Masking antigen binding constructs and methods of making same

Abstract

Antigen binding constructs comprising alpha-helix masking structures and methods of making such masking antigen binding constructs are provided. In one aspect, the methods relate to a computer-aided method for designing an alpha-spiral masking structure.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to antigen binding constructs comprising alpha-helix masking structures and methods of making such masked antigen binding constructs. BACKGROUND

[0002] While antigen binding constructs (e.g., antibodies and antigen binding fragments thereof) typically bind to their intended target with specificity, most target molecules are not specific to the site of disease and can be present in cells or tissues outside of the site of disease.

[0003] These off-target effects can be overcome by adding inhibitory or masking structures to antigen binding constructs using cleavable linkers. These linkers can be designed to be cleaved by enzymes specific to a particular tissue or pathology, thereby preferentially activating the masked antigen binding construct at the desired location. However, inhibitory or masking domains effective for one antigen binding construct often do not work as well for another construct, and the addition of large protein domains to antigen binding constructs often leads to manufacturing issues. Therefore, there is a need to effectively design these inhibitory or masking domains for any given antigen binding construct. SUMMARY

[0004] The present disclosure provides an activatable masked antigen binding construct, wherein an alpha-helix masking structure is added to the antigen binding construct. Alpha helices are one of the major forms of protein secondary structure, with the polypeptide chain backbone arranged in a regular helix around a central axis. In alpha helices, the R groups of the amino acids point outward and downward to avoid steric clashes with the polypeptide chain backbone. The core of the alpha helix is tightly bound with atoms connected by van der Waals forces. Because alpha helices have a stable secondary structure, scoring of screening processes such as structure prediction and molecular docking is more reliable. The present disclosure provides an alpha-helix masking structure that is able to form a stable steric effect with the antigen binding domain of the masked antigen binding construct. This structure effectively blocks the antigen binding ability of the masked antigen binding construct in normal tissues and selectively "activates" the binding activity of the masked antigen binding construct (e.g., an antibody or antigen binding fragment thereof) to the target molecule in diseased tissues.

[0005] The alpha-helix masking structure can effectively prevent the masked antigen binding construct from binding to its intended target until the alpha-helix masking structure is cleaved or otherwise removed. In other words, the alpha-helix masking structure shields the antigen binding portion of the masked antigen binding construct from interacting with the target. In certain embodiments, the alpha-helix masking structure can be removed (e.g., cleaved) by one or more molecules present in the in vivo environment (e.g., a protease) after the masked antigen binding construct is administered to a patient. In other uses, such as non-therapeutic uses, the alpha-helix masking structure can be removed by adding one or more proteases to a culture medium comprising the masked antigen binding construct. Upon removal of the alpha-helix masking structure, the masked antigen binding construct regains the ability to bind to the target and the masked antigen binding construct (e.g., an antibody or antigen binding fragment thereof) can be specifically targeted.

[0006] In one aspect, the disclosure relates to a masked antigen binding domain comprising (a) an antigen binding domain, and (b) an alpha-helix masking structure, wherein the alpha-helix masking structure comprises an alpha-helix structure, is linked to the antigen binding domain, optionally through a single covalent bond, and prevents interaction between the antigen binding domain and its target antigen.

[0007] In one aspect, the disclosure relates to a masked antigen binding domain comprising (a) an antigen binding domain, and (b) an alpha-helix masking structure, wherein the alpha-helix masking structure comprises an alpha-helix structure, is linked to the antigen binding domain through a single covalent bond, and prevents interaction between the antigen binding domain and its target antigen.

[0008] In certain embodiments, the alpha-helix masking structure can comprise two or more alpha-helix structures (e.g., 2, 3, 4, or 5 alpha-helix structures). In certain embodiments, the two or more alpha-helix structures are linked to each other in the polypeptide, optionally comprising a linking sequence between adjacent alpha-helix structures.

[0009] In certain embodiments, the alpha-helix structure does not form a coiled-coil structure with the masked antigen binding construct through multiple covalent bonds.

[0010] In certain embodiments, the alpha-helix masking structure comprises a single alpha-helix structure.

[0011] In one aspect, the disclosure relates to a masked antigen binding domain comprising (a) an antigen binding domain, and (b) an alpha-helix masking structure, wherein the alpha-helix masking structure comprises a polypeptide chain comprising an alpha-helix structure.

[0012] In certain embodiments, the masked antigen binding domain further comprises a cleavable linker between (a) the antigen binding domain and (b) the alpha-helix masking structure.

[0013] In certain embodiments, the cleavable linker is cleavable by an enzyme to release the antigen binding domain masked by the alpha-helix masking structure.

[0014] In certain embodiments, the enzyme is an enzyme present in the tumor microenvironment (e.g., a matrix metalloproteinase).

[0015] In certain embodiments, the antigen binding domain comprises (1) a heavy chain variable region (VH) and a light chain variable region (VL), or (2) a VHH.

[0016] In some embodiments, the alpha-helix masking structure is linked to the N-terminus of the VH, the N-terminus of the VL, or the N-terminus of the VHH. In some embodiments, the alpha-helix masking structure is linked to the C-terminus of the VH, the C-terminus of the VL, or the C-terminus of the VHH.

[0017] In one aspect, the disclosure relates to a masked antigen binding domain comprising (a) an antigen binding domain, and (b) an alpha-helix masking structure, wherein the alpha-helix masking structure comprises an alpha-helix structure, the alpha-helix masking structure is linked to the masked antigen binding construct by a single covalent bond, and the alpha-helix masking structure prevents interaction between the antigen binding domain and its target antigen.

[0018] In certain embodiments, the alpha-helix structure does not form a coiled coil structure with the masked antigen binding construct through multiple covalent bonds.

[0019] In certain embodiments, the alpha-helix masking structure comprises a single alpha-helix structure.

[0020] In one aspect, the disclosure relates to a masked antigen binding construct comprising, (a) an antigen binding domain, and (b) an alpha-helix masking structure, wherein the alpha-helix masking structure comprises a polypeptide chain comprising an alpha-helix structure.

[0021] In certain embodiments, the masked antigen binding construct further comprises a cleavable linker between (a) the antigen binding domain and (b) the alpha-helix masking structure.

[0022] In certain embodiments, the cleavable linker can be cleaved by an enzyme, thereby releasing the a-helix masking structure from the masking antigen binding construct.

[0023] In certain embodiments, the enzyme is an enzyme present in the tumor microenvironment (e.g., a matrix metalloproteinase).

[0024] In certain embodiments, the masking antigen binding construct comprises (a) a first polypeptide comprising an a-helix masking structure, a cleavable linker, a VH; and (b) a second polypeptide comprising a VL, wherein the a-helix masking structure is attached to the N-terminus or C-terminus of the first polypeptide.

[0025] In certain embodiments, the masking antigen binding construct further comprises a third polypeptide comprising a VHH.

[0026] In certain embodiments, the a-helix masking structure is attached to the N-terminus or C-terminus of the third polypeptide.

[0027] In certain embodiments, the masking antigen binding construct further comprises a third polypeptide comprising a VH and a VL.

[0028] In certain embodiments, the a-helix masking structure is attached to the N-terminus or C-terminus of the third polypeptide.

[0029] In certain embodiments, the masking antigen binding construct further comprises: (c) a third polypeptide comprising a VH; and (d) a fourth polypeptide comprising a VL.

[0030] In certain embodiments, the a-helix masking structure is attached to the N-terminus or C-terminus of the third polypeptide.

[0031] In certain embodiments, the masking antigen binding construct comprises (a) a first polypeptide comprising an a-helix masking structure, a cleavable linker, a VL; and (b) a second polypeptide comprising a VH, wherein the a-helix masking structure is attached to the N-terminus or C-terminus of the first polypeptide.

[0032] In certain embodiments, the masking antigen binding construct further comprises a third polypeptide comprising a VHH.

[0033] In certain embodiments, the a-helix masking structure is attached to the N-terminus or C-terminus of the third polypeptide.

[0034] In some embodiments, the masked antigen-binding construct further comprises a third polypeptide comprising VH and VL.

[0035] In some embodiments, the α-helical masking structure is attached to the N-terminus or C-terminus of the third polypeptide.

[0036] In some embodiments, the masked antigen-binding construct further includes: (c) A third polypeptide containing VH; and (d) The fourth polypeptide containing VL.

[0037] In some embodiments, the α-helical masking structure is attached to the N-terminus or C-terminus of the fourth polypeptide.

[0038] In some embodiments, the masking antigen-binding construct includes a first polypeptide comprising an α-helical masking structure, a cleavable linker, VH, and VL, wherein the α-helical masking structure is attached to the N-terminus or C-terminus of the first polypeptide.

[0039] In some embodiments, the masked antigen-binding construct further includes a second polypeptide containing VHH.

[0040] In some embodiments, the α-helical masking structure is attached to the N-terminus or C-terminus of the second polypeptide.

[0041] In some embodiments, the masked antigen-binding construct further includes a second polypeptide comprising VH and VL.

[0042] In some embodiments, the α-helical masking structure is attached to the N-terminus or C-terminus of the second polypeptide.

[0043] In some embodiments, the masked antigen-binding construct further includes a second polypeptide comprising VH and a third polypeptide comprising VL.

[0044] In some embodiments, the masking antigen-binding construct includes a first polypeptide comprising an α-helical masking structure, a cleavable linker, and a VHH, wherein the α-helical masking structure is attached to the N-terminus or C-terminus of the first polypeptide.

[0045] In some embodiments, the masked antigen-binding construct further includes a second polypeptide comprising VHH.

[0046] In some embodiments, the α-helical masking structure is attached to the N-terminus or C-terminus of the second polypeptide.

[0047] In some embodiments, the masked antigen-binding construct further includes a second polypeptide comprising VH and VL.

[0048] In some embodiments, the α-helical masking structure is attached to the N-terminus or C-terminus of the second polypeptide.

[0049] In some embodiments, the masked antigen-binding construct further includes a second polypeptide containing VH and a third polypeptide containing VL.

[0050] In some embodiments, the number of amino acids in the α-helical masking structure does not exceed 150.

[0051] In some embodiments, the number of amino acids in the α-helical masking structure does not exceed 100.

[0052] In some embodiments, the number of amino acids in the α-helix masking structure does not exceed 50.

[0053] In some embodiments, the α-helical masking structure contains 5-40 amino acids.

[0054] In some embodiments, the α-helical masking structure contains 5-20 amino acids.

[0055] In some embodiments, the α-spiral masking structure has one or more parameters that satisfy a threshold level.

[0056] In some embodiments, one or more parameters are selected from the group consisting of: (1) α-helix pLDDT (predicted local distance difference test), (2) α-helix pTM score (predicted template modeling score), (3) ipTM score between the α-helix and the CDR region (interface pTM score), and (4) the average minimum distance between the α-carbon atom of the α-helix and the CDR region.

[0057] In some embodiments, the pLDDT value of the α-helix is ​​greater than 55.

[0058] In some embodiments, the pTM fraction of the α-helix is ​​greater than 0.5.

[0059] In some embodiments, the ipTM score between the α-helix and the CDR region is greater than 0.6, and more preferably greater than 0.8.

[0060] In some embodiments, the average minimum distance between the α-carbon atom of the α-helix and the CDR region is less than 12 Å, and more preferably less than 10 Å.

[0061] In some embodiments, the average minimum distance between the α-carbon atom of the α-helix and the CDR region is less than 12 Å, and the ipTM score is greater than 0.6.

[0062] In some embodiments, the average minimum distance between the α-carbon atom of the α-helix and the CDR region is less than 10 Å, and the ipTM score is greater than 0.8.

[0063] In some embodiments, the masked antigen-binding domain or the masked antigen-binding structure contains only one α-helix masking structure.

[0064] In some embodiments, the masked antigen-binding domain or the masked antigen-binding structure comprises two α-helical masks, wherein the two α-helical masks are respectively connected to two different antigen-binding domains.

[0065] In some embodiments, the α-helix structure comprises an amino acid sequence of any one of SEQ ID NO: 12, 16, 18 or 20, or an amino acid sequence having at least 95%, 97% or 99% identity with an amino acid sequence of any one of SEQ ID NO: 12, 16, 18 or 20.

[0066] In some embodiments, the cleavable linker includes a first linker, a cleavable site, and a second linker, wherein the cleavable site comprises the amino acid sequence of SEQ ID NO:14 or 24, or an amino acid sequence that is at least 95%, 97%, or 99% identical to the amino acid sequence of SEQ ID NO:14 or 24.

[0067] In some embodiments, the first linker and / or the second linker comprises an amino acid sequence of any one of SEQ ID NO: 13, 21 or 23, or an amino acid sequence having at least 95%, 97% or 99% identity with an amino acid sequence of any one of SEQ ID NO: 13, 21 or 23.

[0068] In one aspect, this disclosure relates to a nucleotide comprising a nucleotide sequence encoding the masking antigen-binding domain or the masking antigen-binding construct described herein.

[0069] In one respect, this disclosure relates to vectors containing the nucleic acids described herein.

[0070] On the other hand, this disclosure relates to cells comprising the nucleic acids or vectors described herein.

[0071] In another aspect, this disclosure relates to a method for producing a masked antigen-binding domain or a masked antigen-binding construct, the method comprising culturing the cells described herein.

[0072] In one aspect, this disclosure relates to a masking antibody-drug conjugate (ADC) comprising a therapeutic agent covalently bound to the masking antigen-binding domain or the masking antigen-binding construct described herein.

[0073] In some embodiments, the therapeutic agent is a cytotoxic agent or a cell inhibitor.

[0074] In one respect, this disclosure relates to a nucleotide therapy comprising the nucleotides described herein.

[0075] In one aspect, this disclosure relates to a cell therapy comprising the masked antigen-binding domain or the masked antigen-binding construct described herein.

[0076] In one aspect, this disclosure relates to a fusion protein comprising (1) a cytokine and (2) a masking antigen-binding domain or a masking antigen-binding construct as described herein.

[0077] In one aspect, this disclosure relates to a method of treating a subject with cancer, the method comprising administering to the subject in need a therapeutically effective amount of a composition comprising the masking antigen-binding domain described herein, the masking antigen-binding construct described herein, the antibody-drug conjugate described herein, the nucleotide therapy described herein, the cell therapy described herein, or the fusion protein described herein.

[0078] In one aspect, this disclosure relates to a computer implementation method for generating masked antigen-binding constructs, the method comprising: (a) Provides information on the initiation antigen-binding domain; (b) Generate candidate α-helical masking structures that can bind to the CDR region of the initiating antigen-binding domain; (c) Assemble the candidate α-helix masking structure from step (b) into a masked antigen-binding construct in a computer, the construct comprising an initiating antigen-binding domain and a candidate α-helix masking structure; (d) Perform structural prediction; and (e) Select a masked antigen binding construct with one or more parameters that meet the threshold level.

[0079] In some embodiments, the one or more parameters are selected from the group consisting of: (1) α-helix pLDDT (predicted local distance difference test), (2) α-helix pTM score (predicted template modeling score), (3) ipTM score between the α-helix and the CDR region (interface pTM score), and (4) the average minimum distance between the α-carbon atom of the α-helix and the CDR region.

[0080] In some embodiments, the value of α-helical pLDDT is greater than 55.

[0081] In some embodiments, the value of the α-helical pTM fraction is greater than 0.5.

[0082] In some embodiments, the ipTM fraction between the α-helix and the CDR region is greater than 0.6, more preferably greater than 0.8.

[0083] In some embodiments, the average minimum distance between the α-carbon atom of the α-helix and the CDR region is less than 12 Å, more preferably less than 10 Å.

[0084] In some embodiments, the average minimum distance between the α-carbon atom of the α-helix and the CDR region is less than 10 Å and the ipTM score is greater than 0.8. In some embodiments, the average minimum distance between the α-carbon atom of the α-helix and the CDR region is less than 12 Å and the ipTM score is greater than 0.6.

[0085] In some embodiments, step (b) is performed using a protein language model (PLM) or diffusion model design.

[0086] In some embodiments, step (d) is performed using rapid folding software.

[0087] In some embodiments, the rapid folding software is selected from the group consisting of Alphafold2, ESMFold, and OmegaFold.

[0088] In some embodiments, at least 100 or 1000 antigen-binding constructs are generated on a computer.

[0089] In some embodiments, the method further includes preparing a vector containing a nucleotide sequence encoding a masked antigen-binding construct; and expressing the masked antigen-binding construct.

[0090] In one aspect, this disclosure relates to one or more machine-readable hardware storage devices for storing instructions executable by one or more data processing devices to perform the methods described herein.

[0091] In another aspect, this disclosure relates to a system comprising: one or more data processing devices; and one or more machine-readable hardware storage devices storing instructions executable by the aforementioned one or more data processing devices to perform the methods described herein.

[0092] In some embodiments, the system further includes one or more devices for synthesizing and expressing nucleotide sequences.

[0093] In one aspect, this disclosure provides a masking antigen-binding construct (e.g., an antibody or its antigen-binding fragment) comprising at least one α-helical masking structure. In some embodiments, in a VH and VL pair, the α-helix is ​​attached to the N-terminus or C-terminus of the VH but not to the VL; or the α-helix is ​​attached to the N-terminus or C-terminus of the VL but not to the VH. In some embodiments, the α-helix is ​​attached to the N-terminus or C-terminus of the VHH. In some embodiments, the α-helical masking structure has a coiled-coil structure.

[0094] In some embodiments, in the masked antigen-binding construct, the α-helical masking structure can (1) provide steric hindrance due to its helical structure and (2) have binding affinity to the target antigen of the antigen-binding construct.

[0095] In one aspect, this disclosure relates to a masked antigen-binding domain comprising: Antigen-binding domain; and (b) α-spiral masking structure, The α-helix masking structure contains an α-helix structure that is connected to the antigen-binding domain, optionally via a single covalent bond, and prevents the interaction between the antigen-binding domain and its target antigen.

[0096] In some embodiments, (1) the α-helix structure does not form a coiled helical structure connected to the antigen-binding domain by multiple covalent bonds; and / or (2) the α-helix structure specifically binds to the antigen-binding domain.

[0097] In some embodiments, the α-helix masking structure includes a single α-helix structure, or one or more α-helix structures (e.g., 2, 3, 4 or 5 α-helix structures).

[0098] In one aspect, this disclosure relates to an α-helical masking structure comprising: one or more α-helical peptides, wherein the one or more α-helical peptides are capable of specifically binding to an antigen-binding domain of an antibody or an antibody fragment.

[0099] In some embodiments, the binding of the α-helical peptide to the antigen-binding domain is non-covalent.

[0100] In one aspect, this disclosure relates to a masking structure comprising: one or more α-helical peptides and one or more cyclic peptides, wherein the one or more cyclic peptides are capable of specifically binding to the antigen-binding domain of an antibody or antibody fragment.

[0101] In one aspect, this disclosure relates to a masking antigen-binding domain comprising (a) an antigen-binding domain and (b) the α-helical masking structure described herein, wherein the α-helical masking structure comprises an α-helical structure and the α-helical masking structure is connected to the antigen-binding domain via one or more cleavable linkers, thereby blocking the interaction between the antigen-binding domain and its target antigen.

[0102] In this document, the term "antibody" refers to any antigen-binding molecule that contains at least one (e.g., one, two, three, four, five, or six) complementarity-determining regions (CDRs) (e.g., any one of the three CDRs of the immunoglobulin light chain or any one of the three CDRs of the immunoglobulin heavy chain) and is capable of specifically binding to an epitope. Non-limiting examples of antibodies include: monoclonal antibodies, polyclonal antibodies, multispecific antibodies (e.g., bispecific antibodies), single-chain antibodies, single variable domain (VHH) antibodies, chimeric antibodies, human antibodies, and humanized antibodies. Antibodies may contain the Fc region of human antibodies. The term "antibody" also includes derivatives such as bispecific antibodies, single-chain antibodies, bivalent antibodies, linear antibodies, and multispecific antibodies formed from antibody fragments.

[0103] As used herein, "antigen-binding construct" refers to any substance capable of binding an antigen, such as a polypeptide or polypeptide complex. In some embodiments, an antigen-binding construct is a protein that specifically binds to an antigen of interest. An antigen-binding construct can be a monomer, dimer, multimer, protein, peptide, or protein or peptide complex; an antibody, antibody fragment, or antigen-binding fragment thereof; scFv, etc. An antigen-binding construct can be a single-specific, bispecific, or multispecific polypeptide construct. In some aspects, an antigen-binding construct may comprise, for example, one or more antigen-binding components (such as Fab or scFv) linked to one or more Fc domains. Other embodiments of antigen-binding constructs are also provided below and in the examples. An antigen-binding construct includes an antibody and an antigen-binding fragment of an antibody, such as a bispecific antibody.

[0104] In this document, the term "antigen-binding fragment" refers to a portion of a full-length antibody that specifically binds to an antigen. An antigen-binding fragment may contain at least one variable region (e.g., a variable region of the heavy chain, a variable region of the light chain, or VHH). Non-limiting examples of antibody fragments include Fab, Fab', F(ab')2, and Fv fragments.

[0105] As used herein, the term "antigen-binding domain" refers to the antigen-binding portion of an antibody. In some embodiments, the antigen-binding domain consists of VH and VL. In some embodiments, the antigen-binding domain consists of VHH.

[0106] The term “α-helix masking structure” as used herein refers to a protein structure containing an α-helix that effectively masks the antigen-binding domain, resulting in reduced antigen-binding activity of the antigen-binding domain.

[0107] The terms “subject” and “patient” as used herein are interchangeable throughout the specification and are used to describe an animal, human, or non-human being treated with the methods of this disclosure. This disclosure is intended for both veterinary and non-veterinary applications. Human patients may be adult or minor (e.g., humans under the age of 18). In addition to humans, patients include, but are not limited to, mice, rats, hamsters, guinea pigs, rabbits, mink, cats, dogs, and primates. For example, this includes non-human primates (such as monkeys, chimpanzees, gorillas, etc.), rodents (such as rats, mice, gerbils, hamsters, mink, rabbits), lagomorphs, pigs (such as pigs, miniature pigs), equines, canines, felines, bovines, and other domesticated, farm, and zoo animals.

[0108] In this article, the term "linker" refers to an amino acid sequence of two or more amino acids in length. Linkers can consist of neutral polar or nonpolar amino acids. For example, the length of a linker can range from 2 to 100 amino acids, specifically from 2 to 50 amino acids, such as 3, 5, 10, 15, 20, 25, 30, 35, 40, 45, or 50 amino acids. Linkers can be cleavable linker peptides, for example, through self-cleavage, enzymatic cleavage, or chemical cleavage. This article also describes cleavable sites (e.g., enzymatic cleavage sites) in the amino acid sequence and enzymes and chemicals capable of cleavage at such sites. Enzymatic cleavage of the "linker" or "cleavable site" can be performed using endopeptidases such as Lys-C, Asp-N, Arg-C, V8, Glu-C, chymotrypsin, trypsin, pepsin, papain, thrombin, Genenase, Factor Xa, TEV (tobacco etch virus cysteine ​​protease), enterokinase, HRV C3 (human rhinovirus C3 protease), plasminogen activator, and subtilisin-like protease precursor convertases (e.g., Furin (PC I), PC2, or PC3) or N-arginine dibase convertase. Chemical cleavage may involve the use of hydroxylamine, N-chlorosuccinimide, N-bromosuccinimide, or cyanogen bromide.

[0109] In this paper, the term "curled-helical domain" or "curled-helix" refers to a structure formed by two interacting α-helical structures. These α-helical structures can be right-handed α-helices.

[0110] In this article, the term "cleavable site" may refer to the substrate of the protease. Cleavage of a cleavable site may involve the use of endopeptidases such as Lys-C, Asp-N, Arg-C, V8, Glu-C, chymotrypsin, trypsin, pepsin, papain, thrombin, Genenase, Factor Xa, TEV (tobacco-etch virus cysteine ​​protease), enterokinase, HRV C3 (human rhinovirus C3 protease), prothrombinase, and Bacillus subtilis protease-like precursor convertases (e.g., Furin (PCI), PC2, or PC3) or N-arginine dibasic convertase. Cleavage of a cleavable site may also involve chemical cleavage, such as cleavage using hydroxylamine, N-chlorosuccinimide, N-bromosuccinimide, or cyanogen bromide.

[0111] In the terminology used herein, “polypeptide,” “peptide,” and “protein” are used interchangeably and all refer to amino acid polymers consisting of at least two amino acids.

[0112] In the terminology used herein, “polynucleotide,” “nucleic acid molecule,” and “nucleic acid sequence” are used interchangeably and all refer to nucleotide polymers consisting of at least two nucleotides, including but not limited to DNA, RNA, DNA / RNA hybrids, and their modified forms.

[0113] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. This document describes methods and materials that can be used with the invention; other suitable methods and materials known in the art may also be used. The materials, methods, and embodiments described are for illustrative purposes only and are not intended to be limiting. All publications, patent applications, patents, sequences, database entries, and other references mentioned herein are incorporated herein in their entirety. In the event of conflict, this specification (including definitions) shall be decisive. Details of one or more embodiments are set forth in the accompanying drawings and the description below. Other features, objects, and advantages of the invention will become apparent from the description, drawings, and claims. The contents of any patents, patent applications, and other references cited in this specification are incorporated herein in their entirety. Attached Figure Description

[0114] Figures 1A-1J The structural prediction plots for CEA+TAm-R1-L16, CEA+TAm-R1-L17, N_L_Pani_mask70, CEA+TAm-R1-H1, CEA+TAm-R1-H4, CEA+TAm-R1-H5, N_H_Cet-mask72, N_H_Pani_mask1, Tra-R2-2, and Tra-R2-4 are shown. Figure 1KAn example predicted structure of a masked anti-CD19 antibody is shown.

[0115] Figures 2A-2B The binding map between the example antibody and the mouse CD3 antigen is shown, and the results were obtained by ELISA assay.

[0116] Figure 3 is a schematic diagram of a system used to perform silicon-based screening with an α-helix masking structure.

[0117] Figure 4 shows an example structure of a masked antigen-binding domain.

[0118] Figure 5 shows an example structure of a masked Fab-Fc.

[0119] Figure 6 shows an example structure for masking VHH-Fc.

[0120] Figures 7A-7B show example structures for masking ScFv-Fc.

[0121] Figures 8A-8B show example structures of masked Fab-Fc fusion proteins.

[0122] Figure 9 shows an example structure of the masked ScFv-Fc fusion protein.

[0123] Figure 10 shows an example structure of a masking Fab.

[0124] Figure 11 shows an example structure for masking VHH.

[0125] Figures 12A-12B show example structures for masking ScFv.

[0126] Figures 13A-13B show example structures of masked Fab fusion proteins.

[0127] Figure 14 shows an example structure of a masked ScFv fusion protein.

[0128] Figure 15 shows an example structure for masking the extracellular domain.

[0129] Figure 16 shows a binding map between an example anti-CD19 antibody and Raji cells, the results of which were obtained by flow cytometry.

[0130] Figures 17A-17B show the structural predictions of the first type (Figure 17A) and second type (Figure 17B) structures of the masked anti-HER2 single-domain antibody.

[0131] Figures 18A-18B show the binding patterns between example anti-HER2 single-domain antibodies and H1573 cells, obtained by flow cytometry. Figure 18A shows the results for anti-HER2 single-domain antibodies containing one α-helix; Figure 18B shows the results for anti-HER2 single-domain antibodies containing two α-helices.

[0132] Figure 19 shows the binding pattern between an example anti-HER2 single-domain antibody and H1573 cells after enzymatic digestion, the results of which were obtained by flow cytometry.

[0133] Figure 20 shows the SDS-PAGE results before and after enzymatic digestion of the anti-HER2 single-domain antibody.

[0134] Figure 21 shows the binding pattern between an example anti-CD3 single-domain antibody and Jurkat cells, obtained by flow cytometry.

[0135] Figure 22 shows the binding pattern between an example anti-CD3 single-domain antibody and Jurkat cells after enzymatic digestion, the results of which were obtained by flow cytometry.

[0136] Figure 23 shows the SDS-PAGE results before and after enzymatic digestion of the anti-CD3 single-domain antibody.

[0137] Figure 24 shows the relevant amino acid sequences. Detailed Implementation

[0138] Antibodies are a class of immunoglobulins that specifically bind to antigens. Due to this specific binding mechanism, antibodies can "tag" foreign microorganisms and infected cells, thereby inducing other immune mechanisms to attack them or directly neutralizing their targets. Antibodies are widely used in medical practice and are mainstream targeted therapies for malignant tumors, infectious diseases, autoimmune diseases, transplant rejection, and chronic inflammatory diseases. However, targets that are highly expressed in diseased tissues but lowly expressed in normal tissues are difficult to find. Therefore, antibody drugs may react with normal tissues, causing side effects.

[0139] To address this problem, antibodies can be masked and selectively activated in specific tissues. For example, some antibodies can be masked using coiled coils. A coiled coil is a protein supersecondary structure consisting of 2-7 α-helices (most commonly 2 or 4 α-helices) intertwined to form a helical structure. In the design of coiled coil-masked antibodies, two α-helices are attached to the heavy and light chains of the antibody, respectively. These two α-helices need to have sufficiently high affinity to form a coiled coil near the antibody binding domain, thereby blocking antigen-antibody binding. The key to this design is that the two α-helices attached to the antibody heavy and light chains need to have high affinity (rather than the interaction between the coiled coil and the antibody). When the linker connecting the α-helices to the antibody is digested by an enzyme, the coiled coil dissociates from the antibody binding domain, and the inhibition is released.

[0140] Unlike strategies that mask antigen-binding constructs using coiled helices, this disclosure employs computer-generated virtual design or library screening to obtain α-helices capable of binding to complementarity-determining regions (CDRs). These α-helices are then linked to variable regions of the masking antigen-binding construct via cleavable linkers (e.g., linkers containing cleavable sites). The α-helix (by binding to the CDR of the masking antigen-binding construct) binds to the antigen-binding domain of the masking antigen-binding construct, thereby blocking the interaction between the target antigen and the masking antigen-binding construct. In specific tissues, the cleavable linker (e.g., a linker containing a cleavable site) can be cleaved to selectively activate the masking antigen-binding construct to recognize antigens, thereby reducing off-target toxicity of the antigen-binding drug. This method primarily considers the binding force between the α-helix and the CDR region, rather than the interaction between the α-helices themselves. Therefore, this design only requires linking one linker to one strand of the masking antigen-binding construct to be applied to single-domain antibodies (or other antibodies with single-chain structures). The masking effect of this design can be modulated by adjusting the strength of the interaction between the α-helix and the CDR region.

[0141] In one aspect, this disclosure relates to an α-helical masking structure covalently linked to the N-terminus of either the variable region of an antibody heavy chain or the variable region of an antibody light chain (but not both simultaneously). In some embodiments, the masking antibody described herein may comprise two light chains and two heavy chains, and therefore may comprise a pair of α-helical masking structures described herein. In some embodiments, the two α-helical masking structures do not bind together to form a coiled-helix structure. In some embodiments, the interaction between the α-helical masking structure described herein and the CDR region of a masking antigen-binding construct (e.g., an antibody) is much stronger than the interaction between the two α-helical masking structures.

[0142] In contrast, in traditional bivalent α-helical masking structures (such as coiled helices), two α-helical chains strongly bind to form the bivalent α-helical masking structure. The α-helical chains themselves do not bind to the CDR region of the masked antigen-binding construct (such as an antibody). Traditional bivalent α-helical masking structures (such as coiled helices) can contain two α-helical chains that are strongly bound together through covalent or non-covalent interactions to form the bivalent α-helical masking structure. Each of these two α-helical chains can be attached to the N-terminus of the antibody heavy chain and the N-terminus of the antibody light chain, respectively, thereby blocking the interaction between the CDR and the target antigen. The α-helical chains themselves do not bind to the CDR region. For detailed descriptions of traditional bivalent α-helical masking structures, please refer to, for example, “LUZ-Y, a novel platform for themammalian cell production of full-length IgG-bispecific antibodies” published by Wranik, Bernd J. et al. in the Journal of Biological Chemistry, Vol. 287, No. 52, 2012, pp. 43331-43339; “Acoiled-coil masking domain for selective activation of therapeutic antibodies” published by Trang, Vivian H. et al. in Nature Biotechnology, Vol. 37, No. 7, pp. 761-765; US Patent Publication No. 2016 / 0002356; and US Patent Publication No. 2019 / 0352428. All of the above references are included in full in this article.

[0143] In one aspect, the uncleaved masked antigen-binding constructs provided herein comprise a masked antigen-binding domain and an α-helical masking structure linked by a cleavable linker (e.g., a cleavable linker containing a cleavable site). This linker results in a reduced ability of the antigen-binding domain to bind its target antigen. In an activated state (e.g., a cleavage state), the antigen-binding domain is released from the α-helical masking structure and can regain some or all of its ability to bind the target antigen.

[0144] This invention provides a masked antigen-binding construct (e.g., an antibody and its antigen-binding fragment) comprising a removable α-helical masking structure that prevents the binding of the antigen-binding domain to its antigen target. The removable masking structure can be attached to the N-terminus of one or more heavy and / or light chains of the masked antigen-binding construct via a cleavable linker. The cleavable linker can be a matrix metalloproteinase (MMP) cleavable sequence.

[0145] In the tumor microenvironment, alterations in proteolysis lead to unregulated tumor growth, tissue remodeling, inflammation, tissue invasion, and metastasis. Matrix metalloproteinases (MMPs) are the most prominent family of proteases associated with tumorigenesis, mediating many changes in the tumor microenvironment during tumor progression. When masked antigen-binding constructs (such as antibodies and their antigen-binding fragments) are exposed to MMPs, cleavable linkers are cleaved, thereby removing the mask and allowing the masked antigen-binding constructs to bind their target antigens in a tumor microenvironment-specific manner.

[0146] In some embodiments, the masked antigen-binding construct and its preparation method offer the following advantages.

[0147] First, the efficiency of virtual masking screening is improved. Protein structure determines protein function. When the masking structure consists of small polypeptides with cyclic secondary structures, its structure cannot be accurately predicted in most cases due to the lack of homologous template information, even with AlphaFold2, the best prediction tool in the field of protein folding. However, when the secondary structure of the masking structure is chosen to be α-helix, AlphaFold2 is more likely to predict the correct protein structure during computer-based virtual screening. The correct folding structure is more conducive to the computational analysis of protein-protein interactions and can significantly improve the screening efficiency of masking structures.

[0148] Secondly, the development of single-domain antibody masking structures is simplified. α-helical masking structures can be easily attached to one end of a single-domain antibody, achieving steric hindrance without the need for complex molecular structures. Furthermore, although α-helical structures are easy to fold accurately, no α-helical masking structure with the ability to bind to target proteins to provide steric hindrance has been developed to date. Therefore, such masking structures can reduce the development difficulty of single-domain antibody masking structures.

[0149] Masking structures can be removed more easily through enzymatic digestion. Typically, to achieve antibody masking, sterically hindered masking structures (such as albumin masking structures) or high-affinity circular masking structures are used. However, masking structure molecules are often too large, leading to excessively high immunogenicity. Furthermore, the computational difficulty of circular masking structures makes it challenging to select suitable ones. Moreover, circular masking structures are usually short peptides. To achieve a masking effect, circular masking structures often have extremely strong binding affinity to the target protein, making them difficult to remove enzymatically. In such cases, the affinity of the antibody remains very low after cleavage, and once cleaved, the masking antibody fails to perform its intended function. Unbound by theory, this disclosure employs α-helical masking structures with weak binding affinity to the target protein (e.g., antibodies) to provide steric hindrance. Such α-helical masking structures achieve ideal masking effects and can be easily removed by enzymatic digestion.

[0150] Masking antigen-binding constructs In a masked antigen-binding construct (e.g., an antibody and its antigen-binding fragment), the antigen-binding construct may be coupled to an α-helix masking structure that prevents the masked antigen-binding construct from binding to its target antigen. The masked antigen-binding construct may comprise or consist of an antigen-binding domain. The α-helix masking structure may comprise a single α-helix structure. The α-helix masking structure may comprise two or more α-helix structures (e.g., 2, 3, 4, or 5 α-helix structures). In some embodiments, two or more α-helix structures are linked together by a polypeptide, and a linker may optionally be included between adjacent α-helix structures.

[0151] The α-helical masking structure can be linked to the antigen-binding portion of the masked antigen-binding construct via a cleavable linker. This cleavable linker may contain a cleavable site that can be recognized by an enzyme. In the masked antigen-binding construct, the α-helical structure acts as a masking structure to inhibit the binding between the masked antigen-binding construct and its target antigen. Compared to a control antigen-binding construct without an α-helical masking structure, the binding between the masked antigen-binding construct and its target antigen can be reduced by at least 10%, 20%, or 30%. Compared to an activated antigen-binding construct whose cleavable site has been cleaved (e.g., by enzymatic digestion), the binding between the masked antigen-binding construct and its target antigen can be reduced by at least 10%, 20%, or 30%.

[0152] In a masked antigen-binding construct, an α-helical masking structure can be attached to the antigen-binding construct via a cleavable linker. This masked antigen-binding construct can be an antibody. Typical antibodies include a heavy chain variable region and a light chain variable region. The α-helical masking structure can be attached to either (1) the N-terminus of the heavy chain variable region or (2) the N-terminus of the light chain variable region. Bivalent antibodies have two binding sites, which may be the same or different. In normal monoclonal antibodies, the binding sites are the same, and the antibody has two identical pairs of light and heavy chains. In bispecific antibodies, the binding sites are different, formed by two different pairs of light and heavy chains.

[0153] Bivalent antibodies containing two light chain and heavy chain pairs can be used, wherein the N-terminus of one or more light chains and / or heavy chains is connected to an α-helical masking structure via a cleavable linker, thereby reducing the binding affinity of the variable regions of the light and heavy chains to the target.

[0154] The two light and heavy chain pairs may be identical or different. The light chain may contain a light chain variable region and a light chain constant region, and the heavy chain may contain a heavy chain variable region and a heavy chain constant region. The heavy chain region may include CH1, hinge, CH2, and CH3 regions. The light chain may link to a first heteropeptide, and the heavy chain may link to a second heteropeptide. The cleavable linker may have an MMP1 or MMP2 cleavage site.

[0155] The antigen-binding affinity of the masked antigen-binding construct may be reduced by at least 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 95%, 96%, 97%, 98%, 99%, 99.5%, 99.6%, or 99.7% in the presence of the α-helical masking structure compared to the control antigen-binding construct without the α-helical masking structure. In some embodiments, the ratio (1) of the binding affinity without the α-helical masking structure to (2) the binding affinity with the α-helical masking structure (“affinity reduction ratio”) is at least 1.2, 1.3, 1.4, 1.5, 2, 3, 4, 5, 6, 7, 8, 9, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 200, 300, 400, 500, or 1000. In some embodiments, the affinity reduction rate is less than 1.2, 1.3, 1.4, 1.5, 2, 3, 4, 5, 6, 7, 8, 9, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 200, 300, 400, 500, or 1000.

[0156] The cytotoxicity of the masked antigen-binding construct is reduced by at least 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 95%, 96%, 97%, 98%, 99%, 99.5%, 99.6%, or 99.7% in the presence of the α-helical masking structure. In some embodiments, the ratio (1) to the cytotoxicity without the α-helical masking structure and (2) to the cytotoxicity with the α-helical masking structure (“cytotoxicity reduction ratio”) is at least 1.2, 1.3, 1.4, 1.5, 2, 3, 4, 5, 6, 7, 8, 9, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 200, 300, 400, 500, or 1000. In some implementations, the cytotoxicity reduction rate is less than 1.2, 1.3, 1.4, 1.5, 2, 3, 4, 5, 6, 7, 8, 9, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 200, 300, 400, 500, or 1000.

[0157] Antigen binding affinity can be deduced from the ratio of kinetic rate constants (KD = koff / kon). In some embodiments, the KD of the antigen-binding portion of the masked antigen-binding construct without the α-helical masking structure is less than 1 × 10⁻⁶. - 6 M, less than 1×10 -7 M, less than 1×10 -8 M, less than 1×10 -9 M, less than 1×10 -10 M, less than 1×10 -11 M, less than 1×10 - 12 M, less than 1×10 -13 M or less than 1×10 -14 M. In some embodiments, the KD of the antigen-binding portion of the masked antigen-binding construct is less than 50 nM, 30 nM, 20 nM, 15 nM, 10 nM, 9 nM, 8 nM, 7 nM, 6 nM, 5 nM, 4 nM, 3 nM, 2 nM, or 1 nM. In some embodiments, the KD of the antigen-binding portion of the masked antigen-binding construct is greater than 1 × 10⁻⁶. -7 M, greater than 1×10 -8 M, greater than 1×10 -9 M, greater than 1×10 -10 M, greater than 1×10 -11 M, greater than 1×10 -12 M, greater than 1×10 -13 M or greater than 1×10 -14 M.

[0158] In some embodiments, the KD of the masked antigen-binding construct having an α-helix masking structure is less than 1 × 10⁻⁶. -6 M, less than 1×10 -7 M, less than 1×10 -8 M, less than 1×10 -9 M, less than 1×10 -10 M, less than 1×10 -11 M, less than 1×10 -12 M, less than 1×10 -13 M or less than 1×10 -14 M. In some embodiments, the KD of the masked antigen-binding construct is less than 50 nM, 30 nM, 20 nM, 15 nM, 10 nM, 9 nM, 8 nM, 7 nM, 6 nM, 5 nM, 4 nM, 3 nM, 2 nM, or 1 nM. In some embodiments, the KD of the masked antigen-binding construct is greater than 1 × 10⁻⁶. -7 M, greater than 1×10 -8 M, greater than 1×10 -9 M, greater than 1×10 -10 M, greater than 1×10 -11 M, greater than 1×10 -12 M, greater than 1×10 -13 M or greater than 1×10 -14 M.

[0159] In some embodiments, the α-helical masking structure comprises at least 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, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, or 50 amino acids. In some embodiments, the α-helical masking structure comprises no more than 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, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, or 50 amino acids. In some embodiments, the α-helical masking structure comprises 5-40 amino acids. In some embodiments, the α-helical masking structure comprises 5-20 amino acids. In some embodiments, the α-helical masking structure comprises about 13-28 amino acids. In some embodiments, the α-helical masking structure comprises about 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, or 28 amino acids. In some embodiments, the α-helical masking structure is a single polypeptide. The bispecific antibody described herein may be designed to have an IgG (e.g., IgG1 isoform) structure with a “knobs-into-holes” (KIH) mutation, which helps to promote heterodimer formation and avoid mispairing between the two heavy chains. In some embodiments, the bispecific antibody may have a common light chain.

[0160] The α-helical masking structure can be linked to the variable region of an antibody via a cleavable linker (e.g., containing a cleavable site). The same cleavable linker or site can be used to link each heavy or light chain variable region of an antigen-binding construct (e.g., an antibody or its antigen-binding fragment) to the α-helical masking structure. This cleavable linker or site can be cleaved by proteases present extracellularly in the intended target tissue or pathology (e.g., cancer), thereby releasing the antigen-binding portion of the masked antigen-binding construct from the α-helical masking structure, allowing the antigen-binding portion of the masked antigen-binding construct to bind to its intended target, such as a cell surface antigen or a soluble ligand.

[0161] In addition to the variable region, the masked antigen-binding construct may also contain all or part of a constant region, which may include any one or more of the light chain constant region, CH1, hinge region, CH2, and CH3 regions. As with other antigen-binding constructs, one or more carboxyl-terminal residues may be proteased or derivatized.

[0162] The α-helix can be attached to the N-terminus of the antibody light chain via a breakable linker. For example, structural simulation results for CEA+TAm-R1-L16, CEA+TAm-R1-L17, and N_L_Pani_mask70 are shown below. Figures 1A-1C As shown.

[0163] The α-helix can be attached to the N-terminus of the antibody heavy chain via a breakable linker. For example, structural simulation results for CEA+TAm-R1-H1, CEA+TAm-R1-H4, CEA+TAm-R1-H5, N_H_Cet-mask72, N_H_Pani_mask1, Tra-R2-2, and Tra-R2-4 are shown below. Figures 1D-1J As shown.

[0164] When the α-helical masking structure is linked to the heavy chain, the masking antigen-binding construct may include: The first polypeptide, from the N-terminus to the C-terminus, consists of an α-helical masking structure, a first linker peptide, a cleavable linker peptide, a second linker peptide, and a TAm heavy chain. (b) and a second polypeptide comprising a TAm light chain sequentially from the N-terminus to the C-terminus.

[0165] When the α-helical masking structure is linked to a light chain, the masking antigen-binding construct may include: The first polypeptide, from the N-terminus to the C-terminus, consists of an α-helical masking structure, a first linker peptide, a cleavable linker peptide, a second linker peptide, and a TAm light chain. (b) and a second polypeptide comprising the TAm heavy chain sequentially from the N-terminus to the C-terminus.

[0166] The masking antigen binding construct may have an α-helical masking sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 12, 16, 18, or 20.

[0167] The masked antigen binding construct may have a first linker sequence whose identity with SEQ ID NO: 13, 21 or 23 is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99% or 100%.

[0168] The masked antigen binding construct may have a second linker sequence whose identity with SEQ ID NO: 13 or 21 is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100%.

[0169] The masked antigen binding construct may have a cleavable linker sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 14 or 24.

[0170] The masked antigen binding construct may have a light chain sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 10.

[0171] The masked antigen binding construct may have a heavy chain sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 6.

[0172] The conjugation of an antigen-binding construct to an α-helical masking structure can result in a reduction in binding affinity of at least about 10%, 20%, 30%, 40%, 50%, 100%, 2.5, 5, 10, 15, 20, 25, 30, 40, 50, 100, 200, 500, 750, or 1000 times compared to the same antigen-binding construct without such conjugation or with such conjugation cleaved. In some of these antigen-binding constructs, the reduction in binding affinity is less than about 10%, 20%, 30%, 40%, 50%, 100%, 2.5, 5, 10, 15, 20, 25, 30, 40, 50, 100, 200, 500, 750, or 1000 times. The effector function of a masked antigen-binding construct (such as ADCC, phagocytosis, CDC, or drug-induced cytotoxicity in antibody-drug conjugates (ADCs)) may be reduced by the same factors or to the same extent. After exposing a masked antigen-binding construct by proteolytic cleavage or otherwise removing the mask, the restored antigen-binding construct typically retains affinity or effector function that remains within 2- to 1.5-fold, or more preferably substantially unchanged, within the experimental error range compared to another otherwise identical unmasked control antigen-binding construct.

[0173] Common techniques for measuring the affinity of masked antigen-binding constructs for antigens include, for example, ELISA, RIA, biolayer interferometer (BLI), and surface plasmon resonance (SPR).

[0174] The affinity of the masked antigen-binding construct for the target antigen can be determined by ELISA (e.g., using the method described in Example 4). The binding EC50 of the masked antigen-binding construct can be greater than 1, 2, 3, 4, 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, or 100 μg / ml. The binding EC50 of the masked antigen-binding construct can be less than 1, 2, 3, 4, 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, or 100 μg / ml. The binding EC50 of the masked antigen-binding construct can be 10-50 μg / ml or 10-100 μg / ml. Linking the antigen-binding portion of a masked antigen-binding construct to an α-helical masking structure may result in a decrease in the binding affinity of that antigen-binding portion by at least approximately 10%, 20%, 30%, 40%, 50%, 100%, 2.5, 5, 10, 15, 20, 25, 30, 40, 50, 100, 200, 500, 750, or 1000 times compared to the same antigen-binding construct without such linkage or to an antigen-binding construct after such linkage has been cleaved. Linking the antigen-binding portion of a masked antigen-binding construct to an α-helical masking structure may result in a decrease in the binding affinity of that antigen-binding portion by approximately 9–27 times compared to the same antigen-binding construct without such linkage or to an antigen-binding construct after such linkage has been cleaved.

[0175] The affinity of the masked antigen-binding construct for the target antigen can be determined by BLI (e.g., using the method described in Example 6). The affinity can be calculated from the quotient of the kinetic rate constant (KD = kdis / ka). The KD of the masked antigen-binding construct can be less than 1 × 10⁻⁶. -6 M, less than 1×10 -7 M, less than 1×10 -8 M, less than 1×10 -9 M or less than 1×10 - 10 M. The KD of the masked antigen-binding construct can be less than 50 nM, 30 nM, 20 nM, 15 nM, 10 nM, 9 nM, 8 nM, 7 nM, 6 nM, 5 nM, 4 nM, 3 nM, 2 nM, or 1 nM. The KD of the masked antigen-binding construct may be greater than 1 × 10⁻⁶. -7 M, greater than 1×10 -8 M, greater than 1×10 -9 M, greater than 1×10 -10 M, greater than 1×10 -11 M or greater than 1×10 -12M. Linking the antigen-binding portion of a masked antigen-binding construct to an α-helical masking structure can result in a reduction in the binding affinity of the antigen-binding portion of the masked antigen-binding construct by at least about 10%, 20%, 30%, 40%, 50%, 100%, 2.5 times, 5 times, 10 times, 15 times, 20 times, 25 times, 30 times, 40 times, 50 times, 100 times, 200 times, 500 times, 750 times, or 1000 times compared to the same antigen-binding construct without such linkage or with such linkage cleaved. Linking the antigen-binding portion of a masked antigen-binding construct to an α-helical masking structure can result in a reduction in the binding affinity of the antigen-binding portion of the masked antigen-binding construct by about 20%-60% or 30%-60% compared to the same antigen-binding construct without such linkage or with such linkage cleaved.

[0176] The masking antigen-binding construct can be a masking antibody. The masking antigen-binding construct may have a heavy chain sequence having at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identity with any heavy chain sequence shown in Table 9. The masking antigen-binding construct may have a heavy chain sequence having at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identity with SEQ ID NO: 27, 28, 31, 32, or 35. The masking antigen-binding construct may have a light chain sequence having at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identity with any of the light chain sequences shown in Table 9. The masking antigen-binding construct may have a light chain sequence having at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identity with SEQ ID NO: 26, 30, 34, or 33.

[0177] The masked antigen binding construct may have an amino acid sequence of any one of SEQ ID NO: 3, 4 or 8, or an amino acid sequence that has at least 95%, 97% or 99% identity with the amino acid sequence of any one of SEQ ID NO: 3, 4 or 8.

[0178] The masked antigen-binding construct may have an amino acid sequence of any one of SEQ ID NO: 11, 15, 17, 19, 22, 27, 28, 31, 34 and 35, or an amino acid sequence that is at least 95%, 97% or 99% identical to the amino acid sequence of any one of SEQ ID NO: 11, 15, 17, 19, 22, 27, 28, 31, 34 and 35.

[0179] The masking antigen-binding construct may include a heavy chain constant region, such as a constant region of IgG1, IgG2, IgG3, IgG4, IgA, IgE, IgM, or IgD. Preferably, the heavy chain constant region is an IgG1 heavy chain constant region or an IgG4 heavy chain constant region. Furthermore, the masking antigen-binding construct may also include a light chain constant region, which is a κ light chain constant region or a λ light chain constant region.

[0180] This masked antigen-binding construct may include an engineered Fc region. It is known in the art that antibody effector functions can be altered by replacing amino acid residues in the Fc moiety. The Fc moiety of an antibody mediates a variety of important effector functions, such as cytokine induction, antibody-dependent cytotoxicity (ADCC), phagocytosis, complement-dependent cytotoxicity (CDC), and the half-life / clearance of the antibody and antigen-antibody complex.

[0181] The masking antigen-binding construct can be derivatized or linked to another functional molecule (e.g., another peptide or protein). For example, the labeled antigen-binding constructs described herein can be obtained by functionally linking the antigen-binding construct (through chemical coupling, gene fusion, non-covalent binding, or other means) to one or more other molecular entities, such as another antigen-binding construct (e.g., a bispecific antibody or divalent antibody), a detectable agent, a cytotoxic agent, a pharmaceutical agent, and / or a protein or peptide capable of mediating the binding of the masking antigen-binding construct to other molecules (e.g., a streptavidin core region or a polyhistidine tag).

[0182] This masked antigen-binding construct can be further modified to generate glycosylation site mutants, in which the O-linked or N-linked glycosylation sites of the binding protein have been mutated.

[0183] The glycosylation of a masked antigen-binding construct can be modified. For example, a glycosylated-free masked antigen-binding construct (i.e., a construct lacking glycosylation) can be prepared. Glycosylation modification can be used to improve the affinity of the masked antigen-binding construct for antigens, etc. Such glycan modifications can be achieved by altering one or more glycosylation sites within the sequence of the masked antigen-binding construct. For example, one or more amino acid substitutions can be made to eliminate one or more variable region glycosylation sites, thereby removing the glycosylation at that site. This glycosylation-free modification can improve the affinity of the masked antigen-binding construct for antigens.

[0184] Furthermore, or as an alternative, the masked antigen-binding construct can be further modified with different types of glycosylation, such as a masked antigen-binding construct with low fucosylation (reduced number of fucose residues) or a masked antigen-binding construct with increased bilateral GlcNAc structures. Such altered glycosylation patterns have been shown to enhance the ADCC (antibody-dependent cell-mediated cytotoxicity) capacity of the masked antigen-binding construct. Such glycan modifications can be achieved, for example, by expressing the masked antigen-binding construct in host cells with altered glycosylation mechanisms.

[0185] The masked antigen-binding construct may not have a functional Fc region. For example, it can be Fab, Fab', F(ab')2, VHH, ScFv, or a fusion protein thereof. The masked antigen-binding construct may also have a functional Fc region. For example, its antigen-binding portion can be Fab-Fc, VHH-Fc, ScFv-Fc, or a fusion protein thereof. The Fc region may contain LALA mutations (L234A and L235A mutations in EU designations) or LALA-PG mutations (L234A, L235A, and P329G mutations in EU designations). The masked antigen-binding construct can have any structure shown in Figure 4-14, or it can adopt an extracellular domain to form any structure shown in Figure 15.

[0186] As shown in Figure 4, in some embodiments, the antigen-binding portion of the masked antigen-binding construct may comprise or be composed of Fab. In some embodiments, the masking structure is connected to the VH of the Fab. In some embodiments, the masking structure is connected to the VL of the Fab. In some embodiments, the antigen-binding construct may comprise or be composed of ScFv. In some embodiments, the masking structure is connected to the VH of the ScFv. In some embodiments, the masking structure is connected to the VL of the ScFv. In some embodiments, the antigen-binding portion of the masked antigen-binding construct may comprise or be composed of VHH. In some embodiments, the masking structure is connected to VHH.

[0187] As shown in Figure 5, in some embodiments, the antigen-binding portion of the masked antigen-binding construct may comprise or consist of a full-length antibody (including two Fabs linked to the Fc). In some embodiments, the masking structure is linked to the VH of the full-length antibody. In some embodiments, the masking structure is linked to the VL of the full-length antibody. In some embodiments, the antigen-binding portion of the masked antigen-binding construct may comprise or consist of a Fab-Fc (including one Fab linked to the Fc). In some embodiments, the masking structure is linked to the VH of the Fab. In some embodiments, the masking structure is linked to the VL of the Fab. In some embodiments, the masking structure is linked to the Fc (e.g., through the hinge region of the Fc).

[0188] As shown in Figure 6, in some embodiments, the antigen-binding portion of the masked antigen-binding construct may comprise or consist of a VHH-Fc (e.g., one or two VHHs connected to the Fc). In some embodiments, the masking structure is connected to one VHH of the VHH-Fc. In some embodiments, the masking structure is connected to each of the two VHHs of the VHH-Fc. In some embodiments, the masking structure is connected to the Fc (e.g., through the hinge region of the Fc).

[0189] As shown in Figures 7A-7B, in some embodiments, the antigen-binding portion of the masked antigen-binding construct may comprise or consist of ScFv-Fc (e.g., one or two ScFvs connected to Fc). In some embodiments, the masking structure is connected to one of the ScFv-Fc (e.g., through the VL or VH of the ScFv). In some embodiments, the masking structure is connected to each of the ScFv-Fc (e.g., through the VL or VH of the ScFv). In some embodiments, the masking structure is connected to Fc (e.g., through the hinge region of Fc).

[0190] As shown in Figures 8A-8B, in some embodiments, the antigen-binding portion of the masked antigen-binding construct may comprise or be composed of a Fab-Fc fusion protein. In some embodiments, the Fab-Fc fusion protein comprises: (1) a Fab connected to an Fc; (2) a Fab connected to an Fc and a ScFv connected to the same Fc; or (3) a Fab connected to an Fc and a VHH connected to the same Fc. In some embodiments, the masking structure is connected to the VH of the Fab. In some embodiments, the masking structure is connected to the VL of the Fab. In some embodiments, the masking structure is connected to the Fc (e.g., through the hinge region of the Fc). In some embodiments, the masking structure is connected to the VH of the ScFv. In some embodiments, the masking structure is connected to the VL of the ScFv. In some embodiments, the masking structure is connected to the VHH.

[0191] As shown in Figure 9, in some embodiments, the antigen-binding portion of the masked antigen-binding construct may comprise or consist of the ScFv-Fc fusion protein. In some embodiments, the ScFv-Fc fusion protein comprises an ScFv linked to an Fc and a VHH linked to the same Fc. In some embodiments, the masking structure is linked to the VH of the ScFv. In some embodiments, the masking structure is linked to the VL of the ScFv. In some embodiments, the masking structure is linked to the VHH.

[0192] As shown in Figure 10, in some embodiments, the antigen-binding portion of the masked antigen-binding construct may comprise or consist of F(ab')2 (e.g., two Fabs connected to the hinge region). In some embodiments, the masking structure is connected to the VH of each Fab in F(ab')2. In some embodiments, the masking structure is connected to the VL of each Fab in F(ab')2. In some embodiments, the masking structure is connected to the hinge region.

[0193] As shown in Figure 10, in some embodiments, the antigen-binding portion of the masked antigen-binding construct may comprise or be composed of a Fab' (e.g., a Fab connected to the hinge region). In some embodiments, the masking structure is connected to the VH of the Fab'. In some embodiments, the masking structure is connected to the VL of the Fab'. In some embodiments, the masking structure is connected to the hinge region.

[0194] As shown in Figure 11, in some embodiments, the antigen-binding portion of the masked antigen-binding construct may include or consist of a VHH-hinge region (e.g., one or two VHHs connected to the hinge region). In some embodiments, the masking structure is connected to one VHH of the VHH-hinge region. In some embodiments, the masking structure is connected to each of the two VHHs of the VHH-hinge region. In some embodiments, the masking structure is connected to the hinge region.

[0195] As shown in Figures 12A-12B, in some embodiments, the antigen-binding portion of the masked antigen-binding construct may include or consist of an ScFv-hinge region (e.g., one or two ScFvs connected to the hinge region). In some embodiments, the masking structure is connected to one ScFv of the ScFv-hinge region (e.g., through the VL or VH of the ScFv). In some embodiments, the masking structure is connected to each of the two ScFvs of the ScFv-hinge region (e.g., through the VL or VH of the ScFv). In some embodiments, the masking structure is connected to the hinge region.

[0196] As shown in Figures 13A-13B, in some embodiments, the antigen-binding portion of the masked antigen-binding construct may comprise or be composed of a Fab fusion protein. In some embodiments, the Fab fusion protein comprises: (1) a Fab connected to a hinge region; (2) a Fab connected to a hinge region and an ScFv connected to the same hinge region; or (3) a Fab connected to a hinge region and a VHH connected to the same hinge region. In some embodiments, the masking structure is connected to the VH of the Fab. In some embodiments, the masking structure is connected to the VL of the Fab. In some embodiments, the masking structure is connected to a hinge region. In some embodiments, the masking structure is connected to the VH of the ScFv. In some embodiments, the masking structure is connected to the VL of the ScFv. In some embodiments, the masking structure is connected to the VHH.

[0197] As shown in Figure 14, in some embodiments, the antigen-binding portion of the masked antigen-binding construct may comprise or consist of the ScFv fusion protein. In some embodiments, the ScFv fusion protein comprises an ScFv connected to a hinge region and a VHH connected to the same hinge region. In some embodiments, the masking structure is connected to the VH of the ScFv. In some embodiments, the masking structure is connected to the VL of the ScFv. In some embodiments, the masking structure is connected to the VHH.

[0198] As shown in Figure 15, in some embodiments, the masked antigen-binding construct may be linked to the extracellular domain of a transmembrane protein. In some embodiments, the antigen-binding portion of the masked antigen-binding construct may comprise, or consist of, ScFv, VHH, or a VHH linked to ScFv. In some embodiments, the masking structure is linked to the VH of ScFv. In some embodiments, the masking structure is linked to the VL of ScFv. In some embodiments, the masking structure is linked to the VHH.

[0199] The relevant biological activities and / or antibody properties of the masked antigen-binding construct can be detected. Determining which properties constitute relevant antigen-binding construct characteristics depends on the specific circumstances. Non-limiting examples of antigen-binding construct characteristics include, but are not limited to: antigenicity, immunogenicity, immunomodulatory activity, expression of the masked antigen-binding construct in a homologous host, expression in a heterologous host, expression in plant cells, sensitivity to in vitro post-translational modifications, and sensitivity to in vivo post-translational modifications. Compared to the initial antigen-binding construct, the masked antigen-binding construct may possess comparable or superior relevant biological activities and / or antigen-binding construct characteristics.

[0200] The masked antigen-binding constructs or fragments thereof constructed and / or identified according to this disclosure can be tested in vitro and / or in vivo for their ability to regulate cellular biological activity. Such abilities can be assessed, for example, by: detecting antigen and gene expression; detecting cell proliferation; detecting activation of signaling molecules (such as signal transduction factors and kinases); detecting effector function of cells; or detecting cell differentiation. Techniques known to those skilled in the art can be used to measure these activities. For example, cell proliferation can be detected by a 3H-thymidine incorporation assay and trypan blue cell counting. Antigen expression can be detected by immunoassays, including but not limited to competitive and non-competitive assay systems using the following techniques: Western blotting, immunohistochemical radioimmunoassay, enzyme-linked immunosorbent assay (ELISA), sandwich immunoassay, immunoprecipitation assay, precipitation reaction, gel diffusion precipitation reaction, immunodiffusion assay, agglutination assay, complement fixation assay, immunoradioassay, fluorescence immunoassay, protein A immunoassay, and flow cytometry (FACS) analysis. Activation of signaling molecules can be detected by, for example, kinase assays and electrophoretic mobility shift assays (EMSAs).

[0201] The masked antigen-binding constructs, fragments thereof, or combinations thereof described herein are preferably tested in vitro and then in vivo for their desired therapeutic or preventative activity before use in humans. For example, assays that can be used to determine whether a particular pharmaceutical composition needs to be administered include cell culture assays, in which a patient tissue sample is grown in a culture and exposed to or otherwise contacted with the pharmaceutical composition, and then the effect of the composition on the tissue sample is observed. Tissue samples can be obtained from a patient biopsy. This test allows for the identification of the most therapeutically effective treatment (e.g., a prophylactic or therapeutic agent) for each individual patient. In vitro assays can be performed using representative cells representing cell types involved in a specific condition to determine whether the pharmaceutical compositions described herein have the desired effect on such cell types. For example, cell lines can be used for in vitro assays.

[0202] Design an α-spiral shielding structure An α-helix library can be constructed by extracting α-helix substructures or cyclic helical bundle structures from a protein structure database.

[0203] Candidate α-helices that can bind to a CDR (clustered receptor) of an initiating antigen-binding domain (e.g., an antigen-binding domain in an antigen-binding construct) can be screened from a library using a virtual screening algorithm. Screening of α-helices may include one or more of the following steps: 1) The structural information of the initiation antigen-binding domain (e.g., VH-VL pair, or VHH) to be masked is input into a protein language model (e.g., the ProtGPT2 model) or a diffusion model (e.g., a 3D Equivariant Diffusion model) to generate candidate α-helical masking structures that can bind to the CDR region of the initiation antigen-binding domain. In some cases, supervised fine-tuning can be performed on the input data. A language model trained using a protein database is called a protein language model. Protein language models such as ESM2 and OmegaPLM can be used for structure prediction. Protein language models such as ProtGPT2 and ProGen can be used for the design of novel functional proteins. In this disclosure, the ProtGPT2 architecture has been modified and conditional branches have been added. Supervised fine-tuning (SFT) can be performed on ProtGPT2 by inputting the three-dimensional structure of the antigen-binding domain to generate α-helical candidate sequences that can bind to the CDR region of the antigen-binding domain. Protein language models (e.g., the ProtGPT2 model) can be used for the design and engineering of novel proteins. Protein language models (e.g., the ProtGPT2 model) can be based on transformer architectures (e.g., the GPT2transformer architecture). Fine-tuning training can be performed using paired target sequences and α-helical sequences to enable the model to generate α-helical sequences targeting specific targets (e.g., CDR regions). During model inference, the model can be provided with specified target locations to generate a large number of candidate α-helical sequences. For diffusion models, the structure of the candidate target can be input along with specified binding sites, causing the diffusion model to generate α-helical structures and sequences in the binding site regions.

[0204] 2) In the computer, the candidate α-helical masking structures generated by the model are assembled into the form of "α-helical masking structure + cleavable linker + antigen binding domain", and the structure is predicted using fast folding software such as Alphafold2, ESMFold, and OmegaFold to select α-helical masking structures that can interact with the CDR region of the initiating antigen binding domain. The selection criteria include: (1) the pLDDT (predicted local distance difference test) of the α-helix is ​​greater than 55 (preferably greater than 65, further preferably greater than 80, and even more preferably greater than 90); (2) the pTM (predicted template modeling score) of the α-helix is ​​greater than 0.5 (preferably greater than 0.6, further preferably greater than 0.75, and even more preferably greater than 0.9); (3) the ipTM (interface pTM score) between the α-helix and the antigen-binding domain CDR is greater than 0.45 (preferably greater than 0.6, further preferably greater than 0.75, and even more preferably greater than 0.9); and / or (4) the average minimum distance between the α carbon atom of the α-helix and the CDR is less than 15 Å (preferably less than 14 Å, further preferably less than 12 Å, even more preferably less than 10 Å, even more preferably less than 8 Å, even more preferably less than 6 Å, and even more preferably less than 4 Å).

[0205] The masked antigen-binding domains can then be produced through molecular cloning for virtual screening and use in laboratory testing.

[0206] The calculation methods for pLDDT and ipTM can be found in Jumper et al.'s "Highly accurate protein structure prediction with AlphaFold" (Nature 596.7873(2021):583-589), which is incorporated herein by reference in its entirety. "α-helix pLDDT" refers to calculating pLDDT only for the predicted structure of the masked α-helix, and this value can be greater than 55. "α-helix ptm" is used to assess the accuracy of α-helix structure prediction, and this value can be greater than 0.5. "ipTM" is used to assess the correctness of docking between the masked α-helix sequence and the antigen-binding domain, and this value can be greater than 0.45. "Distance" measures the spatial distance between the masked α-helix and the CDR of the antigen-binding domain, and this value can be less than 15 Å. The calculation methods are as follows: Where dij represents the Euclidean distance from the i-th amino acid of the α-helix to the j-th amino acid of the CDR region of the antigen-binding domain.

[0207] In step (2), various diffusion models can be used. A diffusion model is a generative model. Currently, there are four main types of generative models: Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), flow-based models, and diffusion models. Diffusion models have surpassed the original state-of-the-art (SOTA) GANs in image generation tasks and have produced high-quality image generation models such as DALLE 2, Imagen, and stable-diffusion. It also performs well in many other application areas, such as computer vision, natural language processing, waveform signal processing, multimodal modeling, and molecular graph modeling.

[0208] The diffusion model mainly includes a forward process and a backward process. The forward process (diffusion process) continuously adds noise (standard Gaussian distribution) to the input data. As the time step increases, the added noise increases, and eventually, when the time step approaches positive infinity, the noise becomes pure noise. The backward process starts from random noise and gradually recovers the original data—the denoising process is the process of generating the target.

[0209] Diffusion models have also been introduced into drug design and molecular discovery, for example, for molecular docking (DIFFDOCK), small molecule drug generation (DiffSBDD), and protein design (RFDiffusion). In physics, chemistry, and biology, it is necessary to describe various objects with geometric features, such as position vectors (e.g., atomic coordinates of molecules) and force vectors. The equations describing these geometric features are independent of the choice of coordinate system. The underlying layer of these geometric features is the symmetry information of the system. Therefore, how to maintain these geometric features and symmetries in neural networks is key to the application of artificial intelligence to natural science problems. Through model design, isovariant neural networks can maintain geometric features and isovariance during the data flow from input to output, exhibiting lower data requirements and better performance than traditional neural networks. To this end, a 3D isovariant diffusion model was developed for the co-design of sequence and structure for scaffold functional proteins targeting specific protein regions. The conformation of the antigen-binding domain can be input. The CDR to be bound can be specified. This 3D isovariant diffusion model can be used to generate end-to-end sequences and conformations of candidate α-helices that bind to the CDR region of the antigen-binding domain.

[0210] The design of an α-spiral shielding structure can be completed within 1 minute. The design of an α-spiral shielding structure can be completed within 1 day.

[0211] Fragmentable sites and connectors Cleavable linkers can be used to couple α-helical masking structures to antigen-binding constructs (e.g., antibodies or their antigen-binding fragments). A cleavable linker may contain a cleavable site and optionally two linker fragments (e.g., located before and after the cleavable site). The cleavable linker can be any amino acid fragment conventionally used for linking peptide domains. Suitable cleavable linkers are of variable length, such as 1-20, 2-15, 3-12, 4-10, 6-20, 6-10 amino acids, or specifically 5, 6, 7, 8, 9, or 10 amino acids. In some embodiments, the cleavable linker comprises at least 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, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, or 50 amino acids. In some embodiments, the cleavable linker comprises no more than 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, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, or 50 amino acids. In some embodiments, the cleavable linker comprises about 14 to 30 amino acids. In some embodiments, the cleavable linker comprises about 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, or 30 amino acids. In some embodiments, the cleavable site comprises about 6 to 10 amino acids. In some embodiments, the cleavable site in the cleavable linker contains about 6, 7, 8, 9, or 10 amino acids.

[0212] Cleavable sites can be recognized and cleaved by extracellularly expressed proteases, enabling them to access masked antigen-binding constructs (e.g., antigen-binding domains), releasing the masked antigen-binding constructs and allowing them to access their targets (e.g., receptor extracellular domains or soluble ligands). Multiple matrix metalloproteinase (MMP) recognition sites (MMP1-28) may be applicable. MMPs play a role in tissue remodeling and are associated with tumorigenesis processes such as morphogenesis, angiogenesis, and metastasis. Examples of MMPs can be found in U.S. Patent Publication 2013 / 0309230, International Patent Publications WO 2009 / 025846, WO 2010 / 081173, WO 2014 / 107599, WO2015 / 048329, U.S. Patent Publication 20160160263, and Ratnikov et al. (Proc. Natl. Acad. Sci. USA, 111: E4148-E4155 (2014)); all of which are incorporated herein by reference in their entirety.

[0213] Cleavable sites can be cleaved within the tumor microenvironment. These cleavable sites can be matrix metalloproteinase (MMP) cleavage sites. MMP cleavage sites can be selected from MMP2, MMP7, MMP9, and MMP13 cleavage sites. After MMP cleavage, the heavy and / or light chains of the masking antigen-binding construct can contain amino acid residue fragments from the MMP cleavage site. These residue fragments can contain the sequence LRSG, SG, or VR at the N-terminus of the antigen-binding portion of the masking antigen-binding construct.

[0214] Cleavage sites can also be cleaved at bibasic sites (e.g., arginine-arginine, lysine-arginine, or lysine-lysine sites). Enzymes known in the art to cleave at bibasic sites include, for example, N-arginine bibasic convertases (Chow et al., JBC 275: 19545-19551 (2000)) and subtilisin-like proprotein convertases, such as furin (PCI), PC2, and PC3 (Steiner (1991, in Peptide Biosynthesis and Processing (Fricker ed.), pp. 1-16, CRC Press, Boca Raton, Florida; Muller et al., JBC 275:39213-39222, (2000)).

[0215] Furthermore, one or more linkers can be used to couple α-helical masking structures to antigen-binding constructs. Glycine and glycine-serine polymers are relatively unstructured and therefore can be used as neutral linkers between components. Glycine has significantly more phi-psi space than alanine and is much less restricted than residues with longer side chains (see Scheraga, Rev. Computational Chem. 11173-142 (1992)). Some exemplary linkers are in the form S(G)nS, where n is 5-20. Other exemplary linkers include (G)n, glycine-serine polymers (including, for example, (GS)n, (GSGGS)n (SEQ ID NO: 36)), glycine-alanine polymers, alanine-serine polymers, and other flexible linkers.

[0216] Provide initiator antigen binding construct The method disclosed herein can use an initiating antigen-binding construct (e.g., an antibody or its antigen-binding fragment) as the starting material (“initiating antigen-binding construct”). An α-helical masking structure can be designed targeting the antigen-binding domain (e.g., the initiating antigen-binding domain) in the initiating antigen-binding construct.

[0217] The initiating antigen-binding construct can be non-human, humanized, human, chimeric, or surface-remodeled antibody. The initiating antigen-binding construct can be nanobodies, dAbs, scFb, Fab, etc. The initiating antigen-binding construct can be BiTe, (scFv)2, nanobodies, nanobodies-HSA, DART, TandAb, single-chain biantibodies (scDiabody), single-chain biantibodies-CH3, scFv-CH-CL-scFv, HSAbody, single-chain biantibodies-HAS, or tandem scFv. The antigen-binding portion of the masked antigen-binding construct can be VHH-scAb, VHH-Fab, dual scFab, F(ab')2, diabody, crossMab, DAF (two-in-one), DAF (four-in-one), DutaMab, DT-IgG, knob-in-hole universal light chain, knob-in-hole assembly, charge pair, Fab arm exchanger, SEEDbody, LUZ-Y, Fcab, κλ-body, orthogonal Fab, DVD-IgG, IgG(H)-scFv, scFv-(H)IgG, IgG(L)-scFv, scFv-(L)IgG, IgG(L,H)-Fv, IgG(H)-V, V(H)-IgG, IgG(L)-V, V(L)-IgG, KIH IgG-scFab, 2scFv-IgG, IgG-2scFv, scFv4-Ig, Zybody, DVI-IgG, Bivalent Antibody-CH3, Triple Body, Miniantibody, Minibody, TriBi Minibody, scFv-CH3 KIH, Fab-scFv, F(ab')2-scFv2, scFv-KIH, Fab-scFv-Fc, Quadrivalent HCAb, Single-Chain Bivalent Antibody-Fc, Bivalent Antibody-Fc, Tandem scFv-Fc, Intrabody, Dock and Lock, ImmTAC, IgG-IgG Conjugate, Cov-X-Body or scFv1-PEG-scFv2.

[0218] The initiating antigen-binding construct can be a monoclonal antibody. The initiating antigen-binding construct can have immune specificity against cancer cell antigens (preferably cell surface antigens). The antigen-binding portion of the masking antigen-binding construct can bind therapeutic antigens. Such therapeutic antigens include antigens that can serve as targets for the treatment of any disease or condition, including but not limited to cancer, autoimmune diseases, and infections.

[0219] Antigen-binding constructs can target specific targets, including receptors and their ligands or corresponding receptors on the surface of cancer cells (i.e., tumor-associated antigens). These targets include, but are not limited to, CD3, CD19, CD20, CD22, CD30, CD33, CD34, CD40, CD44, CD47, CD52, CD70, CD79a, CD123, Her-2, EphA2, lymphocyte-associated antigen 1, VEGF or VEGFR, CTLA-4, LIV-1, nectin-4, CD74, SLTRK-6, EGFR, CD73, PD-L1, CD163, CCR4, CD147, EpCam, Trop-2, CD25, C5aR, Ly6D, αv integrin, B7H3, B7H4, Her-3, folate receptor α, GD-2, CEACAM5, CEACAM6, c-MET, CD266, MUC1, CD10, MSLN, sialyl Tn, Lewis... Y, CD63, CD81, CD98, CD166, tissue factor (CD142), CD55, CD59, CD46, CD164, TGFβ receptor 1 (TGFpR1), TGFpR2, TGFpR3, FasL, MerTk, Axl, Clecl2A, CD352, FAP, CXCR3, and CD5.

[0220] Examples of commercial antibodies and their targets applicable to the masking antigen-binding constructs described herein include, but are not limited to, anti-mouse CD3 antibody TAm, anti-human HER2 antibody trastuzumab, anti-human EGFR antibody cetuximab (C225) and panitumumab, Brentuximab or Brentuximab vedotin (CD30), alemtuzumab (CD52), rituximab (CD20), trastuzumab Her / neu, nimotuzumab, cetuximab (EGFR), bevacizumab (VEGF), palivizumab (RSV), abciximab (GpIIb / IIIa), and infliximab. b) Adalimumab, certolizumab, golimumab (TNF-α), basiliximab, daclizumab (IL-2R), omalizumab (IgE), gemtuzumab or vadastuximab (CD33), natalizumab (VLA-4), vedolizumab Anti-vedolizumab (α4β7), belimumab (BAFF), otelixizumab, teplizumab (CD3), ofatumumab, ocrelizumab (CD20), epratuzumab (CD22), alemtuzumab (CD52), eculizumab (C5), canakimumab (IL-1β). Mepolizumab (IL-5), reslizumab, tocilizumab (IL-6R), ustekinumab, briakinumab (IL-12, 23), hBU12 (CD19, US20120294853), humanized 1F6 or 2F12 (CD70, US20120294863), BR2-14a and BR2-22a (LIV-1, WO2012078688).

[0221] The masking antigen-binding construct disclosed herein can be used to treat autoimmune diseases. The antigen-binding portion of the masking antigen-binding construct for treating autoimmune diseases may bind to non-restricted antigens including TNF-α, IL-1, IL-2R, IL-6, IL-12, IL-23, IL-17, IL-17R, BLyS, CD20, CD52, α4β7 integrin, and α4 integrin.

[0222] Knobs-into-holes (KIH) technology can be used for this construct, which involves engineering the CH3 region to create “bumps” or “grooves” on each heavy chain to facilitate heterodimerization. KIH technology is described, for example, in the paper “Production of bispecific antibodies in 'knobs-into-holes' using a cell-free expression system” by Xu, Yiren et al., MAbs, Vol. 7, No. 1, Taylor & Francis, 2015, which is incorporated herein by reference in its entirety. In this invention, one heavy chain may have T366W and / or S354C (bump) substitutions (EU number), and the other heavy chain may have Y349C, T366S, L368A, and / or Y407V (groove) substitutions (EU number). One heavy chain may have one or more of the following substitutions: Y349C and T366W (EU number). The other heavy chain may have one or more of the following substitutions: E356C, T366S, L368A, and Y407V (EU number). Furthermore, substitutions (-ppcpscp-->-ppcppcp-) may be introduced into the hinge region of the two substituted IgGs.

[0223] For example, the initiating antigen-binding construct can be a bispecific antibody containing a cancer antigen-binding domain (e.g., CEA) and an anti-CD3 antigen-binding domain. The cancer antigen-binding domain can be in the form of a single-domain antibody (VHH), and the CD3 antigen-binding domain can be in the form of a Fab. The antigen-binding domain can be located at the N-terminus of the Fc region, and the Fc hinge region can serve as a linker connecting the two fragments. A double-stranded asymmetric structure containing the Fc region can be used. The Fc region can confer a longer half-life and good stability to the masked antigen-binding construct. Simultaneously, the design of a double-stranded Knob into Hole (KIH) structure can reduce the probability of mismatch and improve the uniformity and yield of the target antigen-binding construct.

[0224] In some embodiments, the knock-Fc of the initiating antigen binding construct is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 1, and the hole-Fc is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to sequence identifier 2 (SEQ ID NO: 2).

[0225] In some embodiments, the initiating antigen-binding construct may have a TAm heavy chain variable region (VH) sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 3. The initiating antigen-binding construct may have a CEA nanobody (VHH) sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 4. The initiator antigen-binding construct may have an IgG4 CH1 domain sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to that of SEQ ID NO: 5. The initiator antigen-binding construct may have a TAm heavy chain (knob) sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to that of SEQ ID NO: 6. The initiating antigen binding construct may have a CEA(VHH) heavy chain (hole) sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 7.

[0226] In some embodiments, the initiating antigen-binding construct may have a TAm light chain variable region (VL) sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to that of SEQ ID NO: 8. The initiating antigen-binding construct may have a light chain constant region (CL) sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to that of SEQ ID NO: 9. The initiating antigen-binding construct may have a TAm light chain sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to that of SEQ ID NO: 10.

[0227] In some embodiments, the initiator antigen-binding construct may have a heavy chain sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 25.

[0228] In some embodiments, the initiator antigen-binding construct may have a heavy chain sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to any of the heavy chain sequences shown in Table 9. The initiator antigen-binding construct may have a heavy chain sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 25, 29, or 32. The initiator antigen-binding construct may have a light chain sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to any of the light chain sequences shown in Table 9. The initiator antigen-binding construct may have a light chain sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 26, 30, or 33.

[0229] The initial antigen-binding construct can be prepared using a variety of techniques known in the art, including hybridoma, recombinant, and phage display technologies, or combinations thereof. For example, monoclonal antibodies can be produced using hybridoma technologies known in the art. The term "monoclonal antibody" refers to an antibody derived from a single clone, including any eukaryotic, prokaryotic, or phage clone.

[0230] Antibody fragments that recognize specific epitopes can be generated using known techniques. For example, the Fab and F(ab′)2 fragments described in this specification can be prepared by protease cleavage of immunoglobulin molecules, using enzymes such as papain (for generating the Fab fragment) or pepsin (for generating the F(ab′)2 fragment). The F(ab′)2 fragment contains a variable region, a light chain constant region, and a CH1 region of the heavy chain.

[0231] Fragmentable sites and connectors Cleavable linkers can be used to couple α-helical masking structures to antigen-binding constructs (e.g., antibodies or their antigen-binding fragments). A cleavable linker may contain a cleavable site and optionally two linker fragments (e.g., located before and after the cleavable site). The cleavable linker can be any amino acid fragment conventionally used for linking peptide domains. Suitable cleavable linkers are of variable length, such as 1-20, 2-15, 3-12, 4-10, 6-20, 6-10 amino acids, or specifically 5, 6, 7, 8, 9, or 10 amino acids. In some embodiments, the cleavable linker comprises at least 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, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, or 50 amino acids. In some embodiments, the cleavable linker comprises no more than 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, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, or 50 amino acids. In some embodiments, the cleavable linker comprises about 14 to 30 amino acids. In some embodiments, the cleavable linker comprises about 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, or 30 amino acids. In some embodiments, the cleavable site comprises about 6 to 10 amino acids. In some embodiments, the cleavable site in the cleavable linker contains about 6, 7, 8, 9, or 10 amino acids.

[0232] Cleavable sites can be recognized and cleaved by extracellularly expressed proteases, enabling them to access masked antigen-binding constructs (e.g., antigen-binding domains), releasing the masked antigen-binding constructs and allowing them to access their targets (e.g., receptor extracellular domains or soluble ligands). Multiple matrix metalloproteinase (MMP) recognition sites (MMP1-28) may be applicable. MMPs play a role in tissue remodeling and are associated with tumorigenesis processes such as morphogenesis, angiogenesis, and metastasis. Examples of MMPs can be found in U.S. Patent Publication 2013 / 0309230, International Patent Publications WO 2009 / 025846, WO 2010 / 081173, WO 2014 / 107599, WO2015 / 048329, U.S. Patent Publication 20160160263, and Ratnikov et al. (Proc. Natl. Acad. Sci. USA, 111: E4148-E4155 (2014)); all of which are incorporated herein by reference in their entirety.

[0233] Cleavable sites can be cleaved within the tumor microenvironment. These cleavable sites can be matrix metalloproteinase (MMP) cleavage sites. MMP cleavage sites can be selected from MMP2, MMP7, MMP9, and MMP13 cleavage sites. After MMP cleavage, the heavy and / or light chains of the masking antigen-binding construct can contain amino acid residue fragments from the MMP cleavage site. These residue fragments can contain the sequence LRSG, SG, or VR at the N-terminus of the antigen-binding portion of the masking antigen-binding construct.

[0234] Cleavage sites can also be cleaved at bibasic sites (e.g., arginine-arginine, lysine-arginine, or lysine-lysine sites). Enzymes known in the art to cleave at bibasic sites include, for example, N-arginine bibasic convertases (Chow et al., JBC 275: 19545-19551 (2000)) and subtilisin-like proprotein convertases, such as furin (PCI), PC2, and PC3 (Steiner (1991, in Peptide Biosynthesis and Processing (Fricker ed.), pp. 1-16, CRC Press, Boca Raton, Florida; Muller et al., JBC 275:39213-39222, (2000)).

[0235] Furthermore, one or more linkers can be used to couple α-helical masking structures to antigen-binding constructs. Glycine and glycine-serine polymers are relatively unstructured and therefore can be used as neutral linkers between components. Glycine has significantly more phi-psi space than alanine and is much less restricted than residues with longer side chains (see Scheraga, Rev. Computational Chem. 11173-142 (1992)). Some exemplary linkers are in the form S(G)nS, where n is 5-20. Other exemplary linkers include (G)n, glycine-serine polymers (including, for example, (GS)n, (GSGGS)n (SEQ ID NO: 36)), glycine-alanine polymers, alanine-serine polymers, and other flexible linkers.

[0236] Provide initiator antigen binding construct The method disclosed herein can use an initiating antigen-binding construct (e.g., an antibody or its antigen-binding fragment) as the starting material (“initiating antigen-binding construct”). An α-helical masking structure can be designed targeting the antigen-binding domain (e.g., the initiating antigen-binding domain) in the initiating antigen-binding construct.

[0237] The initiating antigen-binding construct can be non-human, humanized, human, chimeric, or surface-remodeled antibody. The initiating antigen-binding construct can be nanobodies, dAbs, scFb, Fab, etc. The initiating antigen-binding construct can be BiTe, (scFv)2, nanobodies, nanobodies-HSA, DART, TandAb, single-chain biantibodies (scDiabody), single-chain biantibodies-CH3, scFv-CH-CL-scFv, HSAbody, single-chain biantibodies-HAS, or tandem scFv. The antigen-binding portion of the masked antigen-binding construct can be VHH-scAb, VHH-Fab, dual scFab, F(ab')2, diabody, crossMab, DAF (two-in-one), DAF (four-in-one), DutaMab, DT-IgG, knob-in-hole universal light chain, knob-in-hole assembly, charge pair, Fab arm exchanger, SEEDbody, LUZ-Y, Fcab, κλ-body, orthogonal Fab, DVD-IgG, IgG(H)-scFv, scFv-(H)IgG, IgG(L)-scFv, scFv-(L)IgG, IgG(L,H)-Fv, IgG(H)-V, V(H)-IgG, IgG(L)-V, V(L)-IgG, KIH IgG-scFab, 2scFv-IgG, IgG-2scFv, scFv4-Ig, Zybody, DVI-IgG, Bivalent Antibody-CH3, Triple Body, Miniantibody, Minibody, TriBi Minibody, scFv-CH3 KIH, Fab-scFv, F(ab')2-scFv2, scFv-KIH, Fab-scFv-Fc, Quadrivalent HCAb, Single-Chain Bivalent Antibody-Fc, Bivalent Antibody-Fc, Tandem scFv-Fc, Intrabody, Dock and Lock, ImmTAC, IgG-IgG Conjugate, Cov-X-Body or scFv1-PEG-scFv2.

[0238] The initiating antigen-binding construct can be a monoclonal antibody. The initiating antigen-binding construct can have immune specificity against cancer cell antigens (preferably cell surface antigens). The antigen-binding portion of the masking antigen-binding construct can bind therapeutic antigens. Such therapeutic antigens include antigens that can serve as targets for the treatment of any disease or condition, including but not limited to cancer, autoimmune diseases, and infections.

[0239] Antigen-binding constructs can target specific targets, including receptors and their ligands or corresponding receptors on the surface of cancer cells (i.e., tumor-associated antigens). These targets include, but are not limited to, CD3, CD19, CD20, CD22, CD30, CD33, CD34, CD40, CD44, CD47, CD52, CD70, CD79a, CD123, Her-2, EphA2, lymphocyte-associated antigen 1, VEGF or VEGFR, CTLA-4, LIV-1, nectin-4, CD74, SLTRK-6, EGFR, CD73, PD-L1, CD163, CCR4, CD147, EpCam, Trop-2, CD25, C5aR, Ly6D, αv integrin, B7H3, B7H4, Her-3, folate receptor α, GD-2, CEACAM5, CEACAM6, c-MET, CD266, MUC1, CD10, MSLN, sialyl Tn, Lewis... Y, CD63, CD81, CD98, CD166, tissue factor (CD142), CD55, CD59, CD46, CD164, TGFβ receptor 1 (TGFpR1), TGFpR2, TGFpR3, FasL, MerTk, Axl, Clecl2A, CD352, FAP, CXCR3, and CD5.

[0240] Examples of commercial antibodies and their targets applicable to the masking antigen-binding constructs described herein include, but are not limited to, anti-mouse CD3 antibody TAm, anti-human HER2 antibody trastuzumab, anti-human EGFR antibody cetuximab (C225) and panitumumab, Brentuximab or Brentuximab vedotin (CD30), alemtuzumab (CD52), rituximab (CD20), trastuzumab Her / neu, nimotuzumab, cetuximab (EGFR), bevacizumab (VEGF), palivizumab (RSV), abciximab (GpIIb / IIIa), and infliximab. b) Adalimumab, certolizumab, golimumab (TNF-α), basiliximab, daclizumab (IL-2R), omalizumab (IgE), gemtuzumab or vadastuximab (CD33), natalizumab (VLA-4), vedolizumab Anti-vedolizumab (α4β7), belimumab (BAFF), otelixizumab, teplizumab (CD3), ofatumumab, ocrelizumab (CD20), epratuzumab (CD22), alemtuzumab (CD52), eculizumab (C5), canakimumab (IL-1β). Mepolizumab (IL-5), reslizumab, tocilizumab (IL-6R), ustekinumab, briakinumab (IL-12, 23), hBU12 (CD19, US20120294853), humanized 1F6 or 2F12 (CD70, US20120294863), BR2-14a and BR2-22a (LIV-1, WO2012078688).

[0241] The masking antigen-binding construct disclosed herein can be used to treat autoimmune diseases. The antigen-binding portion of the masking antigen-binding construct for treating autoimmune diseases may bind to non-restricted antigens including TNF-α, IL-1, IL-2R, IL-6, IL-12, IL-23, IL-17, IL-17R, BLyS, CD20, CD52, α4β7 integrin, and α4 integrin.

[0242] Knobs-into-holes (KIH) technology can be used for this construct, which involves engineering the CH3 region to create “bumps” or “grooves” on each heavy chain to facilitate heterodimerization. KIH technology is described, for example, in the paper “Production of bispecific antibodies in 'knobs-into-holes' using a cell-free expression system” by Xu, Yiren et al., MAbs, Vol. 7, No. 1, Taylor & Francis, 2015, which is incorporated herein by reference in its entirety. In this invention, one heavy chain may have T366W and / or S354C (bump) substitutions (EU number), and the other heavy chain may have Y349C, T366S, L368A, and / or Y407V (groove) substitutions (EU number). One heavy chain may have one or more of the following substitutions: Y349C and T366W (EU number). The other heavy chain may have one or more of the following substitutions: E356C, T366S, L368A, and Y407V (EU number). Furthermore, substitutions (-ppcpscp-->-ppcppcp-) may be introduced into the hinge region of the two substituted IgGs.

[0243] For example, the initiating antigen-binding construct can be a bispecific antibody containing a cancer antigen-binding domain (e.g., CEA) and an anti-CD3 antigen-binding domain. The cancer antigen-binding domain can be in the form of a single-domain antibody (VHH), and the CD3 antigen-binding domain can be in the form of a Fab. The antigen-binding domain can be located at the N-terminus of the Fc region, and the Fc hinge region can serve as a linker connecting the two fragments. A double-stranded asymmetric structure containing the Fc region can be used. The Fc region can confer a longer half-life and good stability to the masked antigen-binding construct. Simultaneously, the design of a double-stranded Knob into Hole (KIH) structure can reduce the probability of mismatch and improve the uniformity and yield of the target antigen-binding construct.

[0244] In some embodiments, the knock-Fc of the initiating antigen binding construct is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 1, and the hole-Fc is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to sequence identifier 2 (SEQ ID NO: 2).

[0245] In some embodiments, the initiating antigen-binding construct may have a TAm heavy chain variable region (VH) sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 3. The initiating antigen-binding construct may have a CEA nanobody (VHH) sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 4. The initiator antigen-binding construct may have an IgG4 CH1 domain sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to that of SEQ ID NO: 5. The initiator antigen-binding construct may have a TAm heavy chain (knob) sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to that of SEQ ID NO: 6. The initiating antigen binding construct may have a CEA(VHH) heavy chain (hole) sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 7.

[0246] In some embodiments, the initiating antigen-binding construct may have a TAm light chain variable region (VL) sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to that of SEQ ID NO: 8. The initiating antigen-binding construct may have a light chain constant region (CL) sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to that of SEQ ID NO: 9. The initiating antigen-binding construct may have a TAm light chain sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to that of SEQ ID NO: 10.

[0247] In some embodiments, the initiator antigen-binding construct may have a heavy chain sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 25.

[0248] In some embodiments, the initiator antigen-binding construct may have a heavy chain sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to any of the heavy chain sequences shown in Table 9. The initiator antigen-binding construct may have a heavy chain sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 25, 29, or 32. The initiator antigen-binding construct may have a light chain sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to any of the light chain sequences shown in Table 9. The initiator antigen-binding construct may have a light chain sequence that is at least 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to SEQ ID NO: 26, 30, or 33.

[0249] The initial antigen-binding construct can be prepared using a variety of techniques known in the art, including hybridoma, recombinant, and phage display technologies, or combinations thereof. For example, monoclonal antibodies can be produced using hybridoma technologies known in the art. The term "monoclonal antibody" refers to an antibody derived from a single clone, including any eukaryotic, prokaryotic, or phage clone.

[0250] Antibody fragments that recognize specific epitopes can be generated using known techniques. For example, the Fab and F(ab′)2 fragments described in this specification can be prepared by protease cleavage of immunoglobulin molecules, using enzymes such as papain (for generating the Fab fragment) or pepsin (for generating the F(ab′)2 fragment). The F(ab′)2 fragment contains a variable region, a light chain constant region, and a CH1 region of the heavy chain.

[0251] Production of masked antigen-binding constructs Masked antigen-binding constructs can be prepared using a variety of techniques known in the art. For example, they can be expressed in host cells—expression vectors encoding both the heavy and light chains are transfected into host cells using standard techniques. Various forms of the term "transfection" are intended to encompass a variety of commonly used techniques for introducing exogenous DNA into prokaryotic or eukaryotic host cells, such as electroporation, calcium phosphate precipitation, DEAE-dextran transfection, etc. Although the masked antigen-binding constructs described herein (e.g., antibodies or their antigen-binding fragments) can be expressed in prokaryotic or eukaryotic host cells, expression in eukaryotic cells is preferred, and most preferably in mammalian host cells, because such eukaryotic cells (especially mammalian cells) are more likely than prokaryotic cells to assemble and secrete properly folded and immunologically active antigen-binding constructs.

[0252] The variable regions of heavy and light chain DNA sequences can be cloned into mammalian expression vectors and fused with pre-inserted human heavy chain constant regions or κ light chain constant regions within the frame. The target molecule can be produced by co-transfecting Expi 293F™ cells with the mammalian expression vector using polyethyleneimine. At transfection, the ratio of heavy chain vector to light chain vector can be 1:1.5.

[0253] Preferred mammalian host cells for expressing the masked antigen-binding constructs (e.g., antibodies) described herein include Chinese hamster ovary cells (CHO cells), NSO myeloma cells, COS cells, SP2 cells, and HEK293 cells. The masked antigen-binding construct is produced by culturing the host cells for a period sufficient for intracellular expression (more preferably secretion into the culture medium for host cell growth) after the recombinant expression vector encoding the antigen-binding construct has been introduced into the mammalian host cells. The antigen-binding construct can be recovered from the culture medium using standard protein purification methods.

[0254] Host cells can be used to produce functional antibody fragments, such as Fab fragments or scFv molecules. In some cases, it is advantageous to transfect host cells with DNA encoding functional fragments of the light and / or heavy chains of a masking antigen-binding construct. Recombinant DNA technology can also be used to remove some or all of the DNA encoding the light and heavy chains (either or both), which are not essential for binding the antigen of interest. Molecules expressed from such truncated DNA molecules also fall under the category of masking antigen-binding constructs described herein. Furthermore, bifunctional antibodies can be produced, in which one pair of heavy / light chains is derived from the masking antigen-binding constructs described herein, and another pair of heavy / light chains is specific to an antigen other than the antigen of interest. Such bifunctional antibodies can be prepared by crosslinking the masking antigen-binding construct described herein with a second antigen-binding construct using standard chemical crosslinking methods.

[0255] In certain systems for recombinant expression of masked antigen-binding constructs (e.g., antibodies or their antigen-binding moieties), a recombinant expression vector encoding the heavy and light chains of the masked antigen-binding construct is introduced into CHO cells via calcium phosphate-mediated transfection. Within this recombinant expression vector, both the heavy and light chain genes of the masked antigen-binding construct are functionally linked to a CMV promoter or an AdMLP promoter regulatory element to drive high-level transcription of the gene. The vector may also carry a selection marker for screening CHO cells transfected with the vector. The recombinant expression vector can be prepared using standard molecular biology techniques, followed by transfection of host cells, screening of transformants, culturing of host cells, and recovery of the masked antigen-binding construct from the culture medium. Further, this disclosure provides a method for synthesizing the recombinant antigen-binding construct described herein, comprising culturing the host cells described herein in a suitable culture medium until the recombinant antigen-binding construct described herein is synthesized. This method may further include isolating the recombinant antigen-binding construct from the culture medium.

[0256] The masking antigen-binding construct used in the formulation can be recovered and purified from recombinant cell cultures by methods including, but not limited to, protein A purification, ammonium sulfate or ethanol precipitation, acid extraction, anion or cation exchange chromatography, cellulose phosphate chromatography, hydrophobic interaction chromatography, affinity chromatography, hydroxyapatite chromatography, and lectin chromatography. High-performance liquid chromatography (HPLC) can also be used for purification. See, for example, Colligan, *Current Protocols in Immunology* or *Current Protocols in Protein Science*, John Wiley & Sons, New York, NY (1997–2001).

[0257] The masked antigen-binding constructs described in this invention can be expressed in modified forms. For example, an additional amino acid, especially a charged amino acid, can be added to the amino terminus of the masked antigen-binding construct to improve its stability and persistence in host cells, during purification, or during subsequent processing and storage. Furthermore, peptide groups can be added to the masked antigen-binding construct to facilitate purification. These regions can be removed prior to the final preparation of the masked antigen-binding construct. Such methods have been described in numerous standard laboratory manuals, such as Sambrook, supra; Ausubel, et al., ed., Current Protocols In Molecular Biology, John Wiley & Sons, Inc., NY, NY (1987–2001). The masked antigen-binding constructs described in this invention may include purified products, products of chemical synthesis processes, and products generated from eukaryotic hosts (e.g., yeast, higher plants, insects, and mammalian cells) via recombinant technology. Depending on the host used in the recombinant production process, the antigen-binding constructs or masked antigen-binding constructs disclosed herein may be glycosylated or non-glycosylated, with glycosylated forms preferred. Such methods have been described in numerous standard laboratory manuals, such as Sambrook, supra; Ausubel, supra, Colligan, Protein Science, supra. The purity of the masked antigen-binding construct can be measured by size exclusion chromatography (SEC). The purity of the masked antigen-binding construct can be greater than 30%, 40%, 50%, 60%, 70%, 72.5%, 75%, 77.5%, 80%, 82.5%, 85%, 87.5%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, or 99%, for example, as measured by SEC. The purity of the masked antigen-binding construct can be less than 30%, 40%, 50%, 60%, 70%, 72.5%, 75%, 77.5%, 80%, 82.5%, 85%, 87.5%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, or 99%, for example, as measured by SEC. The purity of the masked antigen-binding construct can be 95%–100%, or 95%–100% as measured by SEC.

[0258] The yield of the masked antigen-binding construct can be measured by SEC. The yield of the masked antigen-binding construct can be greater than 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, or 80 mg / L. The yield of the masked antigen-binding construct can be less than 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, or 80 mg / L. The yield of the masked antigen-binding construct can be 40-60 or 50-60 mg / L.

[0259] Methods of using masked antigen-binding constructs The masked antigen-binding constructs described herein can be used for therapeutic purposes. This disclosure provides methods for preventing, managing, treating, or improving a disease, including administering one or more of the masked antigen-binding constructs described herein to a subject in need, either alone or in combination with one or more therapies (e.g., one or more prophylactic or therapeutic agents). Non-limiting exemplary diseases and conditions that can be treated with the masked antigen-binding constructs provided herein include cancer, autoimmune diseases, and infections. Generally, any disease or condition that can be treated with therapeutic antibodies can be treated with the masked antigen-binding constructs provided herein. Compared to unmasked forms, masked antigen-binding constructs reduce side effects, for example, because the masked antigen-binding construct does not bind to its antigen until the masking domain is removed. By selecting a suitable cleavable linker between the α-helical masking structure and the antigen-binding portion of the masked antigen-binding construct, the masked antigen-binding construct remains masked until it reaches the vicinity of its target antigen, particularly the target antigen at the site of the disease or condition. For example, in some cases, by selecting cleavable sites that can be cleaved by high concentrations of proteases near the tumor, masked antigen-binding constructs can have higher safety or a wider therapeutic window because they do not significantly bind antigens before reaching the tumor.

[0260] A method for treating a subject's cancer is provided, comprising: a) determining that the level of MMPs in the subject's cancer is elevated relative to surrounding non-cancerous tissue; and b) administering to the subject a therapeutically effective amount of the masking antigen-binding construct provided herein, wherein each masking domain of the masking antigen-binding construct contains a cleavable site, and the cleavable site is a matrix metalloproteinase (MMP) cleavage site. The MMP cleavage site may be selected from MMP2, MMP7, MMP9, and MMP13 cleavage sites. The MMPs may be selected from MMP2, MMP7, MMP9, and MMP13.

[0261] Step a) may include: i) isolating cancerous and non-cancer tissues from the subject; ii) detecting MMPs in the isolated cancerous and non-cancer tissues; and iii) comparing the staining levels in cancerous and non-cancer tissues.

[0262] This disclosure also provides compositions comprising one or more masking antigen-binding constructs described herein, and methods of using said compositions to prevent, manage, treat, or improve a disease or one or more symptoms thereof. Other therapeutic or preventative agents include, but are not limited to, small molecules, synthetic pharmaceuticals, peptides, polypeptides, proteins, nucleic acids (e.g., DNA and RNA nucleotides, including but not limited to antisense nucleotide sequences, triplets, RNAi, and nucleotide sequences encoding biologically active proteins, polypeptides, or peptides), antibodies, synthetic or natural inorganic molecules, mimics, and synthetic or natural organic molecules.

[0263] Any therapy known to be, or currently used, for the prevention, management, treatment, or improvement of a disease or one or more of its symptoms may be used in combination with the masked antigen binding construct described herein.

[0264] The antigen-binding constructs described herein can be used directly against specific antigens. The antigen-binding portion of these constructs may belong to a subclass or isotype capable of mediating cell lysis. Furthermore, the antigen-binding portion of these constructs may belong to a subclass or isotype that, upon complexing with cell surface proteins, activates serum complement and / or mediates antibody-dependent cytotoxicity (ADCC) by activating effector cells (such as natural killer cells or macrophages).

[0265] As used herein, the term "effective amount" refers to the amount of a compound (e.g., a masking antigen-binding construct) sufficient to produce a beneficial or desired outcome. Effective amounts can be administered by single or multiple applications, doses, or administration, and are not limited to a particular formulation or route of administration. Typically, therapeutically effective amounts of the active ingredient range from 0.01 mg / kg to 100 mg / kg, 0.1 mg / kg to 100 mg / kg, 1 mg / kg to 100 mg / kg, 0.01 mg / kg to 10 mg / kg, 0.1 mg / kg to 10 mg / kg, and 1 mg / kg to 10 mg / kg. The administered dose may vary depending on known factors such as the pharmacokinetic characteristics of the particular agent and its route of administration; the recipient's age, health status, and weight; the type and extent of the disease or indication to be treated, the nature and severity of symptoms, the type of concurrent treatments, the frequency of treatment, and the desired effect. The initial dose may be increased above the upper limit to rapidly achieve the desired blood or tissue levels. Alternatively, the initial dose may be less than the optimal dose, and the daily dose may be gradually increased during treatment. Human dosage can be optimized, for example in a routine Phase I dose-escalation study designed from 0.5 mg / kg to 20 mg / kg. Dosage frequency can vary depending on factors such as route of administration, dose, serum half-life of the masked antigen-binding construct, and the disease being treated. Exemplary dosing frequencies are once daily, once weekly, and once every two weeks.

[0266] This disclosure provides a method for treating cancer in cells, tissues, organs, animals, or patients. The disclosure also provides a method for treating solid cancers in humans. Examples of cancer include, but are not limited to, solid tumors, soft tissue tumors, hematologic malignancies that give rise to solid tumors, and metastatic lesions. Examples of hematologic malignancies with the potential to give rise to solid tumors include, but are not limited to, diffuse large B-cell lymphoma (DLBCL), follicular lymphoma, myelodysplastic syndrome (MDS), lymphoma, Hodgkin's disease, malignant lymphoma, non-Hodgkin's lymphoma, Burkitt lymphoma, multiple myeloma, and Likert syndrome (Likert conversion). Examples of solid tumors include, but are not limited to, malignant tumors of various organ systems, such as sarcomas (including soft tissue sarcomas and osteosarcomas), adenocarcinomas and carcinomas, such as those affecting the head and neck (including the pharynx), thyroid gland, lungs (small cell or non-small cell lung cancer (NSCLC)), breast, lymphatic tissue, gastrointestinal tract (such as the mouth, esophagus, stomach, liver, pancreas, small intestine, colon and rectum, anal canal), reproductive and genitourinary systems (such as kidneys, urothelial cells, bladder, ovaries, uterus, cervix, endometrium, prostate, testes), central nervous system (such as nerve or glial cells, such as neuroblastoma or glioma), skin (such as melanoma), etc. The solid tumor may be an NMDA receptor-positive teratoma. Cancers can include breast cancer, colon cancer, pancreatic cancer (such as pancreatic neuroendocrine tumor (PNET) or pancreatic ductal adenocarcinoma (PDAC)), stomach cancer, uterine cancer, and ovarian cancer. Cancer can also be selected from, but is not limited to, the following leukemias: acute lymphoblastic leukemia (ALL), chronic lymphocytic leukemia (CLL), acute myeloid leukemia (AML), chronic myeloid leukemia (CML), hairy cell leukemia (HCL), T-cell prolymphocytic leukemia (T-PLL), large granular lymphocytic leukemia, adult T-cell leukemia, and acute monocytic leukemia (AMoL).

[0267] The methods described in this specification include administering the masking antigen-binding construct described herein to a subject, as described herein, for example, in combination with one or more other therapies, such as surgery or administration of other therapeutic agents. In the case of cancer, other therapies may include chemotherapy, such as cytotoxic drugs. Other therapies may also include targeted therapies, such as tyrosine kinase inhibitors, proteasome inhibitors, or protease inhibitors. Other therapies may also include anti-inflammatory, anti-angiogenic, anti-fibrotic, or anti-proliferative compounds, such as steroids, biological immunomodulators (such as immune checkpoint molecule inhibitors), monoclonal antibodies, antibody fragments, aptamers, siRNA, antisense molecules, fusion proteins, cytokines, cytokine receptors, bronchodilators, statins, anti-inflammatory agents (such as methotrexate), or nonsteroidal anti-inflammatory drugs (NSAIDs). Other therapies may also include the combined use of different classes of drugs. The masking antigen-binding construct formulation may be administered simultaneously or sequentially with other therapies.

[0268] Proteases in tissues can be monitored using a variety of techniques, including those that monitor protease activity and those that detect proteolytic activity. Traditional methods for detecting the presence of proteases in tissues (which may include both inactive and active forms) include IHC, RNA sequencing, Western blot, or ELISA-based methods. Other techniques can also be used to detect protease activity in tissues, including zymography, in situ fluorescence microscopy zymography, or the use of fluorescent protein hydrolysis substrates. Furthermore, the use of fluorescent protein hydrolysis substrates can be combined with the immunocapture of specific proteases. Antibodies targeting protease active sites are also available using various techniques, including IHC, fluorescence microscopy, Western blot, ELISA, or flow cytometry (see Sela-Passwell et al., *Nature Medicine*, 18: 143-147, 2012; LeBeau et al., *Cancer Research*, 75: 1225-1235, 2015; Sun et al., *Biochemistry*, 42:892-900, 2003; Shiryaev et al., 2:e80, 2013).

[0269] Nucleic acids containing nucleic acid sequences encoding masking antigen-binding constructs can be developed into RNA or DNA drugs for therapeutic purposes. DNA or RNA drugs can be delivered via carriers such as lipid nanoparticles (“LNPs”). After the nucleic acid is translated into a masking antigen-binding construct in a subject, the masking molecule can be activated in a similar manner as described above. Administration of the masking molecule as a DNA or RNA drug can improve its safety or therapeutic window.

[0270] Masked antigen-binding constructs can be used in cell therapies, such as CAR-T therapy. Target cells (e.g., T cells) can be transfected using vectors such as retroviruses and lentiviruses to express the masked antigen-binding construct, which is part of a chimeric antigen receptor (CAR). The CAR expressed on the cell surface may possess a masked antigen-binding domain, which can be activated as described above. Compared to traditional CAR-T therapy, cell therapies incorporating masked antigen-binding constructs may offer higher safety or a wider therapeutic window.

[0271] Similar to ADCs, masked antigen-binding constructs can be part of fusion molecules containing peptides, proteins, cytokines, enzymes, etc.

[0272] Structural analysis of antigen-binding constructs There are two general approaches to predicting the structure of proteins of interest: template-based modeling, which uses known related protein structures to model the unknown target structure; and template-free modeling, which does not rely on global similarity to structures in the Protein Data Bank (PDB) and can therefore be used for proteins with novel folds. Both methods can be used to predict the interaction between α-helical masked structures and antibodies. In some embodiments, template-free modeling methods (e.g., alphafold2 or esmfold) are used.

[0273] Standard template-based modeling involves selecting a suitable structural template; aligning the target sequence with the template structure; and performing molecular modeling to account for mutations, insertions, and deletions in the target-template alignment. Single-sequence search methods (such as BLAST) can be used to scan PDB sequences to detect closely related templates. To detect more distant templates, a database of sequence feature matrices of known structural proteins can be scanned using a target sequence profile (target sequence feature matrix) constructed from multiple sequence alignments, or through profile-profile comparisons, or by matching it to a structural template library to assess sequence-structure compatibility. Template selection methods return an initial target-template alignment, which can be manually adjusted, typically iteratively after model construction. Given an alignment with a template, existing tools can be used to quickly construct a molecular model of the target sequence, optimizing the side-chain only at mutation sites and reconstructing the backbone at insertions and deletions. For target protein sequences that are only distantly related to known structural proteins, more complex methods may be required, relying on multiple templates and aggressive backbone conformation sampling. By combining available crystal structures, template-based modeling methods can provide structural information for approximately two-thirds of known protein families.

[0274] Template-free modeling methods can be used for proteins that do not share global structural similarity with proteins in a protein data bank (PDB). Due to the lack of structural templates, these methods require a conformational sampling strategy to generate candidate models and a ranking criterion to select native-like conformations. Template-free structure prediction typically begins with constructing multiple-sequence alignments (MSAs) of the target protein and its related sequences. Local structural features, such as secondary structures and backbone torsion angles, as well as non-local features, such as residue-residue contacts or inter-residue distances on the polypeptide chain, are then predicted using the target sequence and its homologous sequences. These predicted features guide the construction of a 3D model of the target protein structure, which is subsequently refined, ranked, and compared to select the final prediction. End-to-end deep neural networks directly use embedded multiple sequence alignments (MSAs) and pairwise features to predict the 3D coordinates of all heavy atoms in a given protein. Leveraging insights from learning evolutionary patterns in millions of sequences using language models, single-sequence protein structure prediction methods infer structures directly from primary sequences using large language models, achieving high-resolution structure prediction.

[0275] Structural analysis can be performed on the antigen-binding portion of a masked antigen-binding construct (e.g., an antibody) or on the masked antigen-binding construct itself to identify key framework residues in the starting frame region; these key residues may need to be retained if they do not correspond to their counterparts in the receptor antibody. These key residues can be identified using methods well-known in the art, such as modeling the interaction between the CDR and framework residues to determine framework residues important for antigen binding, and identifying uncommon framework residues at specific sites through sequence comparison. (See, for example, Queen et al., US Pat. No. 5,585,089; Riechmann et al., Nature 332:323 (1988), the entire contents of which are incorporated herein by reference.) Three-dimensional immunoglobulin models are generally available and well-known to those skilled in the art. Computer programs are available capable of displaying and exhibiting the possible three-dimensional conformations of selected candidate immunoglobulin sequences. Examining these three-dimensional conformations allows for analysis of the functional roles of residues in the candidate immunoglobulin sequences, i.e., analyzing residues that influence the candidate immunoglobulin's ability to bind its antigen. In this way, FR residues can be selected and combined from consensus sequences and imported sequences to achieve desired antibody properties, such as increased affinity for target antigens.

[0276] The antigen-binding portion of a masked antigen-binding construct (e.g., an antibody) or the structure of the masked antigen-binding construct can be predicted using deep learning-based folding methods such as Alphafold2, RosettaFold, or VibrantFold. Antibody structures can be modeled using the Antibody Modeler in the Molecular Operating Environment (MOE), 2011.10; Chemical Computing Group Inc., Montreal, QC, Canada). MOE antibody homology modeling considers the specific structural composition of the antibody when searching for template candidates and constructing templates. Therefore, models can be generated based on templates containing frames and CDR loops from different sources (combined in dimer form). A knowledge-based approach can be employed, utilizing the antibody structure database currently stored in the Protein Data Bank (PDB), which is clustered by category, species, subclass, and frame sequence similarity. This database can be enriched by adding more antibody structures and can be continuously updated and re-clustered.

[0277] Multiple structural models can be provided for each initiating antigen-binding construct or each masking antigen-binding construct to generate a single consensus structure. This consensus structure is then used for further structure-based analysis. Any structural models containing missing or gap information in the modeled structure can be discarded.

[0278] Once a suitable structural model is determined, a person with ordinary skill in the art can annotate the modeled structure to identify the CDR or FR by associating the structure with the annotation sequence of the initiating antigen binding construct provided above. For example, if a missing or inserted element appears in the modeled structure, the structural model can be moved or recalibrated to correspond to the structural position of the original initiating antigen binding construct.

[0279] Numerous programs are available for modeling and evaluating the three-dimensional structure of masked antigen-binding constructs (e.g., antibodies) (see, for example, U.S. Patent No. 7,117,096). For instance, molecular mechanics software can be used to achieve these purposes, examples of which include, but are not limited to, Congen, SCWRL, UHBD, GENPOL, and AMBER.

[0280] CONGEN (CONformation GENerator) is a program for conformational searches of protein fragments (RE Bruccoleri (1993) Molecular Simulations 10, 151–174 (1993); REBruccoleri, E. Haber, J. Novotny, (1988) Nature 335, 564–568 (1988); R.Bruccoleri, M. Karplus. (1987) Biopolymers 26, 137–168). It is best suited for problems requiring the construction of undetermined loops or fragments within known structures, i.e., homology modeling. The basic energy functions used include bond, angle, torsion, anomalous angle, van der Waals, and electrostatic interaction terms, using a distance-varying dielectric constant and employing an Amber94 force field, which can be determined via CONGEN. This program can be used to perform conformational searches and structural evaluations using basic or refined scoring functions. The program can also calculate other properties of molecules, such as solvent-accessible surface area and conformational entropy, provided that stereo constraints are given.

[0281] SCWRL is a side-chain placement program that generates side-chain rotomers and combinations thereof using a library of main-chain-dependent rotomers (Dunbrack RL Jr, Karplus M (1993) J Mol Biol 230:543–574; Bower, MJ, Cohen FE, Dunbrack RL (1997) J Mol Biol 267, 1268–1282). The library provides a list of chi1-chi2-chi3-chi4 values ​​and their relative probabilities for residues at a given phi-psi value. The program can further explore these conformations to minimize side-chain-main-chain collisions and side-chain-side-chain collisions. After minimizing steric conflict, the energy of the substituted side chains and main chain can be minimized using CONGEN (Bruccoleri and Karplus (1987) Biopolymers 26:137–168) to alleviate local strain.

[0282] Several automated programs specifically designed for constructing antibody structures can be used to model the structure of masked antigen-binding constructs (e.g., antibodies) or the antigen-binding portion of masked antigen-binding constructs. The ABGEN program is an automated antibody structure generation algorithm for obtaining structural models of antibody fragments (Mandal et al. (1996) Nature Biotech.14:323–328). ABGEN employs a homology-based scaffold technique and includes structural motifs using invariant and strictly conserved residues, known Fab structures, typical features of hypervariable loops, torsional constraints of residue substitutions, and key inter-residue interactions. Specifically, the ABGEN algorithm consists of two main modules: ABalign and ABbuild. ABalign is a program that provides alignment scores between the antibody sequence and all known antibody V-region sequences. The library sequence with the highest score is considered the best match for the test sequence. ABbuild then uses the best-matching model output by ABalign to generate a three-dimensional structure and provides Cartesian coordinates for the desired antibody sequence.

[0283] WAM (Whitelegg NRJ and Rees, AR (2000) Protein Engineering 13, 819–824) is an improved version of ABM, employing a combinatorial algorithm (Martin, ACR, Cheetham, JC, and Rees AR (1989) PNAS 86, 9268–9272) to model CDR conformations using typical conformations of CDR loops from the PDB database and loop conformations generated using CongEN. In short, the modular nature of antibody structures allows for structural modeling through a combination of protein homology modeling and structure prediction.

[0284] The method described in this paper can be used to model antibody structures. Since antibodies are among the most conserved proteins in both sequence and structure, antibody homology models are relatively simple, except for some specific CDR loops that are not yet identified in existing standard structures or that have insertions or deletions. In some cases, these loops can be modeled using algorithms that combine homology modeling with conformational search, or using the deep learning algorithms described in this paper.

[0285] Systems, software and interfaces The methods described herein (e.g., computational screening of α-spiral masking structures) typically require a computer, processor, software, module, or other device. The methods described herein are generally computer-implemented methods, and one or more portions of the methods are sometimes executed by one or more processors. Embodiments of the methods described herein are generally applicable to the same or related processes implemented by instructions in the systems, devices, and computer program products described herein. The processes and methods described herein can be performed by automated methods. Automated methods can be embodied in software, modules, processors, peripheral devices, and / or devices containing similar components. As used herein, "software" refers to computer-readable program instructions that, when executed by a processor, perform the computer operations described herein.

[0286] Data or datasets can be characterized by one or more features or variables, including, for example, the values ​​of amino acid sequences. A sequencing device can be part of a system. A system can include a computing device and a sequencing device, wherein the sequencing device is configured to receive physical nucleic acids and generate sequence reads, and the computing device is configured to process the reads from the sequencing device.

[0287] The subject matter and functional operation of this disclosure can be implemented using digital electronic circuits, computer software or firmware existing in tangible form, computer hardware (including the structures described herein and their equivalents), or combinations of one or more of the above structures. The implementation of the subject matter described herein can be as one or more computer programs, i.e., one or more computer program instruction modules, encoded on a tangible program carrier for execution by a processing device or control of its operation. Alternatively, as a supplement, program instructions can also be encoded on propagation signals, which are artificially generated, such as machine-generated electrical signals, optical signals, or electromagnetic signals, for encoding information and transmitting it to a suitable receiving device for execution by a processing device. The machine-readable medium can be a machine-readable storage device, a machine-readable storage substrate, a random or serial access memory device, or one or more combinations thereof.

[0288] Referring to Figure 3, system 10 processes input data using a processor and outputs information (e.g., computer screening of an α-spiral masking structure). System 10 includes a client device 12, a data processing system 18, a data storage system 20, a network 16, and a wireless device 14. The processor processes the input data according to the methods described herein.

[0289] Data processing system 18 retrieves data 21 representing one or more processor parameter values ​​from data repository 20, including, for example, threshold levels and / or antigen-binding construct sequences. Data processing system 18 inputs the retrieved data to a processor, such as to data processing program 30. In this embodiment, data processing program 30 is programmed to perform computer screening of an α-helix masking structure.

[0290] The data processing system 18 can generate data for a graphical user interface, which, when rendered on the display device of the client device 12, displays a visual representation of the output. The values ​​of these parameters can be stored in the data repository 20 or in memory 22.

[0291] Client device 12 can be any computing device capable of accepting user input and communicating with data processing system 18 and / or other client devices via network 16. Client device 12 can be a mobile device, desktop computer, laptop computer, mobile phone, personal digital assistant (PDA), server, embedded computing system, etc.

[0292] Data processing system 18 can be one of various computing devices capable of receiving data and running one or more services. Data processing system 18 may include servers, distributed computing systems, desktop computers, laptops, mobile phones, etc. Data processing system 18 can be a single server or a group of servers located in the same location or different locations (i.e., different sites). Data processing system 18 and client device 12 can run programs with client-server relationships. Although different modules are shown in the figure, client and server programs may run on the same device.

[0293] Data processing system 18 can receive data from wireless device 14 and / or client device 12 via input / output (I / O) interface 24 and data repository 20. Data repository 20 can store various data values ​​for data processing program 30. The processing program (also referred to as a program, software, software application, script, or code) can be written in any form of programming language, including compiled or interpreted languages, or declarative or procedural languages, and can be deployed in any form, including as a standalone program or as a module, component, subroutine, or other unit suitable for a computing environment. The data processing program may or may not correspond to a file in a file system. The program may be stored as part of a file containing other programs or information (e.g., one or more scripts stored in a markup language document), or it may be stored in a single file dedicated to the program, or it may be stored in multiple co-located files (e.g., a file storing one or more modules, subroutines, or code snippets). The data processing program can be deployed to execute on a single computer, or on multiple computers located in the same location or distributed across multiple locations and interconnected via a communication network.

[0294] Interface 24 can be an interface type capable of receiving data over a network, such as an Ethernet interface, a wireless network interface, a fiber optic network interface, a modem, etc. The data processing system 18 also includes a processing device 28. In this document, "processing device" encompasses all means, devices, and machines used for processing information, such as a programmable processor, a computer, or multiple processors or computers. This device may include dedicated logic circuitry, such as an FPGA (field programmable gate array), an ASIC (application specific integrated circuit), or a RISC (reduce instruction set circuit). The device may also contain, outside of hardware, code for creating an execution environment for related computer programs, such as code constituting processor firmware, a protocol stack, a database management system, an operating system, or one or a combination thereof.

[0295] The data processing system 18 also includes a memory 22 and a bus system 26, such as a data bus and a motherboard, for establishing and controlling data communication between the components of the data processing system 18. The processing device 28 may include one or more microprocessors. Typically, the processing device 28 may include suitable processors and / or logic capable of receiving and storing data and communicating over a network. The memory 22 may include hard disks and random access storage devices, such as dynamic random access memory (DRAM) or other types of non-volatile machine-readable storage devices. The memory 22 stores data processing programs 30 that can be executed by the processing device 28. These computer programs may include a data engine for implementing the operations and / or techniques described herein. The data engine may be implemented as software, hardware, or a combination of software and hardware running on a computing device.

[0296] Various methods and formulas can be implemented as computer program instructions and executed by a processing device. Suitable programming languages ​​for expressing these instructions include, but are not limited to, Python, C, C++, FORTRAN implementations such as FORTRAN77 or FORTRAN90, Java, Visual Basic, Perl, Tcl / Tk, JavaScript, ADA, and statistical analysis software such as SAS, R, MATLAB, SPSS, and Stata. Various aspects of the methods can be written in different computational languages ​​and communicate with each other through appropriate system-level tools available on a given system.

[0297] The processes and logic flows described in this disclosure can be executed by one or more programmable computers, which perform one or more computer programs to realize functions by manipulating input information and generating output. These processes and logic flows can also be executed by dedicated logic circuits, and the devices can also be implemented as dedicated logic circuits, such as FPGAs (field-programmable gate arrays), ASICs (application-specific integrated circuits), or RISCs.

[0298] Suitable for computers that execute computer programs, such as general-purpose or special-purpose microprocessors, or both, or any other type of central processing unit (CPU).

[0299] Computer-readable media suitable for storing computer program instructions and information include various forms of non-volatile memory, media, and storage devices, such as semiconductor memory devices like EPROM, EEPROM, and flash memory devices; magnetic disks, such as internal hard disks or removable disks; magneto-optical disks; and CD-ROMs and (Blue Ray) DVD-ROMs. Processors and memory can be complemented by or integrated into dedicated logic circuits.

[0300] To provide interaction with the user, the subject matter disclosed herein can be implemented on a computer with a display device, such as a CRT (cathode ray tube) or LCD (liquid crystal display) monitor, to display information to the user, and equipped with a keyboard and pointing devices, such as a mouse or trackball, through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user. Furthermore, the computer can also interact with the user by sending and receiving documents to and from the device used by the user; for example, by sending a webpage to a web browser on the user's client device in response to a request received from the web browser.

[0301] The subject matter described herein can be implemented in a computing system containing backend components, such as an information server; or containing middleware components, such as an application server; or containing frontend components, such as a client computer with a graphical user interface or a web browser through which a user can interact with the subject matter; or a combination of one or more of the aforementioned backend, middleware, or frontend components. The components of the system can be interconnected via digital communication in any form or medium, such as a communication network. Examples of communication networks include local area networks (“LANs”) and wide area networks (“WANs”), such as the Internet.

[0302] Computing systems can include clients and servers. Clients and servers are typically remote to each other and usually interact via communication networks. The client-server relationship arises from the programs running on their respective computers and forming a client-server relationship. Servers may reside in the cloud via cloud computing services.

[0303] While this disclosure contains numerous specific implementation details, these should not be construed as limiting any claimable scope, but rather as descriptions of features that may be specific to certain implementations. Certain features described in this disclosure in the context of independent implementations may also be implemented in combination within a single implementation. Conversely, various features described in the context of a single implementation may also be implemented separately in multiple implementations, or in any suitable sub-combination manner. Furthermore, although the aforementioned features may be described as functioning in a particular combination, or even initially claimed in this way, in certain circumstances, one or more features in a claimed combination may be excluded from that combination, and the claimed combination may point to a sub-combination or a variation of a sub-combination.

[0304] Similarly, while operations are described in a specific order, this should not be interpreted as meaning that these operations must be performed in the specific order or sequence shown, or that all operations in the examples must be performed to obtain the desired result. In some cases, multitasking and parallel processing may be advantageous. Furthermore, the separation of various system components in the above implementation should not be interpreted as requiring such separation in all implementations, and should be understood as meaning that the described program components and systems can generally be integrated into a single software product or packaged into multiple software products.

[0305] Specific implementations of this subject matter have been described. Other implementations are also within the scope of the claims. For example, the operations listed in the claims can be performed in different orders and still achieve the desired results. The processes shown in the figures do not necessarily have to be performed in the specific order or sequence shown in the figures to obtain the desired results. In some implementations, multitasking and parallel processing may be advantageous.

[0306] Example The present invention is further described in the following embodiments, but these embodiments do not limit the scope of the invention as set forth in the claims.

[0307] Example 1: Computational screening of α-helical regions blocking antibody CDR regions This embodiment describes the process of computer-screening the α-helices of the CDR regions of the following antibodies: anti-mouse CD3 antibody TAm, anti-human HER2 antibody trastuzumab, anti-human EGFR antibody cetuximab (C225), panitumumab, anti-CD19 antibody blinatumomab, anti-HER2 single-domain antibody ErBB, and anti-CD3 single-domain antibody EP03. For these antibodies, the α-helix is ​​positioned at the N-terminus of either the heavy or light chain, and computer screening is performed using indicators including α-helix pLDDT, α-helix ptm, ipTM, and distance. For a detailed description of the screening process in the embodiment, please refer to the "Designing α-helix Masking Structures" section. Examples 3 and 5 show the specific structures. The following table lists the screening results: Table 1: Evaluation Indicators for α-helix Masked Antigen Binding Constructs The calculation methods for pLDDT and ipTM in the above table can be found in the paper "Highly accurate protein structure prediction with AlphaFold" by Jumper, John et al., published in *Nature*, issue 596.7873 (2021), which is cited in full as a reference. "α-helix pLDDT" refers to the pLDDT value calculated only for the predicted structure after alpha-helix masking; this value should be greater than 55. "α-helix ptm" is used to assess the accuracy of alpha-helix structure prediction; this value should be greater than 0.5. "ipTM" is used to assess the docking correctness between the alpha-helix masked sequence and the antibody; this value should be greater than 0.45. "Distance" measures the spatial distance between the alpha-helix masked sequence and the antibody CDR region; this value should be less than 15 Å. The calculation methods are as follows: Where, dij represents the Euclidean distance from the i-th amino acid of the α-helix to the j-th amino acid of the antibody CDR region; m represents the number of amino acids in the α-helix, and n represents the number of amino acids in the CDR region.

[0308] In some cases, the α-helix is ​​connected to the N-terminus of the light chain via a cuttable connector. Structural simulation results for CEA+TAm-R1-L16, CEA+TAm-R1-L17, and N_L_Pani_mask70 are shown below. Figures 1A-1C As shown.

[0309] In some cases, the α-helix is ​​connected to the N-end of the heavy chain via a cuttable connector. The structural simulation results for CEA+TAm-R1-H1, CEA+TAm-R1-H4, CEA+TAm-R1-H5, N_H_Cet-mask72, N_H_Pani_mask1, Tra-R2-2, and Tra-R2-4 are as follows: Figures 1D-1J As shown.

[0310] Example 2: Preparation, screening, and bispecific antibody construction of CD3 and CEA antibodies Based on the anti-mouse CD3 antibody, the anti-mouse CD3 antibody TAm was obtained through mutation, library construction, humanization, and multiple screenings. The CEA antibody was obtained by immunizing camels. Adult camels were subcutaneously injected with 250 μg of recombinant tumor antigen human CEA every two months for a total of five immunizations. Peripheral blood mononuclear cells (PBMCs) were isolated, and RNA was extracted to construct a phage antibody library. After the library was packaged to form phage particles, it was panned using liquid chromatography. The phages were bound to a biotin-labeled CEA antigen solution and then separated using streptavidin magnetic beads. Through various screening methods (such as ELISA and engineered cells expressing high levels of CEA), the CEA antibody C17 was selected, structurally named CEA. Bispecific antibody structure design: This bispecific antibody contains a CEA antigen-binding domain and a mouse CD3 antigen-binding domain. The CEA antigen-binding domain is in the form of a single-domain antibody (VHH), and the CD3 antigen-binding domain is in the form of a Fab antibody. The TAm and CEA antibodies are located at the N-terminus of the Fc region, with the Fc hinge region serving as a linker between the two fragments. A double-stranded asymmetric structure containing the Fc region is employed. The Fc region imparts a longer half-life and better stability to the antibody. Simultaneously, the double-stranded Knob into Hole (KIH) structure significantly reduces the mismatch probability, improving the uniformity and yield of the target bispecific antibody. The heavy chain structures of the bispecific antibody from the N-terminus to the C-terminus are: (1) TAm VH-CH1-knob-Fc, (2) CEA(VHH)-hole-Fc. The light chain structure is TAm VL-CL (kappa). The two heavy chain IgG4 Fcs are knob-Fc and hole-Fc, respectively. The sequences of knob-Fc and hole-Fc are shown in the table below. The light chain is the kappa light chain.

[0311] Table 2: List of Fc sequences The table below shows the heavy chain variable region (VH) sequence and CH1 sequence of (1) the antibody targeting mouse CD3 (TAm) and (2) the single-domain antibody targeting CEA C17.

[0312] Table 3: List of heavy chain sequences of TAm and CEA antibodies The table below shows the light chain variable region (VL) sequence and light chain constant region (CL) sequence of the antibody against mouse CD3 (TAm).

[0313] Table 4: List of TAm antibody light chain sequences Example 3: Preparation of CEA-TAm bispecific antibody using humanized C17 as anti-CEA antibody and TAm as anti-mouse CD3 antibody. The α-helix is ​​attached to the N-terminus of the TAm heavy or light chain as an antibody masking structure.

[0314] The structure of the masked antigen-binding construct is as follows (N-terminus-C-terminus): α-helix + linker peptide 1 + cleavable site + linker peptide 2 - TAm heavy chain (heavy chain containing "knob" structure) or TAm light chain (light chain).

[0315] The light and heavy chain sequences of the α-helix-TAm antibody are shown in the table below. The letter "L" in the name indicates that the α-helix is ​​attached to the light chain; the letter "H" in the name indicates that the α-helix is ​​attached to the heavy chain ("knob" structure).

[0316] Table 5: Light / Heavy Chain Sequences of α-Helical-TAm Antibody Table 6: Light / Heavy Chain Sequences of α-Helical-TAm Antibody In the table above, the cleavable linker sequence GSWPHGSLQAARGS (SEQ ID NO: 21) contains the cleavable site sequence WPHGSLQAAR (SEQ ID NO: 14), while the cleavable linker sequence GLSGRSDNHGS (SEQ ID NO: 23) contains the cleavable site sequence LSGRSDNH (SEQ ID NO: 24).

[0317] The heavy and light chain variable region DNA sequences were cloned into a mammalian expression vector, which was located in the same reading frame as the pre-inserted human IgG4 heavy chain or κ light chain constant region. Masked antigen-binding constructs were prepared by co-transfecting Expi 293F™ cells (ThermoFisher) with the mammalian expression vector and transfecting with polyethyleneimine (Polyethylenimine “Max” (PEI); Polysciences, catalog number: 24765-2). Cells were transfected with the corresponding expression vectors at a 1:4:2 ratio (“heavy chain (knob) vector”: “heavy chain Fc (hole) vector”: “light chain vector”). The heavy and light chain compositions of the bispecific antibodies are shown in the table below. CEA+TAm is a bispecific antibody without an α-helical structure. The monoclonal antibody TAm mAb (IgG4) was used as a control of a bivalent anti-mouse CD3 antibody without an α-helical structure.

[0318] Table 7: Heavy and light chain composition of CEA-TAm bispecific antibody Expi 293F™ cells were cultured in suspension in CD OptiCHO™ medium at 37°C, 5% CO2, and 135 rpm. One day prior to transfection, 293F cells were seeded into 1L aerosol Ehrenmeier flasks (Corning) at a density of 1.0 × 10⁶ cells / year. 6 Cells / ml, culture volume 200 mL. On the day of transfection, the expected cell density is 1.8-2.0 × 10⁶ cells / ml. 6Cells / ml. On the day of transfection, the cell suspension was centrifuged at 1000 rpm for 5 minutes at room temperature to collect cells, washed once with Expi293 medium, and then resuspended in 200 mL Expi293 medium. 400 μg of plasmid was diluted with 5 mL Opti-MEM medium and vortexed for 15 seconds; 1.2 mg of PEI was diluted with 5 mL Opti-MEM medium and vortexed for 15 seconds. The PEI-containing solution was added dropwise to the DNA-containing solution, gently mixed, and incubated at room temperature for 15 minutes. Subsequently, the plasmid / PEI mixture was added to the cell suspension and incubated in a 37°C, 5% CO2, 85 rpm incubator. After 4 hours, 200 mL of EX-CELL™ 293 medium and 2 mM glutamine (Gibco) were added, and the rotation speed was adjusted to 135 rpm for continued culture. 24 hours later, a cell proliferation inhibitor (3.8 mM VPA) was added, followed by 40 mL of culture medium D after 72 hours. After 7 days of culture, the supernatant was collected by centrifugation at 18,000 rpm for 30 minutes and purified. The solution was aseptically filtered through a 0.22 μm filter membrane, and sodium azide was added to achieve a final concentration of 0.01% (w / v). The solution was stored at 4°C.

[0319] Purify the target protein using protein A. Load cell culture supernatant onto a Mabselect Prism A FF (GE; 17-5498-01) column equilibrated with 20 mL of 25 mM Tris and 150 mM NaCl (pH 7.5). Elute unbound protein with at least 10 column volumes of 25 mM Tris and 150 mM NaCl (pH 7.5). Elute the target protein with 5 column volumes of 20 mM sodium citrate (pH 3.5). Neutralize the protein solution by adding 1 / 10 volume of 1 M Tris (pH 9.5).

[0320] The target protein was exchanged for the desired buffer using a Zeba™ desalting centrifuge column (ThermoFisher) or an ultrafiltration tube (Millipore). Protein concentration and purity were determined by SDS-PAGE electrophoresis and NanoDrop2000, using 2–3 μg of sample. The target protein was aliquoted and stored at -80°C.

[0321] After concentration and filtration of the target protein, it was added to a gel filtration column (Gel Filtration, HiLoad Superdex 200, GE) equilibrated with 20 mM histidine and 140 mM sodium chloride (pH 6.0). The sample was eluted with pH 7.2 PBS buffer at room temperature at a flow rate of 0.5 mL / min to analyze molecular weight, purity, aggregation, and other properties.

[0322] In another purification method, the protein is purified from the cell supernatant by protein A affinity chromatography (MabSelect SuRe, GE), followed by cation exchange chromatography (HiTrap SP HP, GE) of the protein eluent, and then fractionation and analysis via gel filtration (SEC). The protein obtained using this purification method has a target antibody purity >90%.

[0323] Protein yield and purity are shown in the table.

[0324] Table 8: Antibody Yield and Purity Example 4: Binding to mouse CD3 antigen (measured by ELISA) Enzyme-linked immunosorbent assay (ELISA) was used to assess the binding of antibodies to mouse CD3ε antigen. The specific experimental steps are as follows: Antigen coating: Dilute mouse CD3ε-Fc antigen (ACRO systems, CDE-M5256) to 1 μg / ml with PBS, add 100 μl to each well of a microtiter plate, and incubate overnight at 4°C.

[0325] Blocking: Wash the plate three times with 0.05% PBST, add 200 μl of 2% skim milk powder-PBS blocking solution to each well, and incubate at room temperature for 1 hour.

[0326] Primary antibody detection: Wash the plate three times with 0.05% PBST. Dilute the sample antibody to 20 μg / ml, and perform serial dilutions (dilution factor 3) to obtain gradient concentrations of 6.67 μg / ml, 2.22 μg / ml, 0.74 μg / ml, 0.25 μg / ml, 0.08 μg / ml, 0.03 μg / ml, 0.01 μg / ml, 0.003 μg / ml, 0.001 μg / ml, 0.0003 μg / ml, and 0.0001 μg / ml. Add 100 μl of the prepared serially diluted antibody sample to each well of a microtiter plate and incubate at room temperature for 1 hour.

[0327] Secondary antibody detection: Wash the plate three times with 0.05% PBST. Dilute goat anti-human κ light chain (bound and free)-HRP (Sigma-Aldrich, A7164) at a ratio of 1:5000, add 100 μl to each well of the microtiter plate, and incubate at room temperature for 1 hour.

[0328] Color development: Wash the plate 6 times with 0.05% PBST. Add 100 μl of TMB color development solution to each well. After color development for 10 minutes, add 100 μl of 1M hydrochloric acid to each well to stop the reaction.

[0329] Reading: Absorbance was measured at 450 nm using an ELISA reader.

[0330] The standard curve was fitted using GraphPad software, and the EC50 value of antibody-antigen binding was calculated.

[0331] Experimental results showed that, compared with the monoclonal antibody TAm (IgG4), the bispecific antibody CEA+TAm exhibited a similar binding curve to mouse CD3 antigen, with an EC50 of approximately 14.41 μg / ml. The binding of the α-helix-masked bispecific antibody CEA+TAm to mouse CD3 antigen was significantly reduced (compared to CEA+TAm), with an affinity reduction of 9-27 fold calculated from OD values. The affinity reduction of CEA+TAm-R1-H4 exceeded 500 fold.

[0332] Experimental results are as follows Figure 2A , 2B And as shown in the table below.

[0333] Table 9: EC50 (μg / ml) of antibody binding to mouse CD3 antigen and fold decrease in antibody affinity Example 5: Preparation of anti-human HER2 (ERBB2) / EGFR (ERBB1) antibody Using trastuzumab (anti-human HER2 antibody), cetuximab (anti-human EGFR antibody), and panitumumab as controls, masking monoclonal antibodies were prepared according to Example 1. An α-helix was constructed as the antibody masking structure, targeting the N-terminus of either the heavy or light chain. The antibody sequences are shown in the table below.

[0334] Table 10. List of anti-human HER2 (ERBB2) / EGFR (ERBB1) antibody sequences The variable regions of the heavy and light chain DNA sequences were cloned into a mammalian expression vector, maintaining reading frames consistent with the pre-inserted human IgG1 heavy chain constant region or κ light chain constant region.

[0335] The molecule was generated by co-transfection of Expi 293F™ cells (ThermoFisher) with a mammalian expression vector using polyethyleneimine (Polyethylenimine "Max" (PEI); Polysciences, catalog number: 24765-2). Cells were transfected with the corresponding expression vector at a ratio of 1:1.5 ("heavy chain vector" : "light chain vector").

[0336] Protein expression and purification were performed according to the method described in Example 3. The protein yield and purity of the monoclonal antibody are shown in the table below.

[0337] Table 11: Yield and purity of anti-human HER2 / EGFR antibodies Example 6: Determination of antibody affinity for human HER2 and human EGFR by biomembrane interferometry (BLI) Biomembrane Interference (BLI) experiments were performed using an OCTECT RED96e (ForteBio) at 30°C. The running buffer was 0.02% PBST solution (10 mmol / L Na2HPO4; 1.75 mmol / L KH2PO4; 137 mmol / L NaCl; 2.65 mmol / L KCl; pH 7.2-7.4, containing 0.02% surfactant Tween 20).

[0338] First, the sample antibody was captured by anti-human Fc protein immobilized on the surface of the AHC2 sensor chip (Anti-hIgG Fc, Sartorius). Specifically, the sample antibody was diluted to 5 μg / ml, and then a 1.5 nm protein was coupled to the surface of the AHC sensor chip using 0.02% PBST.

[0339] The analytes human HER2 (ACRO Systems, HE2-H5225) and recombinant tumor antigen human EGFR (CEACAM-5 / CD66e, ACRO Systems, EGR-H5222) were diluted to 200 nM with 0.02% PBST.

[0340] After the sample antibody was captured by the chip, it was bound to and dissociated with different concentrations of analyte to obtain the KD value of the interaction. The experimental parameters were as follows: baseline 1: 60 s, loading: 240 s; baseline 2: 180 s, binding: 240 s, dissociation: 600 s; loading response: 1.5–2.0 nm; high-sensitivity kinetics: 2 Hz. The equilibrium dissociation constant (KD) was obtained by fitting the 1:1 Langmuir binding rate equation through numerical integration. The binding-dissociation curves were simultaneously fitted using Date Analysis HT 12 software (sartorius, 50-5029) to calculate Kon and Koff. Affinity KD = Koff / Kon.

[0341] The results showed that the anti-HER2 antibody had an affinity for human HER2 ranging from 3.10 nM to 7.91 nM. Compared with trastuzumab, the binding capacity of the masked antigen-binding constructs Tra-R2-2 and Tra-R2-4, which contain α-helix structures, was reduced by 57% and 55%, respectively, indicating that the α-helix at the N-terminus of the antibody heavy chain significantly hindered binding to the HER2 antigen, and the mass of the bound antigen was reduced by more than 50%. Compared with trastuzumab, the affinity of Tra-R2-4 was reduced by 1.5-fold, while the affinity of Tra-R2-2 remained largely unchanged.

[0342] The anti-EGFR antibodies (cetuximab and N_H_Cet-mask72) showed a relatively small affinity for human EGFR antigen, ranging from 3.18 nM to 3.95 nM. Compared to cetuximab, the α-helix-containing masked antigen-binding construct N_H_Cet-mask72 exhibited a 30% decrease in binding affinity to human EGFR antigen. This indicates that the α-helix at the N-terminus of the antibody heavy chain significantly hinders binding to EGFR antigen, and reduces the mass of bound antigen by 30%.

[0343] The anti-EGFR antibodies (panitumumab, N_L_Pani_mask70, and N_H_Pani_mask1) showed relatively small affinities for human EGFR antigen, ranging from 0.998 nM to 1.10 nM. Compared to panitumumab, the binding affinity of N_L_Pani_mask70 and N_H_Pani_mask1, which contain α-helical structures, decreased by 31% and 28%, respectively. This indicates that the α-helix at the N-terminus of the antibody light chain or the N-terminus of the antibody heavy chain significantly hinders binding to the EGFR antigen, and reduces the mass of bound antigen by approximately 30%.

[0344] Example 7: Preparation of anti-CD19 antibody based on bonnetumab Using the anti-CD19 antibody bonnetumab as a control, masking monoclonal antibodies were prepared according to Example 3. The α-helix was constructed as the N-terminal heavy or light chain of the antibody masking structure. The antibody sequences are shown in the table below.

[0345] Table 12: List of anti-CD19 antibody sequences For each antibody, its sequence was inserted into the pCDNA3.1 vector, and the plasmid was then transfected into HEK 293F cells via PEI (Polysciences, catalog number 24765-2) to achieve transient expression of the target protein. Cell supernatant was collected after 5 days of culture. The target protein was purified using Ni Sepharose excel or Protein A (Cytiva), and its purity was analyzed by SDS-PAGE gel electrophoresis. The protein yield and purity of the monoclonal antibodies are shown in the table below.

[0346] Table 13: Yield and purity of anti-CD19 antibody Example 8: Binding of anti-CD19 antibody to Raji cells (measured by flow cytometry) The binding of anti-CD19 antibody to Raji cells was evaluated using flow cytometry. The specific experimental procedure was as follows: Raji cells were collected, resuspended in PBS, and placed in 96-well plates. 100 μl of serially diluted antibody was added to each well, and the cells were gently mixed by pipetting and incubated at 4°C for 2 hours. After washing twice with PBS, 100 μl of PE-labeled anti-human IgG Fc antibody (rabbit monoclonal antibody against the 6X tag (FITC), abcam, catalog number: ab1206) was added to each well, and the cells were incubated at 4°C for 0.5 hours. After washing twice more with PBS, data were read using a Beckman Cytoflex instrument. The results showed that the binding ability of the designed masking agent—bonatomab antibody—to Raji cells was significantly reduced. Specific results are shown in Figure 16 and the table below. Table 14: EC50 (μg / ml) of anti-CD19 antibody binding to Raji cells Example 9: Preparation of ErBB-based anti-HER2 single-domain antibody A single-domain antibody, ERBB (ErBB-Fc), was constructed based on IgG4-Fc. An α-helix masking structure was added to the N-terminus of the single-domain antibody, namely α-helix + linker peptide 1 + substrate + linker peptide 2 - ErBB heavy chain - Fc. In this embodiment, the α-helix masking structure may have two types. The first type has a single helix structure, and the second type consists of two helical structures, as shown in Figures 17A-17B. The antibody sequence is shown in the table below.

[0347] Table 15: List of anti-HER2 single-domain antibody sequences Protein expression and purification were performed according to the method described in Example 3. The protein yield and purity of the monoclonal antibody are shown in the table below.

[0348] Table 16: Yield and purity of anti-HER2 single-domain antibodies Example 10: Binding of anti-HER2 single-domain antibody to H1573 cells (detected by flow cytometry) The binding of the anti-HER2 single-domain antibody to H1573 cells was evaluated by flow cytometry. The specific experimental steps are as follows: H1573 cells were collected, resuspended in PBS, and seeded into 96-well plates. 50 μl of serially diluted antibody was added to each well. Cells were gently mixed by pipetting and incubated at 4°C for 2 hours. After washing twice with PBS, 100 μl of diluted PE-labeled anti-human IgG Fc antibody (Invitrogen #12-4998-82, batch number: 2533138) was added to each well, and the plates were incubated at 4°C for 0.5 hours. After washing twice with PBS, the plates were analyzed using a Beckman Cytoflex instrument. The results showed that the designed masking structure-ErBB single-domain antibody significantly reduced the binding affinity to H1573 cells (up to 10-fold). Specific results are shown in Figures 18A-18B.

[0349] Example 11: Binding of anti-HER2 single-domain antibody after enzymatic digestion to H1573 cells (measured by flow cytometry) 50 μg of antibody was digested using 250 nmol of enzyme (R&D, catalog number: 911-MP) in a 250 μL reaction volume. Digestion was performed at 37°C for 15 hours. H1573 cells were collected, resuspended in PBS, and seeded into 96-well plates. Serially diluted enzyme-digested antibody was added to each well. Cells were gently mixed by pipetting and incubated at 4°C for 2 hours. After washing twice with PBS, 100 μL of PE-labeled goat anti-human IgG Fc antibody (Invitrogen #12-4998-82, batch number: 2533138) was added to each well, and the cells were incubated at 4°C for 0.5 hours. After washing twice more with PBS, the cells were analyzed using a Beckman Cytoflex flow cytometer. The results showed that the flow cytometry curve of the antibody after digestion was basically consistent with that of the initial control antibody ErBB-Fc, indicating that the masking structure had been removed after digestion. Specific results are shown in Figure 19 and the table below. Table 17: EC50 (μg / ml) of anti-HER2 single-domain antibody binding to H1573 cells after enzymatic digestion In addition, the anti-HER2 single-domain antibody was analyzed by SDS-PAGE before and after enzymatic digestion, and the results are shown in Figure 20. The SDS-PAGE results show that the molecular weight after digestion is smaller than that before digestion, indicating that the masking has been removed after digestion.

[0350] Example 12: Preparation of anti-CD3 single-domain antibody based on EP03 The single-domain antibody EP03 is constructed based on IgG4-Fc (i.e., ErBB-Fc). An α-helix masking structure is added to the N-terminus of the single-domain antibody, specifically the structure α-helix + linker peptide 1 + substrate + linker peptide 2 - EP03 heavy chain - Fc. In this embodiment, this α-helix masking structure may have two types. The first type has a single helix structure, and the second type has two helical structures. The antibody sequence is shown in the table below.

[0351] Table 18: List of anti-CD3 single-domain antibody sequences Protein expression and purification were performed according to the method described in Example 3. The protein yield and purity of the monoclonal antibodies are shown in the table. Compared to the initial EP03-Fc, EP03-H1 and EP03-H2, with a single α-helix, and EP03-dH1 and EP03-dH2, with two α-helices, all showed significant improvements.

[0352] Table 19: Yield and purity of anti-CD3 single-domain antibodies Example 13: Binding of anti-CD3 single-domain antibody to Jurkat cells (measured by flow cytometry) The binding of the anti-CD3 single-domain antibody to Jurkat cells was evaluated by flow cytometry according to the method described in Example 10.

[0353] The results showed that the designed masking-EP03 single-domain antibody significantly reduced its binding ability to Jurkat cells. Specific results are shown in Figure 21 and the table below.

[0354] Table 20: EC50 (μg / ml) of anti-CD3 antibody binding to Jurkat cells Example 14: Binding of anti-CD3 single-domain antibody after enzymatic digestion to Jurkat cells (measured by flow cytometry) The binding of the enzymatically digested anti-CD3 single-domain antibody to Jurkat cells was evaluated by flow cytometry according to the method described in Example 11.

[0355] Experimental results showed that after enzymatic digestion, the flow cytometry curves of the antibody and the masking antibody were basically consistent with those of the initial control antibody EP03-Fc, indicating that the masking agent had been removed after enzyme digestion. Specific results are shown in Figure 22 and the table below: Table 21: EC50 (μg / ml) of anti-CD3 single-domain antibody binding to Jurkat cells after enzymatic digestion. Furthermore, the anti-CD3 single-domain antibody was analyzed by SDS-PAGE both before and after enzyme digestion, and the results are shown in Figure 23. The SDS-PAGE results show that the molecular weight after enzyme digestion is smaller than that before digestion, indicating that the masking structure was removed after enzyme digestion.

[0356] Other embodiments It should be understood that although the invention has been described in conjunction with a detailed description, the foregoing description is intended to illustrate and not limit the scope of the invention, which is defined by the appended claims. Other aspects, advantages, and modifications are within the scope of the appended claims.

Claims

1. A masked antigen binding domain, comprising: (a) an antigen binding domain; and (b) an alpha-helix masking structure, wherein the alpha-helix masking structure comprises an alpha-helix structure, the alpha-helix masking structure is linked to the antigen binding domain, preferably by a single covalent bond, and blocks the interaction between the antigen binding domain and its target antigen.

2. The masked antigen binding domain of claim 1, wherein (1) the alpha-helix structure does not form a coiled coil structure linked to the antigen binding domain by multiple covalent bonds; and / or (2) the alpha-helix structure specifically binds to the antigen binding domain.

3. The masked antigen binding domain of claim 1 or 2, wherein the alpha-helix masking structure comprises a single alpha-helix structure, or more than one alpha-helix structure (e.g., 2, 3, 4, or 5 alpha-helix structures).

4. A masked antigen binding domain comprising (a) an antigen binding domain, and (b) an alpha-helix masking structure, wherein the alpha-helix masking structure comprises a single polypeptide comprising an alpha-helix structure.

5. The masked antigen binding domain of any one of claims 1-4, wherein the masked antigen binding domain further comprises a cleavable linker between (a) the antigen binding domain and (b) the alpha-helix masking structure.

6. The masked antigen binding domain of claim 5, wherein the cleavable linker is cleavable by an enzyme to release the alpha-helix masking structure from the masked antigen binding domain.

7. The masked antigen binding domain of claim 6, wherein the enzyme is an enzyme present in the tumor microenvironment (e.g., a matrix metalloproteinase).

8. The masked antigen binding domain of any one of claims 1-7, wherein the antigen binding domain comprises (1) a heavy chain variable region (VH) and a light chain variable region (VL), or (2) a VHH (a camelid-derived single domain antibody).

9. The masked antigen binding domain of claim 8, wherein the alpha-helix masking structure is linked to the N- or C-terminus of the VH, VL, or VHH.

10. A masked antigen binding construct comprising (a) an antigen binding domain, and (b) an alpha-helix masking structure, wherein the alpha-helix masking structure comprises an alpha-helix structure, and the alpha-helix masking structure is linked to the antigen binding portion of the masked antigen binding construct by a single covalent bond, and blocks the interaction between the antigen binding domain and its target antigen.

11. The masked antigen binding construct of claim 10, wherein the alpha-helix structure does not form a coiled coil structure linked to the antigen binding portion of the masked antigen binding construct by multiple covalent bonds.

12. The masked antigen binding construct of claim 10 or 11, wherein the alpha-helix masking structure comprises a single alpha-helix structure.

13. A masked antigen binding construct comprising (a) an antigen binding domain, and (b) an alpha-helix masking structure, wherein the alpha-helix masking structure comprises a single polypeptide comprising an alpha-helix structure.

14. The masked antigen binding construct of any one of claims 10-13, wherein the masked antigen binding construct further comprises a cleavable linker between (a) the antigen binding domain and (b) the a-helix masking structure.

15. The masked antigen binding construct of claim 14, wherein the cleavable linker is cleavable by an enzyme to release the a-helix masking structure from the masked antigen binding construct.

16. The masked antigen binding construct of claim 15, wherein the enzyme is an enzyme present in the tumor microenvironment (e.g., a matrix metalloproteinase).

17. The masked antigen binding construct of any one of claims 10-16, comprising (a) a first polypeptide comprising the a-helix masking structure, the cleavable linker, and the VH; and (b) a second polypeptide comprising the VL, wherein the a-helix masking structure is attached to the N- or C-terminus of the first polypeptide.

18. The masked antigen binding construct of claim 17, wherein the masked antigen binding construct further comprises a third polypeptide comprising a VHH.

19. The masked antigen binding construct of claim 18, wherein the a-helix masking structure is attached to the N- or C-terminus of the third polypeptide.

20. The masked antigen binding construct of claim 17, wherein the masked antigen binding construct further comprises a third polypeptide comprising a VH and a VL.

21. The masked antigen binding construct of claim 20, wherein the a-helix masking structure is attached to the N- or C-terminus of the third polypeptide.

22. The masked antigen binding construct of claim 17, wherein the masked antigen binding construct further comprises: (c) a third polypeptide comprising a VH; and (d) a fourth polypeptide comprising a VL.

23. The masked antigen binding construct of claim 22, wherein the a-helix masking structure is attached to the N- or C-terminus of the third polypeptide.

24. The masked antigen binding construct of any one of claims 10-16, comprising (a) a first polypeptide comprising the a-helix masking structure, the cleavable linker, and the VL; and (b) a second polypeptide comprising the VH, wherein the a-helix masking structure is attached to the N- or C-terminus of the first polypeptide.

25. The masked antigen binding construct of claim 24, wherein the masked antigen binding construct further comprises a third polypeptide comprising a VHH.

26. The masked antigen binding construct of claim 25, wherein the a-helix masking structure is attached to the N- or C-terminus of the third polypeptide.

27. The masked antigen binding construct of claim 24, wherein the masked antigen binding construct further comprises a third polypeptide comprising a VH and a VL.

28. The masked antigen binding construct of claim 27, wherein the a-helix masking structure is attached to the N- or C-terminus of the third polypeptide.

29. The masked antigen binding construct of claim 24, wherein the masked antigen binding construct further comprises: (c) a third polypeptide comprising a VH; and (d) a fourth polypeptide comprising a VL.

30. The masked antigen binding construct of claim 29, wherein the alpha-helix masking structure is attached to the N-terminus or C-terminus of the fourth polypeptide.

31. The masked antigen binding construct of any one of claims 10-16, comprising a first polypeptide comprising an alpha-helix masking structure, a cleavable linker, a VH, and a VL, wherein the alpha-helix masking structure is attached to the N-terminus or C-terminus of the first polypeptide.

32. The masked antigen binding construct of claim 31, wherein the masked antigen binding construct further comprises a second polypeptide comprising a VHH.

33. The masked antigen binding construct of claim 32, wherein the alpha-helix masking structure is attached to the N-terminus or C-terminus of the second polypeptide.

34. The masked antigen binding construct of claim 31, wherein the masked antigen binding construct further comprises a second polypeptide comprising a VH and a VL.

35. The masked antigen binding construct of claim 34, wherein the alpha-helix masking structure is attached to the N-terminus or C-terminus of the second polypeptide.

36. The masked antigen binding construct of claim 27, wherein the masked antigen binding construct further comprises a second polypeptide comprising a VH and a third polypeptide comprising a VL.

37. The masked antigen binding construct of any one of claims 10-16, comprising a first polypeptide comprising an alpha-helix masking structure, a cleavable linker, and a VHH, wherein the alpha-helix masking structure is attached to the N-terminus or C-terminus of the first polypeptide.

38. The masked antigen binding construct of claim 37, wherein the masked antigen binding construct further comprises a second polypeptide comprising a VHH.

39. The masked antigen binding construct of claim 38, wherein the alpha-helix masking structure is attached to the N-terminus or C-terminus of the second polypeptide.

40. The masked antigen binding construct of claim 37, wherein the masked antigen binding construct further comprises a second polypeptide comprising a VH and a VL.

41. The masked antigen binding construct of claim 40, wherein the alpha-helix masking structure is attached to the N-terminus or C-terminus of the second polypeptide.

42. The masked antigen binding construct of claim 37, wherein the masked antigen binding construct further comprises a second polypeptide comprising a VH and a third polypeptide comprising a VL.

43. The masked antigen binding domain of any one of claims 1-9 or the masked antigen binding construct of any one of claims 10-42, wherein the alpha-helix masking structure has no more than 150 amino acids.

44. The masked antigen binding domain of claim 43, wherein the alpha-helix masking structure has 5-40 amino acids.

45. The masked antigen binding domain of claim 44, wherein the alpha-helix masking structure has 5-20 amino acids.

46. The masked antigen binding domain of any one of claims 1-9 or the masked antigen binding construct of any one of claims 10-42, wherein the alpha-helix masking structure has one or more parameters that satisfy a threshold level.

47. The threshold level of claim 46, wherein the one or more parameters are selected from (1) alpha-helix pLDDT (predicted local distance difference test), (2) alpha-helix pTM score (predicted template modeling score), (3) ipTM score between the alpha-helix and the CDR region (interface pTM score), and (4) average minimum distance between alpha carbon atoms of the alpha-helix and the CDR region.

48. The threshold level of claim 47, wherein the alpha-helix pLDDT value is greater than 55.

49. The threshold level of claim 47, wherein the alpha-helix pTM score is greater than 0.

5.

50. The threshold level of claim 47, wherein the ipTM score between the alpha-helix and the CDR region is greater than 0.6, more preferably greater than 0.

8.

51. The threshold level of claim 47, wherein the average minimum distance between alpha carbon atoms of the alpha-helix and the CDR region is less than 12 A, more preferably less than 10 A.

52. The threshold level of claim 47, wherein the average minimum distance between alpha carbon atoms of the alpha-helix and the CDR region is less than 12 A and the ipTM score is greater than 0.

6.

53. The threshold level of claim 52, wherein the average minimum distance between alpha carbon atoms of the alpha-helix and the CDR region is less than 10 A and the ipTM score is greater than 0.

8.

54. The masked antigen binding domain of any one of claims 1-9 or the masked antigen binding construct of any one of claims 10-42, comprising only one alpha-helix masking structure.

55. The masked antigen binding domain of any one of claims 1-9 or the masked antigen binding construct of any one of claims 10-42, comprising two alpha-helix masking structures, wherein the two alpha-helix masking structures are linked to two different antigen binding domains, respectively.

56. The masked antigen binding domain of any one of claims 1-9 or the masked antigen binding construct of any one of claims 10-42, wherein the alpha-helix structure comprises the amino acid sequence of any one of SEQ ID NOs: 12, 16, 18, or 20, or an amino acid sequence that is at least 95%, 97%, or 99% identical to the amino acid sequence of any one of SEQ ID NOs: 12, 16, 18, or 20.

57. The masked antigen binding domain of any one of claims 5-9 or the masked antigen binding construct of any one of claims 14-42, wherein the cleavable linker comprises a first linker, a cleavable site, and a second linker, wherein the cleavable site comprises the amino acid sequence of SEQ ID NO: 14 or 24, or an amino acid sequence that is at least 95%, 97%, or 99% identical to SEQ ID NO: 14 or 24.

58. The masked antigen binding domain of claim 57, wherein the first linker and / or second linker comprises the amino acid sequence of any one of SEQ ID NOs: 13, 21, or 23, or an amino acid sequence that is at least 95%, 97%, or 99% identical to the amino acid sequence of any one of SEQ ID NOs: 13, 21, or 23.

59. A nucleic acid comprising a nucleic acid sequence encoding the masked antigen binding domain of any one of claims 1-9 or the masked antigen binding construct of any one of claims 10-42.

60. A vector comprising the nucleic acid of claim 59.

61. A cell comprising the nucleic acid of claim 59 or the vector of claim 60.

62. A method of producing a masked antigen binding domain or a masked antigen binding construct, the method comprising culturing the cell of claim 61.

63. A masked antibody-drug conjugate (ADC) comprising a therapeutic agent covalently bound to the masked antigen binding domain of any one of claims 1-9 or the masked antigen binding construct of any one of claims 10-42.

64. The ADC of claim 63, wherein the therapeutic agent is a cytotoxic agent or a cytostatic agent.

65. A nucleotide therapy comprising the nucleic acid of claim 59.

66. A cell therapy comprising the masked antigen binding domain of any one of claims 1-9 or the masked antigen binding construct of any one of claims 10-42.

67. A fusion drug, molecule, or protein comprising (1) a cytokine, and (2) the masked antigen binding domain of any one of claims 1-9 or the masked antigen binding construct of any one of claims 10-42.

68. A method of treating a subject, the method comprising administering to a subject in need thereof a therapeutically effective amount of a composition comprising the masked antigen binding domain of any one of claims 1-9, the masked antigen binding construct of any one of claims 10-42, the ADC of claim 63 or 64, the nucleotide therapy of claim 65, the cell therapy of claim 66, or the fusion protein of claim 67.

69. A computer-implemented method for producing a masked antigen-binding construct, the method comprising: (a) providing information for a starting antigen binding domain; (b) generating candidate alpha-helix masking structures capable of binding CDR regions of the starting antigen binding domain; (c) assembling the candidate alpha-helix masking structures of step (b) in silico in the form of a masked antigen binding construct comprising the starting antigen binding domain and one of the candidate alpha-helix masking structures; (d) performing structure prediction; and (e) selecting masked antigen binding constructs having one or more parameters meeting a threshold level.

70. The method of claim 69, wherein the one or more parameters are selected from (1) alpha-helix pLDDT (predicted local distance difference test), (2) alpha-helix pTM score (predicted template modeling score), (3) ipTM score between alpha-helix and CDR region (interface pTM score), and (4) average minimum distance between alpha carbon atoms of alpha-helix and CDR region.

71. The method of claim 70, wherein the alpha-helix pLDDT value is greater than 55.

72. The method of claim 70, wherein the alpha-helix pTM score is greater than 0.

5.

73. The method of claim 70, wherein the ipTM score between alpha-helix and CDR region is greater than 0.6, more preferably greater than 0.

8.

74. The method of claim 70, wherein the average minimum distance between alpha carbon atoms of alpha-helix and CDR region is less than 12 A, more preferably less than 10 A.

75. The threshold level of claim 70, wherein the average minimum distance between alpha carbon atoms of alpha-helix and CDR region is less than 10 A and the ipTM score is greater than 0.

8.

76. The threshold level of claim 70, wherein the average minimum distance between alpha carbon atoms of alpha-helix and CDR region is less than 12 A and the ipTM score is greater than 0.

6.

77. The method of any one of claims 69-76, wherein step (b) is performed using a protein language model (PLM) or diffusion model design.

78. The method of any one of claims 69-77, wherein step (d) is performed using a fast folding software.

79. The method of claim 78, wherein the fast folding software is selected from Alphafold2, ESMFold, and OmegaFold.

80. The method of any one of claims 69-79, wherein at least 100 or 1000 antigen binding constructs are generated in silico.

81. The method of any one of claims 69-80, wherein the method further comprises preparing a vector comprising a nucleic acid sequence encoding a masked antigen binding construct; and expressing the masked antigen binding construct.

82. One or more machine-readable hardware storage devices for storing instructions executable by one or more data processing devices to perform the method of any one of claims 69-80.

83. A system comprising: one or more data processing devices; and one or more machine-readable hardware storage devices storing instructions executable by the one or more data processing devices to perform the method of any one of claims 69-80.

84. The system of claim 83, wherein the system further comprises one or more devices for preparing and expressing nucleic acid sequences.

85. An alpha-helix masking structure comprising: one or more alpha-helix peptides, wherein the one or more alpha-helix peptides are capable of specifically binding to an antigen binding domain of an antibody or antibody fragment.

86. The alpha-helix masking structure of claim 85, wherein the binding of the alpha-helix peptide to the antigen binding domain is non-covalent.

87. A masking domain (mask) comprising: one or more alpha-helix peptides and one or more loop peptides, wherein the one or more loop peptides are capable of specifically binding to an antigen binding domain of an antibody or antibody fragment.

88. A masked antigen binding domain comprising (a) an antigen binding domain, and (b) the alpha-helix masking structure of any one of claims 85-86, wherein the alpha-helix masking structure comprises an alpha-helix structure, and the alpha-helix masking structure is linked to the antigen binding domain by one or more cleavable linkers and blocks the interaction of the antigen binding domain with its target antigen.

Images