %, at least 97 %, or at least 98 %, of SEQ ID NO:4; (vi) the VH domain has an amino acid sequence that has, in the framework regions, a sequence identity of at least 90 %, particularly at least 91 %, at least 93 %, at least 94 %, at least 95 %, at least 96 %, at least 97 %, or at least 98 %, of SEQ ID NO:5, and the VL domain has an amino acid sequence that has, in the framework regions, a sequence identity of at least 90 %, particularly at least 91 %, at least 92 %, at least 93 %, at least 95 %, at least 96 %, at least 97 %, or at least 98 %, of SEQ ID NO:6;

(vii) the VH domain has an amino acid sequence that has, in the framework regions, a sequence identity of at least 90 %, particularly at least 91 %, at least 93 %, at least 94 %, at least 95 %, at least 96 %, at least 97 %, or at least 98 %, of SEQ ID NO:11 , and the VL domain has an amino acid sequence that has, in the framework regions, a sequence identity of at least 90 %, particularly at least 91 %, at least 92 %, at least 93 %, at least 95 %, at least 96 %, at least 97 %, or at least 98 %, of SEQ ID NO: 12.

[0063] In particular embodiments, the substitutions only relate to non-canonical residues in the framework regions. In the context of the present invention, the term “canonical residue” refers to positions that are involved in stabilizing or supporting the particular three-dimensional presentation and arrangement of the CDR loops. The nature and position of the canonical residues are well known to one of ordinary skill in the art, and may, for example, be found in Knappik et al. (Joe. cit.).

[0064] In particular embodiments, the position(s) in said VH domains or said VL domains that is/are not identical to the corresponding amino acid residue(s) present in the given VH or VL domain sequences is/are selected from (i) positions in the N- terminal part of the VH or VL domain, in particular in positions 1 to 5 according to the AHo numbering, (ii) positions in the framework 4 region of the VH or VL domain, in particular in positions 145 to 149, and/or (iii) framework positions in the VH or VL domains that are shown in Tables 6D and 5B, respectively, of WO 97/08320, as being variable, in particular, wherein the most commonly appearing amino acid residue (“mcaa” in Tables 6D and 5B) has been found with a frequency below 95 % (“oomcaa” in Tables 6D and 5B), in particular below 90 %, more particularly below 80 %. In particular such embodiments, the residue present in the VH or VL of the antibody of the present invention that differs from the corresponding amino acid residue present in the given VH3 or Vlambda2 domain sequences is an amino acid residue being more often found at the corresponding position shown in Tables 6D and 5B, respectively, of WO 97/08320, in particular an amino acid residue that is one of the three most commonly appearing amino acid residues, in particular one of the two most commonly appearing amino acid residues, at the corresponding position in Tables 6D and 5B. In particular embodiments, when the amino acid residue present in one of the sequences of SEQ ID NOs: 7 and 8, 9 and 10, 13 and 14, 1 and 2, 3 and 4, 5 and 6 or 11 and 12 is the most frequently found amino acid residue at the corresponding position in Tables 6D and 5B, it is replaced in the variant sequence by the amino acid residue that is found with the second- or third-highest frequency, in particular the second-highest frequency. For example, the three most commonly appearing amino acid residues in position 1 of VH3 are, according to Table 6D of WO 97/08320, E (most frequent), Q (second most) and V (third most), and the two most commonly appearing amino acid residues in position 1 of Vlambda2 are, according to Table 5B of WO 97/08320, Q (highest frequency) and H (second most).

[0065] In a particular embodiment, said human antibody or antigen-binding fragment thereof is a full-length antibody, particularly a full-length antibody selected from the list of IgA, IgD, IgE and IgG.

[0066] In a particular embodiment, said full-length antibody is an IgG, particularly an IgG selected from the list of IgG 1 , lgG2, lgG3 and lgG4.

[0067] In particular embodiments of the present invention, said full length antibody is an lgG1 antibody with Leu234Ala/Leu235Ala mutations in the Fc region.

[0068] In order to increase the number of specificities/functionalities at the same or lower molecular weight, it is advantageous to use antibody-based compositions comprising antigen-binding fragments of antibodies, such as Fv, scFv, Fab, Fab’ and F(ab’)2 fragments and other antibody fragments. These smaller molecules retain the antigen-binding activity of the whole antibody and can also exhibit improved tissue penetration and pharmacokinetic properties in comparison to the whole immunoglobulin molecules and thus hold the promise for improved efficacy at the same or lower dose.

[0069] Thus, in a particular other embodiment, said human antibody or antigenbinding fragment thereof is an antigen-binding fragment, particularly an antigenbinding fragment selected from the list of: Fab fragment, F(ab)2 fragment, Fv fragment, dsFv fragment, scFv fragment, a dsscFv fragment, and a fusion protein comprising a Fab fragment, a F(ab)2 fragment, an Fv fragment, a dsFv fragment, an scFv fragment or an dsscFv, particularly wherein said antigen-binding fragment is an scFv fragment or a fusion protein comprising an scFv fragment, particularly an scFv-Fc fusion protein or an (scFv)2-Fc fusion protein.

[0070] In the context of the present invention, the term “Fab fragment” refers to a monovalent fragment consisting of the VL, VH, CL and CH1 domains; a “Fab’ fragment” refers to a Fab fragment additionally comprising part of an antibody hinge region, a “F(ab’)2 fragment” is a bivalent fragment comprising two Fab fragments linked by a disulfide bridge at the hinge region; an “Fv fragment” consists of the VL and VH domains of a single arm of an antibody; a “dsFv” refers to an Fv fragment linked by a disulfide bond, the term “scFv fragment” refers to a single-chain Fv fragment, wherein the VL and VH domains are connected by a polypeptide linker, and the term “dsscFv fragment” refers to an scFv fragment, wherein the two variable domains are additionally linked by a disulfide bridge.

[0071 ] In the context of the present invention, the term “polypeptide linker” refers to a linker consisting of a chain of amino acid residues linked by peptide bonds that is connecting two domains, each being attached to one end of the linker. The polypeptide linker should have a length that is adequate to link two molecules in such a way that they assume the correct conformation relative to one another so that they retain the desired activity. In particular embodiments, the polypeptide linker has a continuous chain of between 2 and 30 amino acid residues (e. g., 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 , 12, 13, 14, 15, 16, 17, 18, 19, 20, 21 , 22, 23, 24, 25, 26, 27, 28, 29, or 30 amino acid residues). In addition, the amino acid residues selected for inclusion in the polypeptide linker should exhibit properties that do not interfere significantly with the activity of the polypeptide. Thus, the linker peptide on the whole should not exhibit a charge that would be inconsistent with the activity of the polypeptide, or interfere with internal folding, or form bonds or other interactions with amino acid residues in one or more of the monomers that would seriously impede the binding of receptor monomer domains. In particular embodiments, the polypeptide linker is non-structured polypeptide. Useful linkers include glycine-serine, or GS linkers. By “Gly-Ser” or “GS” linkers is meant a polymer of glycines and serines in series (including, for example, (Gly-Ser)n (SEQ ID NO:26), (GSGGS)n (SEQ ID NO:27), (GGGGS)n (SEQ ID NO:28) and (GGGS)n (SEQ ID NO:29), where n is an integer of at least one), glycine-alanine polymers, alanine- serine polymers, and other flexible linkers such as the tether for the shaker potassium channel, and a large variety of other flexible linkers, as will be appreciated by those in the art. Glycine-serine polymers are preferred since oligopeptides comprising these amino acids are relatively unstructured, and therefore may be able to serve as a neutral tether between components. Secondly, serine is hydrophilic and therefore able to solubilize what could be a globular glycine chain. Third, similar chains have been shown to be effective in joining subunits of recombinant proteins such as single-chain antibodies.

[0072] In a preferred embodiment, the binding domain of an antibody applied in the present invention is a single-chain Fv fragment (scFv).

[0073] In particular embodiments, the two variable domains of an antigen-binding fragment, as in an Fv or an scFv fragment, are stabilized by an interdomain disulfide bond, in particular, said VH domain comprises a single cysteine residue in position 51 (AHo numbering) and said VL domain comprises a single cysteine residue in position 141 (AHo numbering).

[0074] The term “immunoglobulin Fc region”, or “Fc region” or “Fc” in abbreviated form, as used herein, refers to the CH2 and CH3 domains of the heavy chain constant regions.

[0075] In particular embodiments, said human antibody or antigen-binding fragment thereof is an antigen-binding fragment or the multispecific antibody-based binding composition comprises an immunoglobulin Fc region polypeptide. The term “Fc region” herein is used to define a C-terminal region of an immunoglobulin heavy chain, including native-sequence Fc regions and variant Fc regions. Native-sequence Fc regions include human lgG1 , lgG2 (lgG2A, lgG2B), lgG3 and lgG4. “Fc receptor” or “FcR” describes a receptor that binds to the Fc region of an antibody. Particularly, the FcR is a native sequence human FcR, which binds an IgG antibody (a gamma receptor) and includes receptors of the FcyRI, FcyRII, and FcyRIII subclasses, including allelic variants and alternatively spliced forms of these receptors, FcyRII receptors including FcyRIIA (an "activating receptor") and FcyRI IB (an “inhibiting receptor”), which have similar amino acid sequences that differ primarily in the cytoplasmic domains thereof. Activating receptor FcyRIIA contains an immunoreceptor tyrosine-based activation motif (ITAM) in its cytoplasmic domain. Inhibiting receptor FcyRIIB contains an immunoreceptor tyrosine-based inhibition motif (ITIM) in its cytoplasmic domain (see M. Daeron, Annu. Rev. Immunol. 5:203- 234 (1997). FcRs are reviewed in Ravetch and Kinet, Annu. Rev. Immunol. 9: 457-92 (1991 ); Capet et al, Immunomethods 4: 25-34 (1994); and de Haas et al, J. Lab. Clin. Med. 126: 330-41 (1995). Other FcRs, including those to be identified in the future, are encompassed by the term “FcR” herein. The term “Fc receptor” or “FcR” also includes the neonatal receptor, FcRn, which is responsible for the transfer of maternal IgGs to the fetus. Guyer et al., J. Immunol. 117: 587 (1976) and Kim et al., J. Immunol. 24: 249 (1994). Methods of measuring binding to FcRn are known (see, e. g., Ghetie and Ward, Immunol. Today 18: (12): 592-8 (1997); Ghetie etal., Nature Biotechnology 15 (7): 637-40 (1997); Hinton et al., J. Biol. Chem. TJI (8): 6213-6 (2004); WO 2004/92219 (Hinton et al). Binding to FcRn in vivo and serum half-life of human FcRn high-affinity binding polypeptides can be assayed, e. g., in transgenic mice or transfected human cell lines expressing human FcRn, or in primates to which the polypeptides having a variant Fc region are administered. WO 2004/42072 (Presta) describes antibody variants that improved or diminished binding to FcRs. See also, e. g., Shields et al., J. Biol. Chem. 9(2): 6591 -6604 (2001 ).

[0076] In particular embodiments of the particular aspect presented in [0061 ] to [0064], the present invention relates to a variant of an a human antibody or antigenbinding fragment thereof as defined hereinabove that is an antigen-binding fragment, wherein such variant is still specific for RGMb, inhibits binding of human RGMb to human PD-L2, and inhibits binding of human RGMb to (i) neogenin, or to (ii) BMP2, BMP4, or BMP2 and BMP4, but comprises at most eight amino acid residue substitutions compared to the corresponding sequence of SEQ ID NOs: 7 and 8, 9 and 10, 13 and 14, 1 and 2, 3 and 4, 5 and 6 or 11 and 12, particularly at most seven amino acid residue substitutions, particularly at most six amino acid residue substitutions, more particularly at most five, four, three, two or just one amino acid residue substitution(s), wherein such substitutions are performed at positions in a variable domain, which are no longer in their native context, when present in such antigen-binding fragment. Examples include variable domain residues originally present in the interface between variable and constant domain, which become solvent- exposed when present in an Fv or scFv fragment. The identification of such residues and approaches for identifying suitable substitutions are well known to one of ordinary skill in the art (see, for example, Ewert et al., Biochemistry 42 (2003) 1517-1528; Worn & Pluckthun, J. Mol. Biol. 305 (2001 ) 989-1010; Nieba et al., Protein Eng. 10 (1997) 435-444). Additionally, substitutions may be performed at positions that are known or suspected to result in improved biochemical properties (see, for, example, WO 2009/000099) or result in decreased immunogenicity (see, for example, WO 2011/075861 ).

[0077] Suitably, the human antibody or antigen-binding fragment thereof of the present invention is an isolated human antibody or antigen-binding fragment thereof. The term “isolated human antibody or antigen-binding fragment thereof”, as used herein, refers to a composition that is substantially free of other antibody-based products having different antigenic specificities (e. g., an isolated antibody or antigenbinding fragment thereof that specifically binds RGMb is substantially free of antibodies that specifically bind antigens other than RGMb. Moreover, an isolated human antibody or antigen-binding fragment thereof may be substantially free of other cellular material and/or chemicals.

[0078] In particular embodiments of the present invention, the human antibody or antigen-binding fragment thereof is antagonistic to human RGMb.

[0079] In the context of the present invention, the term “is antagonistic to human RGMb” refers to the capacity of antibodies to bind RGMb and block other proteins from reacting with RGMb. An antagonistic property of anti-RGMb may be measured in cell-based assays using RGMb expressing cells. An antagonistic antibody does not induce visible reaction when binds with antigen.

[0080] In particular embodiments of the present invention, the antibody deactivates, reduces, or inhibits an activity of human RGMb.

[0081 ] In the context of the present invention, the term “deactivates, reduces, or inhibits an activity of human RGMb” means inhibition of its interaction with binding partners that may be measured in a biochemical or cellular assays.

[0082] In particular embodiments of the present invention, human antibody or antigen-binding fragment thereof modulates TNFalpha, IFNgamma, and IL-2 expression by immune cells.

[0083] In the context of the present invention, the term “induces TNFalpha, IFNgamma, and IL-2 production by immune cells” refers to the expression of at least one of these cytokines from cells that play a role in immune responses. A cytokine may be measured by RNAseq, real-time PCR or ELISA, and signals are significantly above the background.

[0084] In a second aspect, the present invention relates to a multispecific antibodybased binding composition with specificity for at least RGMb and a second antigen, wherein said multispecific antibody-based binding composition comprises a first human antibody or antigen-binding fragment thereof, which is a human antibody or antigen-binding fragment thereof of the present invention, and at least a second human antibody or antigen-binding fragment thereof with specificity for a different antigen.

[0085] The term “multispecific antibody-based binding composition” as used herein, refers to an antibody-based binding composition that binds to two or more different antigens, including to RGMb. The term “multispecific antibody-based binding composition” includes bispecific, trispecific, tetraspecific, pentaspecific and hexaspecific. The term “bispecific antibody” as used herein, refers to an antibody that binds to at least two different antigens. The term “trispecific antibody” as used herein, refers to an antibody that binds to at least three different antigens.

[0086] In particular embodiments, said multispecific antibody-based binding composition is bispecific.

[0087] In particular embodiments, said second human antibody or antigen-binding fragment thereof is specific for human PD-L1 or PD-1 , in particular for human PD-1 .

[0088] The term “epitope” means a protein determinant capable of specific binding to an antibody. Epitopes usually consist of chemically active surface groupings of molecules such as amino acids or sugar side chains and usually have specific three- dimensional structural characteristics, as well as specific charge characteristics. “Conformational” and “linear” epitopes are distinguished in that the binding to the former but not the latter is lost in the presence of denaturing solvents.

[0089] The term “conformational epitope” as used herein refers to amino acid residues of an antigen that come together on the surface when the polypeptide chain folds to form the native protein.

[0090] The term “linear epitope” refers to an epitope, wherein all points of interaction between the protein and the interacting molecule (such as an antibody) occurring linearly along the primary amino acid sequence of the protein (continuous). [0091 ] The term “recognize” as used herein refers to an antibody antigen-binding fragment thereof that finds and interacts (e. g., binds) with its conformational epitope. [0092] The term “avidity” as used herein refers to an informative measure of the overall stability or strength of the antibody-antigen complex. It is controlled by three major factors: antibody epitope affinity, the valency of both the antigen and antibody, and the structural arrangement of the interacting parts. Ultimately these factors define the specificity of the antibody, that is, the likelihood that the particular antibody is binding to a precise antigen epitope.

[0093] The term “same epitope”, as used herein, refers to individual protein determinants on the same protein capable of specific binding to an antibody, where these individual protein determinants are identical, /. e. consist of identical chemically active surface groupings of molecules such as amino acids or sugar side chains having identical three-dimensional structural characteristics, as well as identical charge characteristics. The term “different epitope”, as used herein in connection with a specific protein target, refers to individual protein determinants on the same protein capable of specific binding to an antibody, where these individual protein determinants are not identical, /. e. consist of non-identical chemically active surface groupings of molecules such as amino acids or sugar side chains having different three- dimensional structural characteristics, as well as different charge characteristics. These different epitopes can be overlapping or non-overlapping.

[0094] As used herein, the term “affinity” refers to the strength of interaction between antibody and antigen at single antigenic sites. Within each antigenic site, the variable region of the antibody “arm” interacts through weak non-covalent forces with antigen at numerous sites; the more interactions, the stronger the affinity.

[0095] “Binding affinity” generally refers to the strength of the total sum of non- covalent interactions between a single binding site of a molecule (e. g., of an antibody) and its binding partner (e. g., an antigen or, more specifically, an epitope on an antigen). Unless indicated otherwise, as used herein, “binding affinity”, “bind to”, “binds to” or “binding to” refers to intrinsic binding affinity that reflects a 1 :1 interaction between members of a binding pair (e. g., an antibody fragment and antigen). The affinity of a molecule X for its partner Y can generally be represented by the dissociation constant (KD). Affinity can be measured by common methods known in the art, including those described herein. Low-affinity antibodies generally bind antigen slowly and tend to dissociate readily, whereas high-affinity antibodies generally bind antigen faster and tend to remain bound longer. A variety of methods of measuring binding affinity are known in the art, any of which can be used for purposes of the present invention. Specific illustrative and exemplary embodiments for measuring binding affinity, /. e. binding strength are described in the following.

[0096] The term “Kassoc”, “K_a” or “K_on”, as used herein, are intended to refer to the association rate of a particular antibody-antigen interaction, whereas the term “Kdis”, “Kd” or “Koff”, as used herein, is intended to refer to the dissociation rate of a particular antibody- antigen interaction. In one embodiment, the term “KD”, as used herein, is intended to refer to the dissociation constant, which is obtained from the ratio of Kd to

Ka (/. e. Kd/Ka) and is expressed as a molar concentration (M). The “KD” or “KD value” or "KD" or "KD value" according to this invention is in one embodiment measured by using surface-plasmon resonance assays.

[0097] In particular embodiments, said second human antibody or antigen-binding fragment thereof comprises a combination of

(i) the VH domain having a sequence according to SEQ ID NO: 18, and the VL domain having a sequence according to SEQ ID NO:19;

(ii) the VH domain having a sequence according to SEQ ID NO:20, and the VL domain having a sequence according to SEQ ID NO:21 ; or

(iii) the VH domain having a sequence according to SEQ ID NO:22, and the VL domain having a sequence according to SEQ ID NO:23.

[0098] In a particular embodiment, said first human antibody or antigen-binding fragment thereof, and said second human antibody or antigen-binding fragment thereof, are present as two separate monospecific constructs.

[0099] In a particular other embodiment, said first human antibody or antigen-binding fragment thereof, and said second human antibody or antigen-binding fragment thereof, are comprised in one multispecific construct, particularly a multispecific construct selected from the list of: a fusion of antibody fragments, in particular scFv fragments, to the N- and/or C-terminal ends of the antibody heavy and/or light chains of full immunoglobulins, in particular Momson-H or Momson-L constructs, diabody; scDb; tandem di-scFv, tandem tri-scFv; Fab-scFv; scFab-dsscFv; Fab-(scFv)2; (Fab’)2; Fab-(Fv)2; triabody; scDb-scFv; tandem tri-scFv, a scFab-dsscFv, and constructs comprising two antigen-binding fragments, each independently selected from Fab fragment, F(ab’)2 fragment, Fv fragment, dsFv fragment, and scFv fragment, wherein each of said two antigen-binding fragments is fused to the N- and/or the C- terminus of one chain of a heterodimerization domain, particularly one chain of a heterodimeric Fc domain.

[0100] The term “diabody” refers to an antibody fragment with two antigen-binding sites, which fragments comprise a VH connected to VL in the same polypeptide chain (VH-VL). By using a linker that is too short to allow pairing between the two domains on the same chain, the domains are forced to pair with the complementary domains of another chain to create two antigen-binding sites. A diabody may be bivalent or bispecific. Diabodies are described more fully in, for example, EP 404 097, WO 93/01161 , Hudson et al., Nat. Med. 9:129-134 (2003), and Holliger et al., Proc. Natl. Acad. Sci. USA 90: 6444-6448 (1993). Triabodies and tetrabodies are also described in Hudson et al., Nat. Med. 9:129-134 (2003).

[0101 ] The term “scDb” refers to a single-chain diabody, which comprises two variable heavy chain domains (VH) or fragments thereof and two variable light chain domains (VL) or fragments thereof connected by linkers L1 , L2 and L3 in the order VHA-L1 -VLB-L2-VHB-L3-VLA, VHA-L1 -VHB-L2-VLB-L3-VLA, VLA-L1 -VLB-L2-VHB- L3-VHA, VLA-L1 -VHB-L2-VLB-L3-VHA, VHB-L1 -VLA-L2-VHA-L3-VLB, VHB-L1 - VHA-L2-VLA-L3-VLB, VLB-L1 -VLA-L2-VHA-L3-VHB or VLB-L1 -VHA-L2-VLA-L3- VHB, wherein the VLA and VHA domains jointly form the antigen-binding site for the first antigen, and VLB and VHB jointly form the antigen-binding site for the second antigen.

[0102] The linker L1 particularly is a peptide of 2-10 amino acids, more particularly 3- 7 amino acids, and most particularly 5 amino acids, and linker L3 particularly is a peptide of 1 -10 amino acids, more particularly 2-7 amino acids, and most particularly 5 amino acids. In particular embodiments, the linker L1 and/or L3 comprises one or two units of four (4) glycine amino acid residues and one (1 ) serine amino acid residue (GGGGS)n (SEQ ID NO:30), wherein n=1 or 2, particularly n=1.

[0103] The middle linker L2 particularly is a peptide of 10-40 amino acids, more particularly 15-30 amino acids, and most particularly 20-25 amino acids. In particular embodiments, said linker L2 comprises two or more units of four (4) glycine amino acid residues and one (1 ) serine amino acid residue (GGGGS)n (SEQ ID NO:31 ), wherein n= 2, 3, 4, 5, 6, 7 or 8, particularly n=4 or 5. [0104] The term “scDb-scFv” refers to an antibody format, wherein a single-chain Fv (scFv) fragment is fused by a flexible Gly-Ser linker to a single-chain diabody (scDb). In one embodiment, said flexible Gly-Ser linker is a peptide of 2-40 amino acids, e. g., 2-35, 2-30, 2-25, 2-20, 2-15, 2-10 amino acids, particularly 10 amino acids. In particular embodiments, said linker comprises one or more units of four (4) glycine amino acid residues and one (1 ) serine amino acid residue (GGGGS)n (SEQ ID NO:30), wherein n=1 , 2, 3, 4, 5, 6, 7 or 8, particularly n=2.

[0105] In particular other embodiments, the multispecific antibody-based binding composition is in a format selected from a multispecific, e. g. at least bispecific, format selected from the list of a tandem scDb (Tandab); tetrabody; scDb-scFv; di-diabody; scFv-Fc-scFv fusion (ADAPTIR); DVD-lg; IgG-scFv fusions, such as CODV-IgG, Morrison (IgG CHs-scFv fusion (Morrison L) or IgG CL-scFv fusion (Morrison H)), bsAb (scFv linked to C-terminus of light chain), Bs1Ab (scFv linked to N-terminus of light chain), Bs2Ab (scFv linked to N-terminus of heavy chain), Bs3Ab (scFv linked to C- terminus of heavy chain), Ts1Ab (scFv linked to N-terminus of both heavy chain and light chain) and Ts2Ab (dsscFv linked to C-terminus of heavy chain); a DART™; a TRIDENT™; a scDb, a tandem tri-scFv, a MATCH (described in WO 2016/0202457; Egan T. et al., MAbs 9 (2017) 68-84) and DuoBodies (bispecific IgGs prepared by the Duobody technology) (MAbs. 9 (2017) 182-212).

[0106] Suitably, the binding domains of the multispecific antibody-based binding composition are operably linked.

[0107] The term “operably linked”, as used herein, indicates that two molecules (e. g., polypeptides, domains, binding domains) are attached in a way that each molecule retains functional activity. Two molecules can be “operably linked” whether they are attached directly or indirectly (e. g., via a linker, via a moiety, via a linker to a moiety). The term “linker” refers to a peptide or other moiety that is optionally located between binding domains or antibody fragments used in the invention. A number of strategies may be used to covalently link molecules together. These include, but are not limited to, polypeptide linkages between N- and C-termini of proteins or protein domains, linkage via disulfide bonds, and linkage via chemical cross-linking reagents. In one aspect of this embodiment, the linker is a peptide bond generated by recombinant techniques or peptide synthesis. Choosing a suitable linker for a specific case where two polypeptide chains are to be connected depends on various parameters, including but not limited to the nature of the two polypeptide chains (e. g., whether they naturally oligomerize), the distance between the N- and the C-termini to be connected if known, and/or the stability of the linker towards proteolysis and oxidation. Furthermore, the linker may contain amino acid residues that provide flexibility.

[0108] In particular embodiments, the present invention relates to a multispecific antibody-based binding composition with specificity for at least a first epitope of RGMb and at least a second epitope of RGMb.

[0109] In particular such embodiments, said first epitope is the epitope 1 (as defined herein), which overlaps with the BPM2- and BMP4-binding regions, and said second epitope is the epitope 2 (as defined herein), which coincides with the RGMb/Neogenin binding region.

[0110] In particular such embodiments, the present invention relates to a multispecific antibody-based binding composition with specificity for at least a first epitope of RGMb and at least a second epitope of RGMb and a second antigen, wherein said multispecific antibody-based binding composition comprises a first human antibody or antigen-binding fragment thereof, which is a human antibody or antigen-binding fragment thereof of the present invention directed at a first epitope of RGMb, a second human antibody or antigen-binding fragment thereof, which is a human antibody or antigen-binding fragment thereof of the present invention directed at a second epitope of RGMb, and at least a third human antibody or antigen-binding fragment thereof with specificity for a different antigen.

[0111 ] In particular such embodiments, said first human antibody of antigen binding fragment thereof comprises an RGMb-binding site selected from 2C11 (SEQ ID NOs: 7 and 8) and 5B05 (SEQ ID Nos: 9 and 10), and said second human antibody of antigen binding fragment thereof comprises an RGMb-binding site selected from 5C10 (SEQ ID Nos: 1 and 2), 3C08 (SEQ ID Nos: 5 and 6) and 2G03 (SEQ ID Nos: 3 and 4).

[0112] In particular embodiments, said multispecific antibody-based binding composition is trispecific.

[0113] In particular embodiments, said third human antibody or antigen-binding fragment thereof is specific for human PD-L1 or PD-1 , in particular for human PD-1 .

[0114] The binding domains of the multispecific antibody-based binding composition of the present invention may be able to bind to RGMb and to the second antigen (and/or to any further epitope(s) and/or antigen(s)) in an alternative binding mode or independent of each other and thus simultaneously. The term” alternative binding mode” refers to a situation, where binding of one antigen by the corresponding binding domain of the multispecific antibody-based binding composition of the present invention blocks binding of a second binding domain of the of said multispecific antibody-based binding composition to its antigen. In contrast, the term “simultaneously”, as used in this connection refers to the simultaneous binding of at least two binding domains to their respective antigens.

[0115] In particular embodiments, the human antibody or antigen-binding fragment thereof, or the multispecific antibody-based binding composition, may further comprise an additional functional domain that is not related to binding to RGMb, and to binding to said second antigen. In a particular embodiment, such additional functional domain is increasing the half-life of the antibody-based binding composition, or the multispecific antibody-based binding composition, respectively, of the present invention. In particular such embodiments, the additional functional domain is a human serum albumin binding domain (hSA-BD) having a specificity to human serum albumin.

[0116] The term “hSA” refers in particular to human serum albumin with UniProt ID number P02768. Human Serum Albumin (hSA) is a 66.4 kDa abundant protein in human serum (50 % of total protein) composed of 585 amino acids (Sugio, Protein Eng, Vol. 12, 1999, 439-446). Multifunctional hSA can transport a number of metabolites such as fatty acids, metal ions, bilirubin and some drugs (Fanali, Molecular Aspects of Medicine, Vol. 33, 2012, 209-290). hSA concentration in serum is around 3.5 - 5 g/dl. Albumin-binding antibodies and fragments thereof may be used for example, for extending the in vivo serum half-life of drugs or proteins conjugated thereto.

[0117] Suitable hSA-BDs comprises or is derived from a binding domain selected from the group consisting of: (i) polypeptides that bind serum albumin (see, for example, Smith et al., 2001 , Bioconjugate Chem. 12:750-756; EP 0 486 525; US 6,267,964; WO 2004/001064; WO 2002/076489; and WO 2001/45746); (ii) anti-serum albumin binding single variable domains described in Holt et al., Protein Engineering, Design & Selection, vol 21 , 5, pp283-288, WO 2004/003019, WO 2008/096158, WO 2005/118642, WO 2006/0591056 and WO 2011/006915; (iii) anti-serum albumin antibodies described in WO 2009/040562, WO 2010/035012 and WO 2011/086091.

[0118] In a third aspect, the present invention relates to a nucleic acid encoding the human antibody or antigen-binding fragment thereof of the present invention, or the multispecific antibody-based binding composition of the present invention.

[0119] The human antibody or antigen-binding fragment thereof of the present invention, or the multispecific antibody-based binding composition of the present invention can be produced using any convenient antibody-manufacturing method known in the art (see, e. g., Fischer, N. & Leger, O., Pathobiology 74 (2007) 3-14 with regard to the production of bispecific constructs; Hornig, N. & Farber-Schwarz, A., Methods Mol. Biol. 907 (2012)713-727, and WO 99/57150 with regard to bispecific diabodies and tandem scFvs). Specific examples of suitable methods for the preparation of the bispecific construct further include, inter alia, the Genmab (see Labrijn et al., Proc. Natl. Acad. Sci. USA 110 (2013) 5145-5150) and Merus (see de Kruif et al., Biotechnol. Bioeng. 106 (2010) 741 -750) technologies. Methods for production of bispecific antibodies comprising a functional antibody Fc part are also known in the art (see, e. g., Zhu et al., Cancer Lett. 86 (1994) 127-134); and Suresh et al., Methods Enzymol. 121 (1986) 210-228).

[0120] In the case of multispecific antibody-based binding composition of the present invention, the combination of the antigen-binding domains or fragments or parts thereof of two or more different monoclonal antibodies can be prepared by conjugating the constituent binding specificities, using methods known in the art. For example, they can be conjugated by sulfhydryl bonding of the C-terminus hinge regions of the two heavy chains. In a particular embodiment, the hinge region is modified to contain an odd number of sulfhydryl residues, for example one, prior to conjugation.

[0121 ] Alternatively, two or more binding specificities can be encoded in the same vector and expressed and assembled in the same host cell. This method is particularly useful where the bispecific molecule is a mAb x mAb, mAb x Fab, Fab x F (ab')2 or ligand x Fab fusion protein. The multispecific antibody used in the invention can be a single chain multispecific antibody comprising at least two binding determinants. The multispecific antibody use in the invention can also comprise at least two of said singlechain molecules. Methods for preparing multispecific antibodies and molecules are described, for example, in U.S. Pat. No. 5,260,203; U.S. Pat. No. 5,455,030; U.S. Pat. No. 4,881 ,175; U.S. Pat. No. 5,132,405; U.S. Pat. No. 5,091 ,513; U.S. Pat. No. 5,476,786; U.S. Pat. No. 5,013,653; U.S. Pat. No. 5,258,498; and U.S. Pat. No. 5,482,858.

[0122] Binding of the multispecific antibodies to their specific targets can be confirmed by, for example, enzyme-linked immunosorbent assay (ELISA), radioimmunoassay (REA), FACS analysis, bioassay (e. g., growth inhibition), or Western Blot assay. Each of these assays generally detects the presence of proteinantibody complexes of particular interest by employing a labeled reagent (e. g., an antibody) specific for the complex of interest.

[0123] In particular embodiments, the nucleic acid comprises a contiguous nucleic acid encoding the complete human antibody or antigen-binding fragment thereof of the present invention. In particular other embodiments, the nucleic acid consists of a single nucleic acid comprising two or more separate nucleic acid regions encoding two or more components of the complete human antibody or antigen-binding fragment thereof of the present invention. In particular other embodiments, the nucleic acid consists of two or more single nucleic acids, each encoding one of two or more components of the complete human antibody or antigen-binding fragment thereof of the present invention.

[0124] In a fourth aspect, the present invention relates to a vector comprising the nucleic acid of the present invention.

[0125] In a fifth aspect, the present invention relates to a host cell comprising the nucleic acid of the present invention or the vector of the present invention.

[0126] In a sixth aspect, the present invention relates to a method for producing the human antibody or antigen-binding fragment thereof of the present invention, or the multispecific antibody-based binding composition of the present invention, comprising a step selected from (a) expressing the nucleic acid of the present invention in an expression system; (b) expressing a nucleic acid from the vector of the present invention; and (c) culturing the host cell of the present invention.

[0127] In a particular embodiment, the method further comprises the step of isolating the human antibody or antigen-binding fragment thereof of the present invention, or the multispecific antibody-based binding composition of the present invention.

[0128] In a seventh aspect, the present invention relates to a multifunctional antibodybased binding composition, wherein said multifunctional antibody-based binding composition comprises a human antibody or antigen-binding fragment thereof, which is a human antibody or antigen-binding fragment thereof of the present invention, and at least a second non-antibody-based agent that selectively inhibits or blocks the expression or activity of PD-1 or PD-L1 .

[0129] In a particular embodiment, said multifunctional antibody-based binding composition is bifunctional.

[0130] In an eighth aspect, the present invention relates to a pharmaceutical composition comprising the human antibody or antigen-binding fragment thereof of the present invention, the multispecific antibody-based binding composition of the present invention, or the multifunctional antibody-based binding composition of the present invention and a pharmaceutically acceptable carrier.

[0131 ] "Pharmaceutically acceptable carrier" means a medium or diluent that does not interfere with the structure of the antibodies. Pharmaceutically acceptable carriers enhance or stabilize the composition, or facilitate preparation of the composition. Certain of such carriers enable pharmaceutical compositions to be formulated as, for example, tablets, pills, dragees, capsules, liquids, gels, syrups, slurries, suspension and lozenges for the oral ingestion by a subject. Certain of such carriers enable pharmaceutical compositions to be formulated for injection, infusion or topical administration. For example, a pharmaceutically acceptable carrier can be a sterile aqueous solution.

[0132] Pharmaceutically acceptable carriers include but are not limited to solvents, buffer solutions, dispersion media, coatings, antibacterial and antifungal agents, isotonic and absorption delaying agents, and the like that are physiologically compatible. In the context of pharmaceutical compositions based on mRNA, said pharmaceutically acceptable carriers include but are not limited to lipids and polymeric compounds forming liposomes and polymersomes, respectively.

[0133] The pharmaceutical compositions of the present invention can be prepared in accordance with methods well known and routinely practiced in the art. See, e. g., Remington: The Science and Practice of Pharmacy, Mack Publishing Co., 20th ed., 2000; and Sustained and Controlled Release Drug Delivery Systems, J. R. Robinson, ed., Marcel Dekker, Inc., New York, 1978. Pharmaceutical compositions are preferably manufactured under GMP conditions. Typically, a therapeutically effective dose or efficacious dose of the human antibody or antigen-binding fragment thereof of the present invention, or the multispecific antibody-based binding composition of the present invention is employed in the pharmaceutical compositions of the invention. The multispecific antibodies are formulated into pharmaceutically acceptable dosage forms by conventional methods known to those of skill in the art. Dosage regimens are adjusted to provide the optimum desired response (e. g., a therapeutic response). For example, a single bolus may be administered, several divided doses may be administered over time, or the dose may be proportionally reduced or increased as indicated by the exigencies of the therapeutic situation. It is especially advantageous to formulate parenteral compositions in dosage unit form for ease of administration and uniformity of dosage. Dosage unit form as used herein refers to physically discrete units suited as unitary dosages for the subjects to be treated; each unit contains a predetermined quantity of active compound calculated to produce the desired therapeutic effect in association with the required pharmaceutical carrier.

[0134] Actual dosage levels of the active ingredients in the pharmaceutical compositions or the kit can be varied so as to obtain an amount of the active ingredient which is effective to achieve the desired therapeutic response for a particular patient, composition, and mode of administration, without being toxic to the patient. The selected dosage level depends upon a variety of pharmacokinetic factors including the activity of the particular compositions used in the present invention employed, or the ester, salt or amide thereof, the route of administration, the time of administration, the rate of excretion of the particular compound being employed, the duration of the treatment, other drugs, compounds and/or materials used in combination with the particular compositions employed, the age, sex, weight, condition, general health and prior medical history of the patient being treated, and like factors.

[0135] The pharmaceutical compositions or the kit of the invention can be administered by a variety of methods known in the art. In the case of administering the multispecific antibody components of the kits, administration may be done concomitantly or sequentially. In the case of a sequential administration, the multispecific antibody components may be administered based on individually adjusted administration schemes and regimens. The route and/or mode of administration vary depending upon the desired results. Administration can be intravenous, intramuscular, intraperitoneal, or subcutaneous, or administered proximal to the site of the target. In the context of the present invention, administration by the intranasal or inhalative route is particularly preferred. The pharmaceutically acceptable carrier should be suitable for intravenous, intramuscular, subcutaneous, parenteral, spinal or epidermal administration (e. g., by injection or infusion). Depending on the route of administration, the active compound, /. e., the human antibody or antigenbinding fragment thereof of the present invention, the multispecific antibody-based binding composition of the present invention applied in the pharmaceutical composition of the invention, may be coated in a material to protect the compound from the action of acids and other natural conditions that may inactivate the compound. [0136] The pharmaceutical composition of the invention is usually administered on multiple occasions. Intervals between single dosages can be daily, weekly, monthly or yearly. Intervals can also be irregular as indicated by measuring blood levels of the human antibody or antigen-binding fragment thereof of the present invention or the multispecific antibody-based binding composition of the present invention in the patient. Alternatively, the pharmaceutical composition of the invention can be administered as a sustained release formulation, in which case less frequent administration is required. Dosage and frequency vary depending on the half-life of the antibodies in the patient. The dosage and frequency of administration can vary depending on whether the treatment is prophylactic or therapeutic. In prophylactic applications, a relatively low dosage is administered at relatively infrequent intervals over a long period of time. Some patients continue to receive treatment for the rest of their lives. In therapeutic applications, a relatively high dosage at relatively short intervals is sometimes required until progression of the disease is reduced or terminated, and preferably until the patient shows partial or complete amelioration of symptoms of disease. Thereafter, the patient can be administered a prophylactic regime.

[0137] In an eighth aspect, the present invention relates to a pharmaceutical composition comprising the human antibody or antigen-binding fragment thereof of the present invention, the multispecific antibody-based binding composition of the present invention, or the multifunctional antibody-based binding composition of the present invention and a pharmaceutically acceptable carrier.

[0138] In a nineth aspect, the present invention relates to the human antibody or antigen-binding fragment thereof of the present invention, the multispecific antibodybased binding composition of the present invention, the multifunctional antibody- based binding composition of the present invention, or the pharmaceutical composition of the present invention, for use in the prevention or treatment of cancer.

[0139] The terms “treatment”, “treating”, “treat”, “treated”, and the like, as used herein, refer to obtaining a desired pharmacologic and/or physiologic effect. The effect may be therapeutic in terms of a partial or complete cure for a disease and/or adverse effect attributable to the disease or delaying the disease progression. “Treatment”, as used herein, covers any treatment of a disease in a mammal, e. g., in a human, and includes: (a) inhibiting the disease, /. e., arresting its development; and (b) relieving the disease, /. e., causing regression of the disease.

[0140] The term “therapeutically effective amount” or “efficacious amount” refers to the amount of an agent that, when administered to a subject for treating a disease, is sufficient to affect such treatment for the disease. The “therapeutically effective amount” will vary depending on the agent, the disease and its severity and the age, weight, etc., of the subject to be treated.

[0141 ] In a tenth aspect, the present invention relates to a method of treating an individual having a condition, where RGMb is expressed and that would benefit from upregulation of an immune response, in particular cancer, comprising the step of administering to said individual a therapeutically effective amount of the human antibody or antigen-binding fragment thereof of the present invention, the multispecific antibody-based binding composition of the present invention, the multifunctional antibody-based binding composition of the present invention, or the pharmaceutical composition of the present invention.

[0142] In particular embodiments, the method further comprises the step of administering to said individual a therapeutically effective amount of at least one agent that selectively inhibits or blocks the expression or activity PD-1 such that the condition that would benefit from upregulation of an immune response is treated.

[0143] In particular embodiments, the method comprises the step of administering to said individual a therapeutically effective amount of the multispecific antibody-based binding composition of section [0087] or section [0113],

[0144] In particular embodiments, the method further comprises the step of administering to said individual a therapeutically effective amount of one or more additional therapeutic agents. [0145] In particular embodiments, said one or more additional therapeutic agents are selected from the group consisting of immunotherapeutic agents, immune checkpoint inhibitors, vaccines, chemotherapeutic agents, radiation therapy, and epigenetic modifiers.

Sequence listing (CDRs shown in bold/underlined are defined according to the Kabat system)

Table 1. Anti-RGMb antibodies (CDR residues shown in bold/underlined letters).

[0146] Throughout the text of this application, should there be a discrepancy between the text of the specification (e. g., Table 1 ) and the sequence listing, the text of the specification shall prevail.

[0147] It is appreciated that certain features of the invention, which are, for clarity, described in the context of separate embodiments, may also be provided in combination in a single embodiment. Conversely, various features of the invention, which are, for brevity, described in the context of a single embodiment, may also be provided separately or in any suitable sub-combination. All combinations of the embodiments pertaining to the invention are specifically embraced by the present invention and are disclosed herein just as if each and every combination was individually and explicitly disclosed. In addition, all sub-combinations of the various embodiments and elements thereof are also specifically embraced by the present invention and are disclosed herein just as if each and every such sub-combination was individually and explicitly disclosed herein.

[0148] The present invention is not to be limited in scope by the specific embodiments described herein. Indeed, various modifications of the invention in addition to those described herein will become apparent to those skilled in the art from the foregoing description. Such modifications are intended to fall within the scope of the appended claims.

[0149] To the extent possible under the respective patent law, all patents, applications, publications, test methods, literature, and other materials cited herein are hereby incorporated by reference.

[0150] The following Examples illustrate the invention described above, but is not, however, intended to limit the scope of the invention in any way. Other test models known as such to the person skilled in the pertinent art can also determine the beneficial effects of the claimed invention. Examples

Example 1. Discovery of antibodies to human RGMb

[0151 ] The Huscl2^TM library (Mabqi, France) was used to discover antibodies specific for human RGMb. Huscl2^TM is a proprietary synthetic human scFv phage display library of 1x1O¹⁰ size (see EP 2134841 A2; US 2009/0143247; US 2010/0167957) based on a single soluble and highly stable human framework known as 13R4 (see Martineu J Mol Biol. (1998) v. 280; Martineu, J Mol Biol (1999) v.292: 921 -929; Philibert BMC Biotech. (2007), v.7), with side-chain diversity incorporated at positions that contribute most to the antigen-binding energy and least to scFv intra-chain contacts (Robin J Mol Biol (2014) v.426). Diversity was introduced in all six CDR regions (Mabqi).

[0152] Monovalent scFvs were displayed on the pill coat protein at the surface of M13 phages containing ampicillin-resistant phagemid vector (scFv fragments comprise a single polypeptide with the VH and VL domains linked by a flexible glycine-serine linker GGGGSGGGGSGGGGS; SEQ ID NO: 15). A schematic drawing of the discovery and selection workflow is shown in Figure 1. Biopanning strategies on recombinant human RGMb-His (huRGMb-His, R&D Systems, cat. no. 3630-RG-050) and mouse RGMb- His (muRGMb-His, R&D Systems, cat. no. 3597-RG-050) were performed to enrich human RGMb- and mouse RGMb-reactive antibodies. After each biopanning, wells were extensively washed, and bound phages were eluted and used to infect E. coli for amplification. The final panning round was performed on cells from the Calu-1 cell line expressing low levels of RGMb. Pools of scFv phages after the last biopanning round were confirmed for their binding to huRGMb and muRGMb. Briefly, huRGMb-His and muRGMb-His (3 pg/ml) proteins were immobilized on Ni-NTA-microplates overnight at 4°C. After blocking with PBS containing 1 % milk for 1 h at RT, scFv phages (10¹⁰ phages/ml) were added for 90 min at RT. Bound scFv phages were detected by anti- M13-HRP (Sino Biological, cat. no.11973 MM05T-H). Four hundred sixty (460) individual E. coli colonies were picked and cultured. The expression of soluble scFvs was induced using 1 mM isopropyl beta-D-1 -thiogalactopyranoside, and culture supernatants containing soluble scFvs were collected to assess their binding to huRGMb by ELISA. Bound scFvs were detected using anti-c-Myc antibody 9E10 with HRP conjugate (Santa Cruz Biotech, cat. no. sc-40 HRP). One hundred ninety-eight (198) clones bound to huRGMb recombinant protein. Sequencing of 198 RGMb- binders revealed 80 scFvs with diverse CDR sequences.

[0153] Antibody discovery and selection workflow and the critical screening assays are outlined in Figure 1 . The selected 80 scFvs were solubly expressed in E. coli and screened for their binding to huRGMb, muRGMb, human RGMa (huRGMa, cat. no. 2459-RG-050, R&D Systems), and human RGMc (huRGMc, cat. no. 3720-RG-050, R&D Systems) by ELISA as described above. Briefly, recombinant proteins were immobilized at a concentration of 2 pg/mL in PBS overnight at 4°C onto Ni-NTA treated Maxisorp 96-well microplates. After blocking with PBS containing 1 % milk for 1 h at RT, pre-diluted scFv supernatants were added for 1 h at RT. RGMb-binding scFvs were detected by anti- c-Myc-HRP antibody. Fifty-eight of 80 scFvs specifically bound huRGMb, muRGMb and did not cross-react with huRGMa and huRGMc. Then, 58 scFvs were screened for their binding to cynomolgus RGMb (cyRGMb, example 11 ). Fifty-three (53) scFvs of 58 bound to huRGMb, muRGMb, cyRGMb and recognized cell-expressed RGMb.

Example 2. Recognition of human RGMb by scFvs and full-chain antibodies

[0154] The selected 53 scFvs were expressed as full-chain human lgG1 antibodies. DNA sequences encoding the respective VH and VL chains were cloned into a human lgG1 expression system encoding gamma heavy and kappa lights chains on separate pcDNA3.4 plasmids. Constructs were sequenced to validate the correct insertion of VH and VL sequences. Plasmids were transiently transfected into ExpiCHO cells, and IgG antibodies were purified using protein-A chromatography. Fifty-one of 53 antibodies were producible in sufficient amounts. Two antibodies were not producible and detectable in the supernatants for unknown reasons.

[0155] To confirm the specificity of the selected anti-RGMb antibodies, ELISA was performed using monomeric and dimeric forms of huRGMb. A dimeric huRGMb was expressed as a rabbit-Fc fusion protein. A DNA construct encoding recombinant soluble RGMb covering amino acids Gly46 to Ser411 of human RGMb was fused with a rabbit Fc fragment and expressed in CHOEBNALT85 cells (Icosagen). Dimeric RGMb was purified using Protein-A chromatography followed by size exclusion chromatography on a HiLoad 26/600 Superdex 200 pg column. The protein corresponding to a dimeric RGMb was collected, and its identity was confirmed by SDS-PAGE (sodium dodecyl-sulfate polyacrylamide gel electrophoresis) and ELISA using a positive control antibody 9D1 by Western blotting using an anti-rabbit-HRP antibody. The binding of the selected clones to monomeric and dimeric huRGMb proteins was tested in both scFv and IgG formats. Briefly, monomeric huRGM-His or dimeric huRGMb-rabbit-Fc were immobilized at a concentration of 2 pg/mL in PBS overnight at 4°C onto Maxisorp 96-well microplates. After blocking with blocking buffer 1 % BSA (Fisher Scientific, cat.no. 10341944), 0.05 % Tween-20 (Fisher Scientific, cat.no. 17146058), 50 pl of scFv supernatant or 1 pg/ml of full chain human lgG1 anti- RGMb antibodies were added in an incubation buffer (0.5 % BSA, 0.05 % Tween-20, PBS pH 7.4 (Fisher Scientific, cat.no. 11503387). ScFvs were detected with an anti-c- Myc-HRP antibody (Santa Cruz Biotech, cat. no. sc-40 HRP), and huRGMb-binding full-chain antibodies were detected by anti-hulgG1 -HRP (Invitrogen cat. no. 17166078). The ELISA was developed with a 3,3',5,5'-tetramethylbenzidine (TMB) substrate solution (Fisher Scientific, cat. no. 34028). The reaction was stopped by adding a stop solution (Thermo Fisher Scientific, cat. no. SS04), and the absorbance was measured at 450 nm/600 nm on a GloMax microplate reader (Promega). All fifty- one anti-RGMb scFvs and full chain hulgG1 antibodies specifically bound both the monomeric and dimeric forms of huRGMb. Exemplary RGMb-binding scFvs, IgG antibodies, and positive control antibody 9D1 are shown in Figures 2A, B and C.

[0156] Anti-RGMb antibody selection workflow is shown in Figure 1 . Twenty-one (21 ) top-performing antibodies out of 51 were selected based on the following properties: i) best binders by ELISA and flow cytometry; ii) high affinity (KD) binders; iii) best RGMb/PD-L2 blockers; iv) lack of cross-reactivity against huRGMa and huRGMc; iv) ability to bind well human, mouse and cynomolgus RGMb proteins. These properties were examined using various assays that are described in the following sections.

[0157] Then variable heavy chain sequences of the 21 selected antibodies were subjected to cluster analysis using Clustal Omega (version 1 .2.4). The BLOSUM matrix method was used to assign scores corresponding to the similarity between the amino acid sequences. Seven antibodies with diverse CDR sequences were selected for further analysis.

Example 3. Antibodies recognize cell surface-expressed human RGMb

[0158] Human RGMb is expressed on dendritic cells, macrophages, neutrophils, naive CD4 and CD8 T cells, epithelial cells, and neural tissues (Xiao et al., loc. cit. Nie et al., Cell & Mol Immunol, 15, 2018; Sekiya et al., Nat Sc Reports, 2019). However, its expression levels in physiological conditions are low; therefore, the above-listed cells cannot be used for flow cytometry-based screening. Hence RGM-overexpressing cells were generated as an antibody screening tool. The cDNA of the RGMb gene (amino acids 46 to 413) inserted in a hygromycin resistance mammalian vector (NM_001012761 .2, Sinobiological cat. no. HG17194-UT) was amplified by RT-PCR and subcloned into the lentiviral vector pRNATin. The construction was transfected into HEK 293T cells using the calcium phosphate method to generate lentiviral particles containing the RGMb genomic information. These lentiviral particles were used to infect the Raji cell line and HEK293 cell line. Single-cell cloning was also performed from RGMb-expressing HEK293 and Raji cell lines to select individual clones stably expressing high levels of RGMb. The clones HEK-293RGMb.1 and Raji-RGMb.1 that expressed high levels of RGMb were used as tools for antibody screening.

[0159] HEK293-RGMb.1 cells and Raji-RGMB.1 cells (2x10⁵ cells/sample, 50 pl) were stained with anti-RGMb antibodies at 1 mg/ml for 40 min at 4°C. FACS binding buffer (Biolegend, cat.no. 420201 ) was used for all cell incubation and washing steps. After washing with FACS buffer, the cells were incubated in the presence of 2.5 pg/ml anti-huIgG-FITC (BioLegend, cat.no. 410720). The cells were then washed twice with cell staining buffer (BioLegend, cat. no 420201 ) at 300g for 3min in RT) and analyzed by CytoFLEX System B3 (Beckman Coulter). All tested anti-RGMb antibodies and positive control antibody 9D1 bound HEK-RGMb.1 cells, and Raji-RGM.1 cells, whilst isotype control hulgG1 antibody and secondary antibody alone did not bind. No binding was observed on non-transduced HEK293 cells. The histogram plots for HEK-RGMb.1 cells stained with anti-RGMb antibodies are shown in Figure 3. Exemplary antibodies 3C09, 3C08, 5B05, 3E07, 5C10, 2C11 and 2G03 specifically stained HEK-RGMb.1 cells.

[0160] Similar results were obtained with anti-RGMb antibodies in scFv format. HEK- RGMb cells were incubated in the presence of soluble scFvs for 40 min at 4°C. After washing, bound scFvs were detected with anti-Myc/c-Myc antibody conjugated to Alexa Fluor® 647 (Santa Cruz, cat. no. SC-40 AF647) diluted 1 :50. Fifty-one anti- RGMb scFvs specifically bound HEK-RGMb cells but not non-transduced HEK293 cells. Thus, selected anti-RGMb antibodies of both scFv and hulgG1 formats bound to cell surface-expressed RGMb.

Example 4. Antibodies recognize the natural form of RGMb

[0161 ] The binding of anti-RGMb antibodies to the natural form of RGMb on tumor cell line MDA-MB-231 (DSMZ, cat. no. ACC732) was determined by flow cytometry. The cells (2x10⁵ cells/sample, 50 ml) were stained with anti-RGMb antibodies at 1 pg/ml for 40 min at 4°C. After washing, the cells were incubated in the presence of 2.5 pg/ml anti-huIgG-FITC (BioLegend, cat. no. 410720) and analyzed by CytoFLEX System B3 (Beckman Coulter). The histogram plots for binding of exemplary antibodies 3C09, 5B05, 5C10, 2C11 and 2G03 and positive control antibody 9D1 to MDA-MB231 cells are shown in Figure 4. The binding fluorescence intensity for 2C11 was 1 ,990, while for 9D1 , 1 ,302 (Figure 4). Thus, the present invention's anti-RGMb antibodies can bind the natural form RGMb on MDA-MB-231 cells, and the anti-RGMb antibody 2C11 bound much stronger than 9D1 antibody.

Example 5. Inhibition of RGMb interaction with PD-L2

[0162] The ability of monoclonal anti-RGMb antibodies to inhibit the interaction of human RGMb to its binding partners was assessed by three different ELISA assay designs. Experiments were performed in PBS pH 7.4 containing 0.5 % BSA and 0.05 % Tween-20 on a shaker at 250 rpm and RT. [0163] The interaction of huRGMb with huPD-L2 and muPD-L2 and the interaction of muRGMb with muPD-L2 have previously been characterized (Xiao et al., J Exp Med (2014), v.211 ; 2014; Nie et al., Cell. & Mol. Immunol. (2018) v.15; US 2020/0095325 A1 ). Notably, the interaction of huRGMb with huPD-L2 is weaker than with muPD-L2. We established three types of binding assays that are outlined in Figure 5.

[0164] First, we assessed anti-RGMb antibodies in the huRGMb/huPD-L2 assay. Dimeric human RGMb protein was expressed as a fusion of extracellular domain (amino acids 46-411 ) with rabbit Fc in CHO cells (Icosagen). The protein huRGMb-rFc was immobilized onto Maxisorp 96-well microplates at 2 pg/ml in PBS overnight at 4°C. Nonspecific binding was blocked using 1 % BSA, 0.05 % Tween-20, PBS pH 7.4. Increasing amounts of anti-RGMb antibodies (from 10 ng/ml to 10 pg/ml) were added and incubated for 1 h, then a constant amount of 600 ng/ml huPD-L2-Fc-biotin (R&D Systems, cat. no. BT1224-050) was added for an additional 1 h. After washing, bound huPD-L2-Fc-biotin was detected by Streptavidin Poly-HRP (ThermoFisher, cat. no. 10571164). Anti-RGMb antibody 9D1 , a rat Ig2a antibody, which is known to inhibit the interaction of muRGMb with muPD-L2 (Xiao et al., J Exp Med (2014) 211 ), and MAB3630, a monoclonal mouse lgG2b antibody, which recognizes human RGMb, but not mouse RGMb (R&D systems, MAB3630-SP), were used as controls. In order to compare with novel anti-RGMb antibodies, 9D1 was expressed as a human lgG1 antibody as described above. DNA sequences encoding the VH and VL chains of 9D1 were cloned into a human IgG 1 expression system encoding gamma heavy and kappa lights chains. Its specificity and the binding activity to RGMBN was verified. The huRGMb/huPD-L2 interaction was inhibited by the antibodies of the present invention 3C09, 3C08, 5B05, 3E07, 5C10, 2C11 and 2G03 (Figure 6). The assay design, however, did not permit the determination of ICso values.

[0165] We then established an huRGMb/muPD-L2 inhibition assay. A recombinant mouse PD-L2-Fc fusion protein (muPD-L2-Fc, R&D Systems, cat. no. 1022-PL-100) was immobilized onto Maxisorp 96-well microplates at 2 pg/mL in PBS overnight at 4°C. Nonspecific binding was blocked using 1 % BSA, 0.05 % Tween-20, PBS pH 7.4. In separate microplates, increasing amounts of anti-RGMb antibodies (from 10 ng/ml to 10 pg/ml) were pre-incubated with a constant amount of 200 ng/ml huRGMb-rFc- biotin for 1 h at RT. The mixture was added to immobilized muPD-L2 and incubated for an additional 1 h at RT. The bound huRGMb-rFc-biotin was quantified by Streptavidin Poly-HRP (ThermoFisher, cat.no. 10571164). The ELISA was developed with a TMB substrate solution (Fisher Scientific, cat. no. 34028). The reaction was stopped by adding a stop solution (Thermo Fisher Scientific, cat. no. SS04), and the absorbance was measured at 450 nm/600 nm on a GloMax microplate reader (Promega). Inhibition curves were plotted, and the ICso values were calculated using GraphPad Prism® software. The ICso indicates the concentration of an antibody that blocks 50 % of RGMb binding to its ligand. The results and ICso values are shown in Figure 6. Anti-RGMb clone 9D1 was used as a positive control. The inhibitory potencies of antibodies 3C09, 5B05, 3E07, and 2C11 were higher (1.20 nM, 0.47 nM, 1.2 nM, 0.33 nM, respectively) than that of clone 9D1 (3.47 nM). MAB3630 inhibited the huRGMb/muPD-L2 with an ICso of 1 .0 nM. Markedly, antibodies 3C08, 5C10 and 2G03 blocked the huRGMb/muPD-L2 with superior potencies, 130 pM, 70 pM and 130 pM, respectively (Figure 7).

[0166] Thus, anti-RGMb antibodies of the present invention (3C08, 5C10, 2G03, 3C09, 5B05, 3E07 and 2C11 ) block RGMb/PD-L2 interaction with higher potency than 9D1 . The lead anti-RGMb 2C11 was 10.5-fold superior to the 9D1 antibody (Figure 7, lower panel).

[0167] As anti-RGMb antibodies of the present invention bind muRGMb, their ability to block the muRGMb/muPD-L2 interaction was assessed. We established an muRGMb/muPD-L2 inhibition assay. A recombinant mouse RGMb-His (R&D Systems, cat. no. 3597-RG-050) was immobilized onto Maxisorp 96-well microplates at 2 pg/mL in PBS overnight at 4°C. Nonspecific binding was blocked as described above and 10 pg/ml anti-RGMb antibodies were added and incubated for 1 h at RT. After washing, the constant concentration of biotinylated muPD-L2-His (Sinobiologicals, cat. no. 50804-M08H-B) was added. The bound muPD-L2-biotin was quantified by Streptavidin Poly-HRP. The results are shown in Figure 8. Anti-RGMb clone 9D1 was used as a positive control. Thus, the seven anti-RGMb antibodies of the present invention are capable of blocking the interaction of muRGMb with muPD-L2. These results would also permit the use of syngeneic mouse disease models to assess the in vivo efficacy of antibodies of the present invention.

[0168] Thus, inhibition of the RGMb/PD-L2 interaction by anti-RGMb antibodies of the present invention was demonstrated by three different assays. The inhibitory potency (IC50) for anti-RGMb antibodies of the present invention (3C08, 5C10, 2G03 C09, 5B05, 3E07 and 2C11 ) was higher than that of the 9D1 antibody.

Example 6. Antibody binding to RGMb determined by Interferometry & FACS

[0169] Bio-Layer Interferometry (BLI, Octet Red96) was used to determine the binding characteristics of full-chain (hulgG1 ) anti-RGMb antibodies. Anti-RGMb antibodies were captured by anti-human Fab biosensors (FAB2G) (Sartorius, cat.no.18-5125) and incubated with different concentrations of analyte huRGMb-rabbit Fc. The 1 :2 fit indicated affinity (KD) values in bivalent format, and the kinetic rates are shown in Table 2. Affinities (KD values) of anti-RGMb antibodies, except for 3C08 (25.2 nM), spanned from 0.724 nM to 8.15 nM. The clone 9D1 bound huRGMb with a lower affinity (20.5 nM) than did novel anti-RGMb antibodies (Table 2).

Table 2. Affinity constants and kinetic rates determined from BLI experiments.

In the subsequent studies, RGMb-overexpressing HEK293 cells were used to compare antibody binding strength using flow cytometry of live cells. Increasing amounts of anti- RGMb antibodies (100 ng/ml to 30 pg/ml) were incubated with a constant amount (2x10⁵ cells/sample) of HEK-huRGMb.1 cells for 40 min at 4°C. After washing, the cells were incubated in the presence anti-huIgG-FITC (clone M1310G05, Biolegend, cat. no. 410720) and analyzed by CytoFLEX System B3 (Beckman Coulter). The geometric mean fluorescence values were plotted, and the ECso values (or apparent affinities) were calculated using GraphPad Prism® software. The ECso indicates the concentration of an antibody that gives half-maximal binding relative to the point at which the antibody shows saturation behavior. The cell-binding ECso values for anti- RGMb, except for 3C08 (56.63 nM), ranged from 2.92 nM to 14.17 nM (Table 3). The cell-binding ECso value for the 9D1 control antibody was greater, 39.57 nM. These results are in agreement with the BLI-based KD measurements and confirm that anti- RGMb antibodies 3C09, 5B05, 3E07, 5C10, 2C11 and 2G03 bind RGMb with a superior affinity than that of 9D1 . The lead anti-RGMb clone 2C11 showed a superior binding affinity in both BLI-based (KD~1 .40 nM) and cell-binding measurements (ECso~2.92 nM) compared to antibody 9D1 (KD~20.50 nM and ECso~39.57 nM respectively).

Table 3. Anti-RGMb binding affinity determined by flow cytometry.

Example 7. Inhibition of RGMb interaction with neogenin

[0170] In addition to PD-L2, RGMb interacts with several proteins, including neogenin, BMP2, and BMP4. Antibodies blocking RGMb/PD-L2 interaction were assessed for their impact on the interaction of RGMb with neogenin, BMP2 and BMP4 by ELISA assays. Experiments were performed in PBS pH 7.4 containing 0.5 % BSA, 0.05 % Tween-20 on a shaker at 250 rpm and at RT.

[0171 ] In an experimental setup A, a recombinant human neogenin 6-His tag (Neogenin-His; R&D Systems, cat.no. 8607-NE-050) was immobilized onto Maxisorp 96-well microplates at a 2 pg/ml concentration in PBS overnight at 4°C. Nonspecific binding was blocked using 1 % BSA, 0.05 % Tween-20, PBS pH 7.4. In separate microplates, a constant amount of 75 ng/ml huRGMb-rFc was incubated with increasing amounts of anti-RGMb antibodies (10 ng/ml to 10 pg/ml) for 1 h at RT. The mixture was added to wells and incubated for another 1 h at RT. After washing, the bound huRGMb-rFc was detected by using anti-rabbit-Fc-HRP (Genescript, cat. no. A01856). In an experimental setup B, a recombinant human neogenin fusion with human Fc fragment (Neogenin-Fc; R&D Systems, cat. no. 9699-NE-050) was immobilized onto Maxisorp 96-well microplates followed by the addition of the preincubated mixture of huRGMb-rFc and anti-RGMb antibodies (from 10 ng/ml to 10 pg/ml). An anti-RGMb clone 9D1 and MAB3630 were used as positive controls. RGMb binding to Neogenin-His was completely abolished in the presence of 5C10, 2G03 and 3C08 antibodies, while 9D1 and MB3630 antibodies showed only a partial blocking effect (Figure 9A). The inhibitory potencies were 0.42 nM, 0.42 nM and 0.58 nM, respectively, for the experimental setup A. Similar results were obtained when the dimeric neogenin-Fc was used to bind huRGMb; 5C10, 2G03 and 3C08 antibodies completely inhibited Neogenin/RGMb interaction with ICso values of 1.46 nM, 1.88 nM and 1.43 nM, respectively (Figure 9B). Antibodies 3C09, 5B05, 2C11 and 3E07 had only a weak effect on RGMb interaction with neogenin. Based on these results RGMb/PD-L2 blocking antibodies can be divided into two groups; group 1 antibodies (5C10, 2G03, 3C08) can completely block the RGMb/neogenin, whilst group 2 antibodies (3C09, 5B05, 3E07, 2C11 ) show either weak or no impact on the RGMb/neogenin interaction. These and the results of the previous sections suggest that PD-L2 and Neogenin binding sites on RGMb are either shared or partially overlapping.

Example 8. Inhibition of RGMb interaction with BMP2 and BMP4

[0172] We investigated whether anti-RGMb antibodies interfere with the interaction of RGMb and BMP2 and BMP4. HuRGMb-His (R&D Systems, cat. no. 3630-RG-050) was immobilized onto Maxisorp 96-well microplates at 2 pg/mL in PBS overnight at 4°C. Nonspecific binding was blocked using 1 % BSA, 0.05 % Tween-20, PBS pH 7.4. Increasing amounts of anti-RGMb antibodies (10 ng/ml to 10 pg/ml) were added and incubated for 1 h, followed by the addition of a constant amount of 10 ng/ml huBMP4 (R&D Systems, cat. no. 314-BP-010) for an additional 1 h. After washing, bound huBMP4 was detected by 1 pg/ml anti-BMP4-biotin (R&D Systems, cat. no. BAM7572) and SAV-HRP at 1 :5000 dilution (ThermoFisher, cat. no. 10571164). Anti-RGMb clone 9D1 and MAB3630 were used as positive controls. Interestingly, antibodies 5B05, 3C09 and 2C11 potently blocked RGMb/BMP4 interaction (Figure 10A). In contrast, antibodies 2G03 and 3C08 did not influence the binding of RGMb to BMP4, similar to MAB3630. 3E07 antibody behaved as the 9D1 control, /. e. partially blocked RGMb/BMP4 interaction. The biological significance of such a weak inhibition effect remains to be investigated. The anti-RGMb clone 2C1 1 inhibited RGMb/BMP4 with an ICso value of 6.5 nM, while the reference 9D1 antibody showed an IC50 of 21.0 nM. Hence, antibodies 5B05, 3C09 and 2C11 were found to be potent RGMb/BMP4 blockers.

[0173] Anti-RGMb antibodies were tested in a similar RGMb/BMP2 inhibition assay. HuRGMb-His was immobilized and incubated in the presence of 10 pg/ml antibodies. After washing, 8 ng/ml huBMP2 (R&D Systems, cat. no. 355-BM-010CF) was added and followed by the addition of 0.1 pg/ml anti-BPM2-biotin (Peprotech, cat. no. 500P195BT) and SAV-HRP. Antibodies 5B05 and 2C11 almost completely blocked the RGMb/BMP2 interaction, while 3C09 and 9D1 partially blocked the RGMb/BMP2 interaction (Figure 10B). Thus, anti-RGMb antibodies 5B05, 2C11 and 3C09 equally well block both BMP2 and BMP4 interactions with RGMb. The anti-RGMb 2C11 inhibited RGMb/BMP2 by 91 % while the 9D1 clone by 44 % compared to isotype control (Figure 10B).

[0174] Taken together, the above results demonstrate that anti-RGMb 2C11 is superior in inhibiting both RGMb/BMP2 and RGMb/BMP4 interactions. These results also indicate that BMP2 and BMP4 interact with the same site of RGMb. Group 1 antibodies 5C10, 2G03 and 3C08 (RGMB/PD-L2 and RGMb/neogenin blockers) as defined above, did not block RGMb/BMP2 and RGMb/BMP4 interactions. Group 2 antibodies are either complete (3C09 and 2C11 ) or partial (3C09 and 3E07) RGMb/BMP2/4 blockers. These and the results of the previous sections suggest that PD-L2- and neogenin-binding sites on RGMb are either shared or partially overlapping.

Example 9. Two anti-huRGMb antibodies compete with 9D1 antibody

[0175] We next investigated whether antibodies of the present invention compete with 9D1 antibody (anti-mouse-RGMb). As previously reported (Xiao et al., 2014), 9D1 antibody significantly block muRGMb interactions with muBMP2/muBMP4 and partially block muRGMb interaction with mouse neogenin.

[0176] HuRGMb-His was immobilized onto microplate wells at 2 pg/ml. After blocking and washing, increasing amounts of RGMb inhibitory antibodies (30 ng/ml to 10 pg/ml) were added and incubated for 1 h at RT. Ten (10) ng/ml biotinylated 9D1 was added and the bound 9D1 -biotin was detected using Streptavidin-HRP. As expected, unlabelled 9D1 inhibited the binding of 9D1 -biotin to RGMb (/. e. compete with 9D1 - biotin) (Figure 11 ). Antibodies 2C11 and 5B05 fully competed with 9D1 -biotin. In contrast, antibody 3E07 only partially competed with 9D1 -biotin while antibodies 3C09, 3C08, 5C10 and 2G03 did not influence the binding of 9D1 to RGMb (Figure 11 ). These results indicate that epitopes of 9D1 and 3E07 are in close distance. 3C09, 3C08, 5C10 and 2G03 antibodies recognize different regions of the RGMb protein than the epitope of 9D1 . The fact that 9D1 blocks both RGMb/Neogenin and RGMb/BMP2/4 interaction and competes with 2C11 and 5B05 antibodies (RGMb/BMP2/4 blockers) may indicate that BMP2/BMP4 binding sites in human and mouse RGMb proteins differ. Additional competition experiments confirmed that antibodies 2C11 and 5B05 bind an epitope 1 and do not compete with 2G03, 3C08 and 5C10 antibodies. The latter antibodies bind epitope 2. These data allow us to schematically draw anticipated epitopes of anti- RGMb antibodies of the present invention in relation to the 9D1 epitope and BMP2/4- neogenin- and PD-L2-binding sites, as shown with dotted lines in Figure 12.

Example 10. Stability of RGMb/PD-L2 blocking antibodies

[0177] The stability of anti-RGMb antibodies of the invention was evaluated based on protein aggregation. Antibodies were expressed as full-chain mouse lgG2a antibodies with Leu234Ala/Leu235Ala (LALA) mutations. DNA sequences encoding VH and VL chains were cloned into a mouse lgG2a expression system encoding gamma heavy and lambda lights chains on separate pQMCF-LC- and pLIC-mlgG2a-LALA plasmids. Coding regions were verified by nucleotide sequencing. The plasmids were chemically transiently transfected into CHOEBNALT85-1 E9 cells (Icosagen). IgG antibodies were affinity purified using HiTrap MabSelect column, followed by size-exclusion chromatography using HiLoad26/60 Superdex 200 column and sterile filtration. Purified antibodies were quality controlled by SDS-PAGE electrophoresis, and ELISA against huRGMb (R&D Systems, cat.no. 3630-RG-050).

[0178] Antibodies were formulated in PBS pH 7.4 at 2 mg/ml and exposed to three freeze/thaw cycles in liquid nitrogen and thawed at RT. Visual inspection showed no detectable protein precipitation. Analytical HPLC-SEC was performed on a Waters BioSuite 250 4 pm UHR SEC 4.6x300 mm. The results demonstrated that after three freeze/thaw cycles, 5B05 and 3E07 antibodies were 95 % monomeric, 3C09 was 97.25 % monomeric, while 5C10, 3C08, 2C11 , and 2G03 antibodies were above 99 % monomeric.

[0179] The stability of the 2C11 antibody in human plasma and PBS was assessed by testing its binding properties to huRGMb-rFc coated protein after pre-incubation in human plasma or PBS for up to 24 h. Briefly, the 2C11 antibody or an isotype control antibody were incubated at 50 pg/ml in human plasma (Zenbio, cat. no 088SER- PLE10ML-EDTA, 1 male donor) or PBS pH 7.4 at 37°C. Samples were collected at 0 h, 2 h, 6 h, 8 h and 24 h of incubation and added at 5 pg/ml for 1 h to huRGMb-rFc protein previously immobilized onto Maxisorp 96-well microplates at 2 pg/mL in PBS overnight at 4°C. After washing, the bound 2C11 antibody was detected using an anti- kappa-HRP antibody (Jackson ImmunoResearch, cat. no 115-035-071 ). The ELISA was developed with a TMB substrate solution (Fisher Scientific, cat. no. 34028). The reaction was stopped by adding a stop solution (Thermo Fisher Scientific, cat. no. SS04), and the absorbance was measured at 450 nm/600 nm on a GloMax microplate reader (Promega). The binding properties of the 2C11 were preserved up to the 24 h incubation time in human plasma and PBS (Figure 13), reflecting antibody stability in human plasma. Additional in vitro assays were performed to assess the stability of the full human 2C11 -hulgG1 -LALA antibody produced in CHO cells (Icosagen, concentration 2 mg/ml) in different stress conditions (Charles River Laboratories) using the UNchained Labs UNcle instrument with incorporation of a DLS module, the ThermoFisher Scientific Vanquish™ HPLC system and TSKgel G3000 SWXL SEC column. Hence, thermal stability was assessed by exposure of the antibody to 45°C for 1 , 3, 5, 8, and 15 days. Chemical stability was assessed through different tests; for pH stress, the 2C11 antibody was subjected to buffer exchange: 100 mM glycine HCI pH 2.0, 50 mM sodium acetate pH 5.5 (acidic pH stress) or 20 mM Tris pH 9.0 (Alkaline pH stress). For oxidative stress test, the antibody was exposed to H2O2 (at 0.02, 0.01 or 0.005 %), to TBHP (at 0.05, 0.3 or 0.7 %) or to AAPH (at 1 , 3, or 5 mM). Shear stress conditions were performed by exposing the 2C11 antibody to rapid shaking at 250 rpm at 30°C for 1 , 3, 5, 8 and 15 days. An oligomerization was observed for the AAPH oxidative stress test, and a peak shift was noticed in the shear stress test. Thus, anti-RGMb 2C11 antibody is stable at physiological conditions and minor instabilities were observed in extremely harsh shear stress and oxidative stress conditions.

Example 11. Species cross-reactivity & selectivity of RGMb/PD-L2 blocking antibodies.

[0180] ELISA was used to measure the relative binding of anti-RGMb antibodies to ortholog proteins from different species. Human RGMb-His (cat.no. 3630-RG-050) and mouse RGMb-His (cat.no. 3597-RG-050) were purchased from R&D systems.

[0181 ] Cynomolgus RGMb (cyRGMb) was expressed with a polyhistidine tag at Icosagen. In brief, a DNA construct encoding recombinant soluble cynomolgus RGMb covering amino acids Gly46 to Ser411 of the cynomolgus RGMb was fused with His- tag and transiently expressed in CHOEBNALT-85-1 E9 cells (Icosagen). CyRGMb was purified using HisTrap chromatography followed by size exclusion chromatography on a HiLoad 26/600 Superdex 200 pg column and sterile filtration. The protein corresponding to a monomeric cyRGMb was collected, and its identity was confirmed by SDS-PAGE and by western blotting using an anti-His antibody (GenScript, cat.no. A00186-100).

[0182] For the ELISA binding assay, proteins were immobilized at a concentration of 2 pg/mL in PBS overnight at 4°C onto Maxisorp 96-well microplates. All the subsequent steps were performed as described above. The bound antibodies were detected using anti-huIgG-HRP and developed with a TMB substrate solution. The reaction was stopped by the addition of a stop solution and the absorbance was measured on a GloMax microplate reader. All anti-RGMb antibodies bound equally well human, mouse and cynomolgus RGMb proteins (Figure 14). These data permit performing in vivo efficacy and safety studies in mice and non-human primates.

[0183] Selectivity of anti-RGMb antibodies was assessed against human RGMa (huRGMa, R&D Systems, cat. no. 2459-RM-050) and human RGMc (huRGMc, R&D Systems, cat.no. 3720-RG-050). The paralog proteins share about 50 % amino acid identity and several structural motifs (P. Rotwein, Physiological Reports (2019) v.7). HuRGMa, huRGMb and huRGMc proteins were immobilized at a concentration of 2 pg/mL in PBS overnight at 4°C onto Maxisorp 96-well microplates. The subsequent steps were performed as described above, and the results we measured on a GloMax microplate reader. All anti-RGMb antibodies of the present invention specifically bound huRGMb and did not bind to huRGMa and huRGMc (Figure 15).

Example 12. In vivo determination of the maximum tolerated dose (MTD) and pharmacokinetic (PK) properties of the 2C11 antibody.

[0184] For the MTD study, C57BL/6 mice (Charles River) aged 5-8 weeks were housed in IVC cages (up to 5 per cage) with individual mice identified by the tail mark. All animals were allowed free access to a standard certified commercial diet and sanitized water during the study. The holding room was maintained under standard conditions: 18-24°C, 55-70 % humidity, and a 12 h light/dark cycle. All protocols have been approved by the Axis Bio Animal Welfare and Ethical Review Committee, and all procedures were carried out under the guidelines of the Animal (Scientific Procedures) Act 1986. Animals were randomized and assigned to treatment groups with either 100 pg, 200 pg, or 400 pg of 2C11 -mlgG2a antibody diluted in sterile PBS pH 7.4. The dosing volume was 200 pl per dose injected intraperitoneally and there were four dosing occasions on days 1 , 4, 7, and 10. Health and body weight were monitored daily during the study. Treatment with the 2C11 anti-RGMb antibody at 100, 200, and 400 pg was well tolerated over the 14-day study period (Figure 16A). The average body weight of these mice at completion of the study was 106.2 %, 107.1 % and 100.3 % (respectively) of their initial weight. A slight dip in average body weight was evident following administration of the 400 pg dose on day 10 but this was recovered quickly. There were no signs of pain or distress in the mice, and no mice had to be terminated early from the study due to welfare concerns.

[0185] For the PK study, male CD-1 mice (Charles River) aged 5-6 weeks, weighing approximately 30-40 g were housed in the conditions stated above. All protocols used in this study have been approved and procedures were carried out under the mentioned guidelines. Mice were assigned to treatment groups in parallel with 10 mg/kg of 2C11 or the control antibody 9D1 , intravenously. Animals were bled according to a composite PK sampling design. Hence, animals divided into three subgroups (n = 3) were randomly assigned to the following time points: 5 min, 8 h, and 5 d for the first group of mice; 1 h, 24 h, and 7 d for the second group; 4 h and 72 h for the third group, for a maximum of three blood collections per mouse. At the relevant timepoints, approximately 120 pl whole blood was collected from the lateral tail vein, via the opposite vein used for IV drug administration. Blood samples were placed in ice-cold K2-EDTA coated microvette tubes, gently mixed with the anticoagulant coating, and immediately placed on wet ice. The blood samples were transferred to the lab on wet ice for centrifugation, within 30 min of collection. Samples were centrifuged at 3,000 ref for 5 min at 4°C, and the entire resultant plasma was transferred to a chilled labelled Eppendorf tube. Samples were immediately placed in a -80°C freezer and stored at - 80°C prior to ELISA analysis. A human lgG1 ELISA kit (Abeam, ab100548) was used to determine the quantity of each compound in the plasma samples. Eight total representative plasma samples were initially inputted into the ELISA, at dilution factors of 1 :3,000, 1 :4,000, and 1 :5,000. A dilution factor of 1 :3,000 was selected as optimal for the analysis of all samples. On the day of the assay, plasma samples were thawed on wet ice, and prepared for the assay by a two-step dilution in ELISA assay diluent. In terms of body weight, both treatments appeared to be generally well tolerated, with the average body weight of animals remaining largely similar to pre-treatment levels. No symptoms of acute toxicity or adverse responses to treatment were noted in animals during the 3-to-7-day observation periods after dosing. The composite PK profiles for each compound are shown in Figure 16B. A quantifiable mean IgG signal was obtained within the dynamic range of the ELISA at all time points, for both test compounds. The 2C11 antibody displayed a distinct PK profile compared to 9D1 , broadly characterized by larger a CMAX and AUC values reflecting a much higher exposure of the 2C11 antibody compared to the control 9D1 antibody.

Example 13. In vivo activity of anti-RGMb 2C11 antibody combined with anti-PD- L1 antibody.

[0186] The B16 OVA syngeneic tumor model was used to evaluate the in vivo efficacy of the 2C11 anti-RGMb antibody in combination with an anti-PD-L1 antibody (WuXi AppTec). Height weeks female C57BL/6 mice (Shanghai Lingchang Biological Technology Co., Ltd.) were used for the study. The mice, individually marked with ear tag, were kept in individual ventilation cages at constant temperature (20-26°C) and humidity (40-70 %) with 4 animals in each cage at a 12 h light/dark cycle. Animals had free access to irradiation sterilized dry granule food and sterile drinking water during the entire study period. All the procedures related to animal handling, care and the treatment in the study were performed according to the guidelines approved by the Institutional Animal Care and Use Committee (IACUC) of WuXi AppTec following the guidance of the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). The B16F10-OVA cells (B16F10: ATCC-CRL-6475) were generated by transduction of OVA coding sequence into B16F10 cell through lentiviral system (WuXi AppTec). The B16F10-OVA cells were maintained in vitro as a monolayer culture and each mouse was inoculated subcutaneously at the right upper flank with 0.5x10⁶ B16F10-OVA tumor cells in 0.1 mL of PBS. Antibody treatments were started on day 8 after tumor inoculation when the average tumor size reached approximately 50 mm³. The animals were randomized and assigned into groups. Each group consisted of 8 tumor-bearing mice. The test antibody anti-RGMb 2C11 mulgG2a- LALA, anti-muPD-L1 (clone 10F.9G2, BioXcell) antibody and an isotype control (Rat lgG2b, BioXcell) were administrated to the mice on days 8, 11 , 14 and 17 post implantation.

[0187] The animals were daily checked for any effects of tumor growth and treatments on normal behavior such as mobility, food and water consumption, eye/hair matting and any other abnormal effects. The body weight gain/loss were measured three times weekly. Death and observed clinical signs were recorded on the basis of the numbers of animals within each group. The major endpoint was to see if the tumor growth could be delayed, or mice could be cured. Tumor size was measured three times weekly in two dimensions using a caliper, and the volume was expressed in mm³ using the formula: V = 0.5 x a x b². Individual animals were euthanized immediately when its tumor volume reached 2,000 mm³ or diameter exceeded 2 cm in size in any direction.

[0188] The tumor size was then used for calculations of both T/C and TGI values. The T/C value (in percent) is an indication of antitumor effectiveness; T and C are the mean volumes of the treated and control groups, respectively, on a given day. TGI was calculated for each group using the formula: TGI (%) = [1 -(Ti-To)/ (Vi-Vo)] x 100; Ti is the average tumor volume of a treatment group on a given day, To is the average tumor volume of the treatment group on the day of treatment start, is the average tumor volume of the vehicle control group on the same day with Ti, and Vo is the average tumor volume of the vehicle control group on the day of treatment start. Comparison between groups were carried out with two-way ANOVA with Bonferroni’s multiple comparisons test. All data were analyzed using GraphPad Prism 8.0, p < 0.05 was considered to be statistically significant. [0189] Compared with anti-mPD-L1 -single treatment group, no weight loss and no significant change in the anti-mPD-L1-2C11 combo treatment group was observed.

[0190] The tumor growth results are shown in Figure 17A and Table 4. Treating mice with anti-RGMb 2C11 in combination with an anti-PD-L1 significantly reduced tumor growth as compared to anti-PD-L1 alone and isotype control groups.

Table 4: Tumor growth after treatment with anti-RGMb 2C11 antibody and anti- PD-L1 antibody

Tumor Size (mm³) T/C TGI%

Treatment at day 10 p value

(%) (%) (Mean +/- SEM)

Isotype Control 1013 ± 289 anti-mPD-L1 950 ± 406 93.8 6.5 >0.9999

2C11+ anti-mPD-L1 , 262 ± 166 25.9 78.1 0.0050

[0191] These results suggest that blocking RGMb/PD-L2 and PD-L1/PD-1 interactions with anti-RGMb 2C11 and anti-muPD-L1 enhances anti-tumor responses.

[0192] The survival of the control group mice (isotype control) ranged from 10 to 20 days with an MST (median survival time) of 15 days. The combination of 2C11 and anti-mPD-L1 increased survival of animals, and resulted in an MST value of 23 days (p = 0.0025) when compared with the control group (Figure 17B). Thus, anti-RGMb 2C11 in combination with anti-mPD-L1 showed significant antitumor activity and increased survival against the B16F10-OVA tumor.

Example 14. Design and expression of anti-PD-1xanti-RGMb and anti-PD- L1xanti-RGMb and biparatopic bispecific antibodies

[0193] Based on the findings that anti-RGMb promoted anti-tumor response with anti- PD-L1 treatment, we designed bispecific antibodies simultaneously targeting RGMb and PD-1/PD-L1 pathways. Asymmetric and symmetric bispecific antibodies (bsAbs) were designed, Fab-scFv-Fc format (bivalent, Figure 18A, left) with single binding elements for PD-1/L1 and RGMb and (Fab)2-scFv2-Fc format (tetravalent, Figure 18A, right) with two binding elements for each PD-1/L1 and RGMb.

[0194] The VH (SEQ ID NO: 7) and VL (SEQ ID NO: 8) chains were fused onto full- length mouse lgG2a or human lgG1 scaffolds. The Leu234Ala/Leu235Ala (LALA) mutations were introduced into all bsAbs. Knob-into-holes (KiH) mutations were introduced in the Fc fragment of asymmetric bivalent constructs to promote heterodimerization. Anti-RGMb scFvs were fused by using a flexible GGGGSGGGGSGGGGS linker; (SEQ ID NO: 15).

[0195] Two anti-human-PD-1 antibodies and two anti-human PD-L1 antibodies were used to build bsAbs (Table 5). Anti-human PD-1 EH12 VH (SEQ ID NO: 18) and VL chains (SEQ ID NO: 19), and anti-human PD-1 NIV VH (SEQ ID NO: 20) and VL chains (SEQ ID NO: 21 ) were used to build anti-huPD-1xanti-RGMb bsAbs.

[0196] Anti-human PD-L1 Ab23 VH (SEQ ID NO: 22) and VL chains (SEQ ID NO: 23), and anti-human PD-L1 Atezolizumab VH (SEQ ID NO: 24) and VL chains (SEQ ID NO: 25) were used to build anti-huPD-L1xanti-RGMb bsAbs (Table 5).

Table 5. Anti-huPD-1 and anti-huPD-L1 antibodies used to design bsAbs (CDR residues shown in bold/underlined are defined according to the Kabat system).

[0197] BsAbs were transiently expressed in CHO-Express suspension cells (GenScript) and purified with two-step purification procedures; protein A chromatography (MabSelect PrismA) followed by size exclusion SEC-HPLC (HiLoad 26/600 Superdex 200pg) chromatography. Tetravalent bsAbs showed higher production yield. Furthermore, tetravalent bsAbs were easily purifiable by chromatography methods. The resulting bsAbs specifically bound the respective huPD-1 , huPD-L1 and huRGMb proteins in ELISA assays. Flow cytometry assays confirmed the binding of bsAbs to cells expressing huPD-1 , huPD-L1 and huRGMb. The results of the above analyses are summarized in Table 6.

Table 6. Properties of anti-huPD-1xanti-RGMb and anti-huPD-L1/anti-RGMb

BsAbs.

" indicates that the binding and blocking properties are confirmed; indicates no blocking observed.

[0198] Most bivalent and tetravalent anti-huPD-1xanti-RGMb bsAbs and anti-huPD-

L1xanti-RGMb bsAbs were capable of blocking RGMb/PD-L2 and PD-1/PD-L1 interactions (Table 6) Exemplary bsAbs composed of anti-huPD-1 EH12 and anti- RGMb 2C11 are shown in Figure 18B. For the RGMb/PD-L2 blocking assay, increasing amounts of bispecific 2C11/EH12 bivalent and tetravalent antibodies were preincubated with huRGMb-rabbit-Fc-biotin and added to immobilized muPD-L2-huFc protein. The bound huRGMb-rFc-biotin was detected by using streptavidin-poly-HRP. For the PD-1/PD-L1 blocking assay, increasing amounts of bispecific 2C11/EH12 bivalent and tetravalent antibodies were added to immobilized huPD-1 -huFc protein. HuPD-L1 -huFc-Biotin was added, and the bound protein was detected by using streptavidin-HRP. The results demonstrate that both bivalent and tetravalent formats of the 2C11/EH12 BsAbs are capable of blocking both RGMb/PD-L2 and PD-1/PD-L1 interactions. The unexpected finding was that although 2C11xanti-PD-L1 ATZ specifically bound RGMb in ELISA and FACS assays, it did not block the RGMb/PD- L2. Moreover, this particular bsAb did not block RGMb/BMP2 interaction. Hence, when 2C11 fused with ATZ antibody, the 2C11 arm of this bsAb preserved RGMb-binding activity; however, it lost RGMb/PDL2 and RGMb/BMP2 blocking capacity. These data suggest that anti-RGMb 2C11 is suitable for generating bi- or multi-specific antibodies with some but not all IgG antibodies.

Biparatopic antigen targeting by antibodies may be more advantageous than targeting a single epitope. Thus, we constructed biparatopic bsAbs composed of VHs and VLs of anti-RGMb 2C11 and anti-RGMb 2G03, binding on the one hand to N-terminal epitope 1 and on the other hand to C-terminal epitope 2 of RGMb illustrated in Figure 12. Bivalent and tetravalent 2C11x2G03 bsAb constructs were designed as shown in Figure 18A. The VH (SEQ ID NO: 7), VL (SEQ ID NO: 8), VH (SEQ ID NO:3) and VL (SEQ ID NO: 4) chains were fused onto a full-length human lgG1 scaffold. The Leu234Ala/Leu235Ala (LALA) mutations were introduced into both bivalent and tetravalent constructs. Knob-into-holes (KiH) mutations were introduced in the Fc fragment of an asymmetric bivalent construct to promote heterodimerization. Anti- RGMb scFvs were fused by using a flexible GGGGSGGGGSGGGGS linker; (SEQ ID NO: 15). The bsAb constructs were expressed in CHO cells and purified using protein A chromatography followed by ion exchange chromatography. The biparatopic 2C11x2G03 bsAbs exhibited monomeric profiles in HPLC-SEC throughout the production and purification phases. The purity of both bivalent and tetravalent bsAbs analyzed by HPLC-SEC and SDS-PAGE was above 99%. Both biparatopic 2C11x2G03 bsAb specifically bound huRGMb protein in ELISA assays. These results indicate that the amino acid sequences of the clones 2C11 and 2GO3 are suitable for engineering stable and monomeric biparatopic bsAbs.

As described in the previous sections, monospecific 2C11 and 2G03 block RGMb/PD- L2 interaction. Additionally, 2C11 preferentially blocks RGMb/BMP2/4 interaction, while 2G03 preferentially blocks RGMb/neogenin interaction. Consequently, biparatopic 2C11x2G03 bsAbs possess the functions of 2C11 and 2G03; inhibit all three interactions of RGMb with PD-L2, BMP2/4 and neogenin. Such complete blockade of interactions RGMb with its partners may be beneficial in some pathological conditions.

References:

1. Lewkowich, I. P., Lajoie, S., Staffers, S. L., Suzuki, Y., Richgels, P. K., Dienger, K., Sproles, A. A., Yagita, H., Hamid, Q., & Wills-Karp, M. (2013). PD-L2 modulates asthma seventy by directly decreasing dendritic cell IL-12 production. Mucosal Immunology, 6(4), 728-739. https://doi.org/10.1038/mi.2012.111

Nie, X., Chen, W., Zhu, Y., Huang, B., Yu, W., Wu, Z., Guo, S., Zhu, Y., Luo, L., Wang, S., & Chen, L. (2018). B7-DC (PD-L2) costimulation of CD4+ T-helper 1 response via RGMb. Cellular and Molecular Immunology, 15( 0), 888-897. https://doi.Org/10.1038/cm i .2017.17

Bell, C. H., Healey, E., Erp, S. Van, Bishop, B., Gilbert, R. J. C., Aricescu, A. R., & Pasterkamp, R. J. (2013). Structure of the Repulsive Guidance Molecule (RGM)- Neogenin Signaling Hub. Science. https://doi.Org/10.1126/science.1232322. Structure

Yu, S., Leung, K. M., Kim, H.-J., & DeKruyff, R. H. (2019). Blockade of Repulsive guidance molecule b (RGMb) inhibits allergen-induced airways disease. Journal of Allergy and Clinical Immunology. https://doi.Org/10.1016/j . jaci .2018.12.1022. Blockade

Healey, E. G., Bishop, B., Elegheert, J., Bell, C. H., Padilla-Parra, S., & Siebold, C. (2015). Repulsive guidance molecule is a structural bridge between neogenin and bone morphogenetic protein. Nature Structural and Molecular Biology, 22(6), 458-465. https://d0i.0rg/l 0.1038/nsmb.3016

Li, J., Ye, L., Kynaston, H. G., & Jiang, W. G. (2012). Repulsive guidance molecules, novel bone morphogenetic protein co-receptors, are key regulators of the growth and aggressiveness of prostate cancer cells. International Journal of Oncology, 40(2), 544-550. https://doi.org/10.3892/ijo.2011.1251

Li, P., Li, J., Yang, R., Zhang, F., Wang, H., Chu, H., Lu, Y., Dun, S., Wang, Y., Zang, W., Du, Y., Chen, X., Zhao, G., & Zhang, G. (2015). Study on expression of IncRNA RGMB-AS1 and repulsive guidance molecule b in non-small cell lung cancer. Diagnostic Pathology, 10( ), 1-7. https://doi.org/10.1186/s13000-015- 0297-x

Li, J., Ye, L., Shi, X., Chen, J., Feng, F., Chen, Y., Xiao, Y., Shen, J., Li, P., Jiang, W.

G., & He, J. (2016). Repulsive guidance molecule B inhibits metastasis and is associated with decreased mortality in non-small cell lung cancer. Oncotarget, 7(13), 15678-15689. https://doi.org/10.18632/oncotarget.7463

Liu, W., Chen, B., Wang, Y., Meng, C., Huang, H., Huang, X. R., Qin, J., Mulay, S. R., Anders, H. J., Qiu, A., Yang, B., Freeman, G. J., Lu, H. J., Lin, H. Y., Zheng, Z.

H., Lan, H. Y., Huang, Y., & Xia, Y. (2018). RGMb protects against acute kidney injury by inhibiting tubular cell necroptosis via an MLKL-dependent mechanism. Proceedings of the National Academy of Sciences of the United States of America, 115(7), E1475-E1484. https://doi.org/10.1073/pnas.1716959115

Malinauskas, T., Peer, T. V., Bishop, B., Mueller, T. D., & Siebold, C. (2020). Repulsive guidance molecules lock growth differentiation factor 5 in an inhibitory complex. Proceedings of the National Academy of Sciences of the United States of America, 117(27), 15620-15631. https://doi.org/10.1073/pnas.2000561117

Nie, X., Chen, W., Zhu, Y., Huang, B., Yu, W., Wu, Z., Guo, S., Zhu, Y., Luo, L., Wang, S., & Chen, L. (2018). B7-DC (PD-L2) costimulation of CD4+ T-helper 1 response via RGMb. Cellular and Molecular Immunology, 15(10), 888-897. https://doi.Org/10.1038/cm i .2017.17

Pauken, K. E., Torchia, J. A., Chaudhri, A., Sharpe, A. H., & Freeman, G. J. (2021 ). Emerging concepts in PD-1 checkpoint biology. Seminars in Immunology, 52(xxxx), 101480. https://doi.Org/10.1016/j.smim.2021.101480

Shi, Y., Zhong, L., Li, Y., Chen, Y., Feng, S., Wang, M., Xia, Y., & Tang, S. (2021 ). Repulsive Guidance Molecule b Deficiency Induces Gut Microbiota Dysbiosis and Increases the Susceptibility to Intestinal Inflammation in Mice. Frontiers in Microbiology, 72(April), 1-11. https://doi.org/10.3389/fmicb.2021.648915

Healey, E. G., Bishop, B., Elegheert, J., Bell, C. H., Padilla-Parra, S., & Siebold, C. (2015). Repulsive guidance molecule is a structural bridge between neogenin and bone morphogenetic protein. Nature Structural and Molecular Biology, 22(6), 458-465. https://doi.Org/10.1038/nsmb.3016

Tian, C., & Liu, J. (2013). Repulsive guidance molecules (RGMs) and neogenin in bone morphogenetic protein (BMP) signaling. Molecular Reproduction and Development, 80(9), 700-717. https://doi.org/10.1002/mrd.22199

Yu, S., Leung, K. M., Kim, H. Y., Umetsu, S. E., Xiao, Y., Albacker, L. A., Lee, H. J., Umetsu, D. T., Freeman, G. J., & DeKruyff, R. H. (2019). Blockade of RGMb inhibits allergen-induced airways disease. Journal of Allergy and Clinical Immunology, 144(1 ), 94-108.e11. https://doi.org/10.1016/j aci.2018.12.1022

Meng, C., Liu, W., Huang, H., Wang, Y., Chen, B., Freeman, G. J., Schneyer, A., Lin, H. Y., & Xia, Y. (2016). Repulsive guidance molecule b (RGMb) is dispensable for normal gonadal function in mice. Biology of Reproduction, 94(4), 1-10. https://doi.Org/10.1095/biolreprod.115.135921

Xiao, Y., Yu, S., Zhu, B., Bedoret, D., Bu, X., Duke-Cohan, L. M. F., Urnetsu, D. T., Sharpe, A. H., DeKruyff, R. H., & Freeman, G. J. (2014). RGMb is a novel binding partner for PD-I2 and its engagement with PD-I2 promotes respiratory tolerance. Journal of Experimental Medicine, 211(5), 943-959. https://doi.org/10.1084/jem.20130790

Siebold, C., Yamashita, T., Monnier, P. P., Mueller, B. K., & Pasterkamp, R. J. (2017). RGMs: Structural Insights, Molecular Regulation, and Downstream Signaling. Trends in Cell Biology, 27(5), 365-378. https://doi.Org/10.1016/j.tcb.2016.11.009 Samad, T. A., Srinivasan, A., Karchewski, L. A., Jeong, S. J., Campagna, J. A., Ji, R.

R., Fabrizio, D. A., Zhang, Y., Lin, H. Y., Bell, E., & Woolf, C. J. (2004). DRAGON: A Member of the Repulsive Guidance Molecule-Related Family of Neuronal- and Muscle-Expressed Membrane Proteins Is Regulated by DRG11 and Has Neuronal Adhesive Properties. Journal of Neuroscience, 24(8), 2027-2036. https://doi.Org/10.1523/JNEUROSCI.4115-03.2004

Samad, T. A., Rebbapragada, A., Bell, E., Zhang, Y., Sidis, Y., Jeong, S. J., Campagna, J. A., Perusini, S., Fabrizio, D. A., Schneyer, A. L., Lin, H. Y., Brivanlou, A. H., Attisano, L., & Woolf, C. J. (2005). DRAGON, a bone morphogenetic protein co-receptor. Journal of Biological Chemistry, 280(14), 14122-14129. https://doi.Org/10.1074/jbc.M410034200

Sanders, A. J., Ye, L., Li, J., Mason, M. D., & Jiang, W. G. (2014). Tumour angiogenesis and repulsive guidance molecule b: A role in HGF- And BMP-7- mediated angiogenesis. International Journal of Oncology, 45(3), 1304-1312. https://doi.Org/10.3892/ijo.2014.2508

Sekiya, T., & Takaki, S. (2019). RGMB enhances the suppressive activity of the monomeric secreted form of CTLA-4. Scientific Reports, 9(1 ), 1-7. https://doi.Org/10.1038/s41598-019-43068-y

Severyn, C. J., Shinde, U., & Rotwein, P. (2009). Molecular biology, genetics and biochemistry of the repulsive guidance molecule family. Biochemical Journal, 422(3), 393-403. https://doi.org/10.1042/BJ20090978 Shi, Y., Chen, G. Bin, Huang, X. X., Xiao, C. X., Wang, H. H., Li, Y. Sen, Zhang, J. F., Li, S., Xia, Y., Ren, J. L., & Guleng, B. (2015). Dragon (repulsive guidance molecule b, RGMb) is a novel gene that promotes colorectal cancer growth. Oncotarget, 6(24), 20540-20554. https://doi.org/10.18632/oncotarget.4110

Shi, Y., Huang, X. X., Chen, G. Bin, Wang, Y., Zhi, Q., Liu, Y. S., Wu, X. L., Wang, L. F., Yang, B., Xiao, C. X., Xing, H. Q., Ren, J. L., Xia, Y., & Guleng, B. (2016). Dragon (RGMb) induces oxaliplatin resistance in colon cancer cells. Oncotarget, 7(30), 48027-48037. https://doi.org/10.18632/oncotarget.10338

Siebold, C., Yamashita, T., Monnier, P. P., Mueller, B. K., & Pasterkamp, R. J. (2017). RGMs: Structural Insights, Molecular Regulation, and Downstream Signaling. Trends in Cell Biology, 27 5), 365-378. https://doi.Org/10.1016/j.tcb.2016.11.009 Robinson, R. A., Griffiths, S. C., van de Haar, L. L., Malinauskas, T., van Battum, E.

Y., Zelina, P., Schwab, R. A., Karia, D., Malinauskaite, L., Brignani, S., van den Munkhof, M. H., Dudukcu, 0., De Ruiter, A. A., Van den Heuvel, D. M. A., Bishop,

B., Elegheert, J., Aricescu, A. R., Pasterkamp, R. J., & Siebold, C. (2021 ).

Simultaneous binding of Guidance Cues NET1 and RGM blocks extracellular NEO1 signaling. Cell, 184(8), 2103-212O.e31 . https://doi.Org/10.1016/j . cell.2021 .02.045

Rotwein, P. (2019). Variation in the repulsive guidance molecule family in human populations. Physiological Reports, 7(3), 1-16. https://doi.Org/10.14814/phy2.13959

Xia, Y., Cortez-Retamozo, V., Niederkofler, V., Salie, R., Chen, S., Samad, T. A., Hong,

C. C., Arber, S., Vyas, J. M., Weissleder, R., Pittet, M. J., & Lin, H. Y. (2011 ).

Dragon (Repulsive Guidance Molecule b) Inhibits IL-6 Expression in Macrophages. The Journal of Immunology, 186(3), 1369-1376. https://doi.Org/10.4049/jimmunol.1002047 Xiao, Y., Yu, S., Zhu, B., Bedoret, D., Bu, X., Duke-Cohan, L. M. F., llmetsu, D. T., Sharpe, A. H., DeKruyff, R. H., & Freeman, G. J. (2014). RGMb is a novel binding partner for PD-I2 and its engagement with PD-I2 promotes respiratory tolerance. Journal of Experimental Medicine, 211(5), 943-959. https://doi.org/10.1084/jem.20130790

Zhang, S., He, Y., Liu, C., Li, G., Lu, S., Jing, Q., Chen, X., Ma, H., Zhang, D., Wang, Y., Huang, D., Tan, P., Chen, J., Zhang, X., Liu, Y., & Qiu, Y. (2020). MiR-93-5p enhances migration and invasion by targeting RGMB in squamous cell carcinoma of the head and neck. Journal of Cancer, 77(13), 3871-3881. https://doi.Org/10.7150/jca.43854

Claims

CLAIMS A human antibody or antigen-binding fragment thereof that is specific for human RGMb, inhibits binding of human RGMb to human PD-L2, and inhibits binding of human RGMb to (i) BMP2, BMP4, or BMP2 and BMP4, or to (ii) neogenin. The human antibody or antigen-binding fragment thereof of claim 1 , wherein said human antibody or antigen-binding fragment thereof is characterized by one or more of the following properties selected from the following list: i) binds to human RGMb with a KD of between 10 nM and 100 pM, when expressed as full chain hulgG1 and tested Bio-Layer Interferometry captured on an anti-human Fab biosensor and incubation with different concentrations of analyte huRGMb-rabbit Fc; ii) binds specifically to murine RGMb; iii) binds specifically to cynomolgus RGMb; iv) inhibits binding of human RGMb to murine PD-L2, particularly with an ICso value of between 2 nM and 50 pM, when expressed as full chain hlgG 1 and tested against dimeric human RGMb protein expressed as a fusion of the extracellular domain comprising amino acids 46-413 with rabbit Fc that has been immobilized onto Maxisorp 96-well microplates at 2 pg/ml by using increasing amounts of said human antibodies from 10 ng/ml to 10 pg/ml, followed by a constant amount of 600 ng/ml huPD-L2-Fc-biotin and detection of bound huPD-L2-Fc-biotin by Streptavidin Poly-HRP; v) does not bind to RGMa or RGMc; and vi) exhibits a monomeric content of 95 % of higher, when expressed as fullchain mouse lgG2a antibodies with Leu234Ala/Leu235Ala (LALA) mutations and formulated in PBS pH 7.4 at 2 mg/ml and exposed to three freeze/thaw cycles in liquid nitrogen and thawing at RT. The human antibody or antigen-binding fragment thereof of claim 1 or 2, wherein said human antibody or antigen-binding fragment thereof comprises a combination of a VH domain belonging to the VH3 family and a VL domain belonging to the Vlambda2 family, wherein

(vii) the VH domain comprises a set of CDR1 , CDR2 and CDR3 sequences as comprised in the VH domain sequence according to SEQ ID NO: 11 shown in Table 1 , and the VL domain comprises a set of CDR1 , CDR2 and CDR3 sequences as comprised in the VL domain sequence according to SEQ ID NO: 12 shown in Table 1 . The human antibody or antigen-binding fragment thereof of claim 3, wherein said human antibody or antigen-binding fragment thereof comprises a combination of a VH domain belonging to the VH3 family and a VL domain belonging to the Vlambda2 family, wherein

(ii) the VH domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO:9, and the VL domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO: 10;

(iii) the VH domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO: 13, and the VL domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO: 14;

(vii) the VH domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO:11 , and the VL domain comprises at least positions 5 to 138, according to the AHo numbering system, of the VH domain sequence according to SEQ ID NO: 12. The human antibody or antigen-binding fragment thereof of claim 3, wherein said human antibody or antigen-binding fragment thereof comprises a combination of

(ii) the VH domain having a sequence according to SEQ ID NO:3, and the VL domain having a sequence according to SEQ ID NO:9;

(iii) the VH domain having a sequence according to SEQ ID NO:5, and the VL domain having a sequence according to SEQ ID NO: 10;

(iv) the VH domain having a sequence according to SEQ ID NO: 13, and the VL domain having a sequence according to SEQ ID NO: 14;

(v) the VH domain having a sequence according to SEQ ID NO:1 , and the VL domain having a sequence according to SEQ ID NO:2; (vi) the VH domain having a sequence according to SEQ ID NO:3, and the VL domain having a sequence according to SEQ ID NO:4;

(vii) the VH domain having a sequence according to SEQ ID NO:11 , and the VL domain having a sequence according to SEQ ID NO: 12; The human antibody or antigen-binding fragment thereof of any one of claims 1 to 5, wherein said antibody is a full-length antibody, particularly a full-length antibody selected from the list of IgA, IgD, IgE and IgG, particularly an IgG, more particularly an IgG selected from the list of lgG1 , lgG2, lgG3 and lgG4. The human antibody or antigen-binding fragment thereof of any one of claims 1 to 5, wherein said human antibody or antigen-binding fragment thereof is an antigen-binding fragment selected from the list of: Fab fragment, Fab’ fragment, F(ab’)2 fragment, Fv fragment, dsFv fragment, scFv fragment, dsscFv fragment and a fusion protein comprising a Fab fragment, a Fab’ fragment, a F(ab’)2 fragment, an Fv fragment, a dsFv fragment, an scFv fragment and a dsscFv fragment, particularly wherein said antigen-binding fragment is an scFv fragment or a fusion protein comprising an scFv fragment, particularly an scFv-Fc fusion protein or an (scFv)2-Fc fusion protein. A multispecific antibody-based binding composition with specificity for at least RGMb and a second antigen, wherein said multispecific antibody-based binding composition comprises a first human antibody or antigen-binding fragment thereof, which is a human antibody or antigen-binding fragment thereof of any one of claims 1 to 7, and at least a second human antibody or antigen-binding fragment thereof with specificity for a different antigen, particularly wherein said multispecific antibody-based binding composition is bispecific. The multispecific antibody-based binding composition of claim 8, wherein said second human antibody or antigen-binding fragment thereof is specific for human PD-L1 or PD-1. The multispecific antibody-based binding composition of claim 9, wherein said second human antibody or antigen-binding fragment thereof comprises a combination of

(i) the VH domain having a sequence according to SEQ ID NO: 18, and the VL domain having a sequence according to SEQ ID NO: 19;

(ii) the VH domain having a sequence according to SEQ ID NQ:20, and the VL domain having a sequence according to SEQ ID NO:21 ; or

(iii) the VH domain having a sequence according to SEQ ID NO:22, and the VL domain having a sequence according to SEQ ID NO:23. The multispecific antibody-based binding composition of any one of claims 8 to 10, wherein said first and said second human antibody or antigen-binding fragment thereof are comprised in one multispecific construct, particularly a multispecific construct selected from the list of: a fusion of antibody fragments, in particular scFv fragments, to the N- and/or C-terminal ends of the antibody heavy and/or light chains of full immunoglobulins, in particular Morrison-H or Morrison- L constructs, diabody; scDb; tandem di-scFv, tandem tri-scFv; Fab-scFv; scFab- dsscFv; Fab-(scFv)2; Fab2; Fab-(Fv)2; triabody; scDb-scFv; tandem tri-scFv, a scFab-dsscFv, and constructs comprising two antigen-binding fragments selected from Fab fragment, Fab’ fragment, F(ab’)2 fragment, Fv fragment, dsFv fragment, scFv fragment and dsscFv fragment, wherein each of said two antigenbinding fragment is fused to the N- and/or the C-terminus of one chain of a heterodimerization domain, particularly one chain of a heterodimeric Fc domain. A nucleic acid encoding the human antibody or antigen-binding fragment thereof of any one of claims 1 to 7, or the multispecific antibody-based binding composition of any one of claims 8 to 11 . A vector comprising the nucleic acid of claim 12. A host cell comprising the nucleic acid of claim 12 or the vector of claim 13. A method for producing the human antibody or antigen-binding fragment thereof of any one of claims 1 to 7, or the multispecific antibody-based binding composition of any one of claims 8 to 11 , comprising a step selected from (a) expressing the nucleic acid of claim 12 in an expression system; (b) expressing a nucleic acid from the vector of claim 13; and (c) culturing the host cell of claim

14. A pharmaceutical composition comprising the human antibody or antigen-binding fragment thereof of any one of claims 1 to 7, or the multispecific antibody-based binding composition of any one of claims 8 to 11 , and a pharmaceutically acceptable carrier. The human antibody or antigen-binding fragment thereof of any one of claims 1 to 7, the multispecific antibody-based binding composition of any one of claims 8 to 11 , or the pharmaceutical composition of claim 16, for use in the treatment of conditions, where RGMb is expressed and that would benefit from upregulation of an immune response, in particular cancer.